id sid tid token lemma pos bracis-28380 1 1 interpreting interpret VERB bracis-28380 1 2 convolutional convolutional ADJ bracis-28380 1 3 neural neural ADJ bracis-28380 1 4 networks network NOUN bracis-28380 1 5 for for ADP bracis-28380 1 6     SPACE bracis-28380 1 7 brain brain NOUN bracis-28380 1 8 tumor tumor NOUN bracis-28380 1 9 classification classification NOUN bracis-28380 1 10 : : PUNCT bracis-28380 1 11 an an DET bracis-28380 1 12 explainable explainable ADJ bracis-28380 1 13 artificial artificial ADJ bracis-28380 1 14 intelligence intelligence NOUN bracis-28380 1 15 approach approach NOUN bracis-28380 1 16 | | ADV bracis-28380 1 17 springer springer NOUN bracis-28380 1 18 nature nature PROPN bracis-28380 1 19 link link PROPN bracis-28380 1 20 ( ( PUNCT bracis-28380 1 21 formerly formerly ADV bracis-28380 1 22 springerlink springerlink NOUN bracis-28380 1 23 ) ) PUNCT bracis-28380 1 24 skip skip VERB bracis-28380 1 25 to to ADP bracis-28380 1 26 main main ADJ bracis-28380 1 27 content content NOUN bracis-28380 1 28 advertisement advertisement NOUN bracis-28380 1 29 account account NOUN bracis-28380 1 30 menu menu NOUN bracis-28380 1 31 find find VERB bracis-28380 1 32 a a DET bracis-28380 1 33 journal journal NOUN bracis-28380 1 34 publish publish VERB bracis-28380 1 35 with with ADP bracis-28380 1 36 us we PRON bracis-28380 1 37 track track VERB bracis-28380 1 38 your your PRON bracis-28380 1 39 research research NOUN bracis-28380 1 40 search search NOUN bracis-28380 1 41 cart cart NOUN bracis-28380 1 42 home home NOUN bracis-28380 1 43 intelligent intelligent ADJ bracis-28380 1 44 systems system NOUN bracis-28380 1 45 conference conference NOUN bracis-28380 1 46 paper paper NOUN bracis-28380 1 47 interpreting interpret VERB bracis-28380 1 48 convolutional convolutional ADJ bracis-28380 1 49 neural neural ADJ bracis-28380 1 50 networks network NOUN bracis-28380 1 51 for for ADP bracis-28380 1 52     SPACE bracis-28380 1 53 brain brain NOUN bracis-28380 1 54 tumor tumor NOUN bracis-28380 1 55 classification classification NOUN bracis-28380 1 56 : : PUNCT bracis-28380 1 57 an an DET bracis-28380 1 58 explainable explainable ADJ bracis-28380 1 59 artificial artificial ADJ bracis-28380 1 60 intelligence intelligence NOUN bracis-28380 1 61 approach approach NOUN bracis-28380 1 62 conference conference NOUN bracis-28380 1 63 paper paper NOUN bracis-28380 1 64 first first ADV bracis-28380 1 65 online online ADV bracis-28380 1 66 : : PUNCT bracis-28380 1 67 12 12 NUM bracis-28380 1 68 october october NOUN bracis-28380 1 69 2023 2023 NUM bracis-28380 1 70 pp pp SYM bracis-28380 1 71 77–91 77–91 NUM bracis-28380 1 72 cite cite VERB bracis-28380 1 73 this this DET bracis-28380 1 74 conference conference NOUN bracis-28380 1 75 paper paper NOUN bracis-28380 1 76 download download NOUN bracis-28380 1 77 book book NOUN bracis-28380 1 78 pdf pdf PROPN bracis-28380 1 79 download download NOUN bracis-28380 1 80 book book NOUN bracis-28380 1 81 epub epub PROPN bracis-28380 1 82 intelligent intelligent ADJ bracis-28380 1 83 systems system NOUN bracis-28380 1 84 ( ( PUNCT bracis-28380 1 85 bracis bracis NOUN bracis-28380 1 86 2023 2023 NUM bracis-28380 1 87 ) ) PUNCT bracis-28380 1 88 interpreting interpret VERB bracis-28380 1 89 convolutional convolutional ADJ bracis-28380 1 90 neural neural ADJ bracis-28380 1 91 networks network NOUN bracis-28380 1 92 for for ADP bracis-28380 1 93     SPACE bracis-28380 1 94 brain brain NOUN bracis-28380 1 95 tumor tumor NOUN bracis-28380 1 96 classification classification NOUN bracis-28380 1 97 : : PUNCT bracis-28380 2 1 an an DET bracis-28380 2 2 explainable explainable ADJ bracis-28380 2 3 artificial artificial ADJ bracis-28380 2 4 intelligence intelligence NOUN bracis-28380 2 5 approach approach NOUN bracis-28380 2 6 download download NOUN bracis-28380 2 7 book book NOUN bracis-28380 2 8 pdf pdf PROPN bracis-28380 2 9 download download NOUN bracis-28380 2 10 book book PROPN bracis-28380 2 11 epub epub PROPN bracis-28380 2 12 dieine dieine PROPN bracis-28380 2 13 estela estela PROPN bracis-28380 2 14 bernieri bernieri PROPN bracis-28380 2 15 schiavon schiavon PROPN bracis-28380 2 16     SPACE bracis-28380 2 17 orcid orcid NOUN bracis-28380 2 18 : : PUNCT bracis-28380 2 19 orcid.org/0000-0001-8090-23869 orcid.org/0000-0001-8090-23869 NOUN bracis-28380 2 20 , , PUNCT bracis-28380 2 21 carla carla PROPN bracis-28380 2 22 diniz diniz PROPN bracis-28380 2 23 lopes lopes PROPN bracis-28380 2 24 becker becker PROPN bracis-28380 2 25     SPACE bracis-28380 2 26 orcid orcid PROPN bracis-28380 2 27 : : PUNCT bracis-28380 2 28 orcid.org/0000-0002-0403-49169 orcid.org/0000-0002-0403-49169 PROPN bracis-28380 2 29 , , PUNCT bracis-28380 2 30 viviane viviane PROPN bracis-28380 2 31 rodrigues rodrigues PROPN bracis-28380 2 32 botelho botelho PROPN bracis-28380 2 33     SPACE bracis-28380 2 34 orcid orcid NOUN bracis-28380 2 35 : : PUNCT bracis-28380 2 36 orcid.org/0000-0003-1864-59269 orcid.org/0000-0003-1864-59269 NOUN bracis-28380 2 37 & & CCONJ bracis-28380 2 38 … … PUNCT bracis-28380 2 39 thatiane thatiane NOUN bracis-28380 2 40 alves alves PROPN bracis-28380 2 41 pianoski pianoski PROPN bracis-28380 2 42     SPACE bracis-28380 2 43 orcid orcid PROPN bracis-28380 2 44 : : PUNCT bracis-28380 2 45 orcid.org/0000-0001-9552-69939 orcid.org/0000-0001-9552-69939 NOUN bracis-28380 2 46     SPACE bracis-28380 2 47 show show VERB bracis-28380 2 48 authors author NOUN bracis-28380 2 49 part part NOUN bracis-28380 2 50 of of ADP bracis-28380 2 51 the the DET bracis-28380 2 52 book book NOUN bracis-28380 2 53 series series NOUN bracis-28380 2 54 : : PUNCT bracis-28380 2 55 lecture lecture NOUN bracis-28380 2 56 notes note NOUN bracis-28380 2 57 in in ADP bracis-28380 2 58 computer computer NOUN bracis-28380 2 59 science science NOUN bracis-28380 2 60 ( ( PUNCT bracis-28380 2 61 ( ( PUNCT bracis-28380 2 62 lnai lnai ADJ bracis-28380 2 63 , , PUNCT bracis-28380 2 64 volume volume NOUN bracis-28380 2 65 14196 14196 NUM bracis-28380 2 66 ) ) PUNCT bracis-28380 2 67 ) ) PUNCT bracis-28380 2 68 included include VERB bracis-28380 2 69 in in ADP bracis-28380 2 70 the the DET bracis-28380 2 71 following follow VERB bracis-28380 2 72 conference conference NOUN bracis-28380 2 73 series series NOUN bracis-28380 2 74 : : PUNCT bracis-28380 2 75 brazilian brazilian ADJ bracis-28380 2 76 conference conference NOUN bracis-28380 2 77 on on ADP bracis-28380 2 78 intelligent intelligent ADJ bracis-28380 2 79 systems system NOUN bracis-28380 2 80 613 613 NUM bracis-28380 2 81 accesses access NOUN bracis-28380 2 82 10 10 NUM bracis-28380 2 83 citations citation NOUN bracis-28380 2 84 abstract abstract ADJ bracis-28380 2 85 brain brain NOUN bracis-28380 2 86 tumors tumor NOUN bracis-28380 2 87 pose pose VERB bracis-28380 2 88 a a DET bracis-28380 2 89 complex complex ADJ bracis-28380 2 90 medical medical ADJ bracis-28380 2 91 challenge challenge NOUN bracis-28380 2 92 , , PUNCT bracis-28380 2 93 requiring require VERB bracis-28380 2 94 a a DET bracis-28380 2 95 specific specific ADJ bracis-28380 2 96 approach approach NOUN bracis-28380 2 97 for for ADP bracis-28380 2 98 accurate accurate ADJ bracis-28380 2 99 diagnosis diagnosis NOUN bracis-28380 2 100 and and CCONJ bracis-28380 2 101 effective effective ADJ bracis-28380 2 102 treatment treatment NOUN bracis-28380 2 103 . . PUNCT bracis-28380 3 1 early early ADJ bracis-28380 3 2 detection detection NOUN bracis-28380 3 3 can can AUX bracis-28380 3 4 significantly significantly ADV bracis-28380 3 5 improve improve VERB bracis-28380 3 6 outcomes outcome NOUN bracis-28380 3 7 and and CCONJ bracis-28380 3 8 quality quality NOUN bracis-28380 3 9 of of ADP bracis-28380 3 10 life life NOUN bracis-28380 3 11 for for ADP bracis-28380 3 12 patients patient NOUN bracis-28380 3 13 with with ADP bracis-28380 3 14 brain brain NOUN bracis-28380 3 15 tumors tumor NOUN bracis-28380 3 16 . . PUNCT bracis-28380 4 1 magnetic magnetic ADJ bracis-28380 4 2 resonance resonance NOUN bracis-28380 4 3 ( ( PUNCT bracis-28380 4 4 mri mri NOUN bracis-28380 4 5 ) ) PUNCT bracis-28380 4 6 is be AUX bracis-28380 4 7 a a DET bracis-28380 4 8 powerful powerful ADJ bracis-28380 4 9 diagnostic diagnostic ADJ bracis-28380 4 10 tool tool NOUN bracis-28380 4 11 , , PUNCT bracis-28380 4 12 and and CCONJ bracis-28380 4 13 convolutional convolutional ADJ bracis-28380 4 14 neural neural ADJ bracis-28380 4 15 networks network NOUN bracis-28380 4 16 ( ( PUNCT bracis-28380 4 17 cnns cnns PROPN bracis-28380 4 18 ) ) PUNCT bracis-28380 4 19 are be AUX bracis-28380 4 20 efficient efficient ADJ bracis-28380 4 21 deep deep ADJ bracis-28380 4 22 learning learn VERB bracis-28380 4 23 algorithms algorithm NOUN bracis-28380 4 24 for for ADP bracis-28380 4 25 image image NOUN bracis-28380 4 26 analysis analysis NOUN bracis-28380 4 27 . . PUNCT bracis-28380 5 1 in in ADP bracis-28380 5 2 this this DET bracis-28380 5 3 study study NOUN bracis-28380 5 4 , , PUNCT bracis-28380 5 5 we we PRON bracis-28380 5 6 explored explore VERB bracis-28380 5 7 using use VERB bracis-28380 5 8 two two NUM bracis-28380 5 9 cnn cnn PROPN bracis-28380 5 10 models model NOUN bracis-28380 5 11 for for ADP bracis-28380 5 12 brain brain NOUN bracis-28380 5 13 tumor tumor NOUN bracis-28380 5 14 classification classification NOUN bracis-28380 5 15 and and CCONJ bracis-28380 5 16 applied apply VERB bracis-28380 5 17 hyperparameter hyperparameter NOUN bracis-28380 5 18 optimization optimization NOUN bracis-28380 5 19 and and CCONJ bracis-28380 5 20 data datum NOUN bracis-28380 5 21 augmentation augmentation NOUN bracis-28380 5 22 techniques technique NOUN bracis-28380 5 23 to to PART bracis-28380 5 24 achieve achieve VERB bracis-28380 5 25 an an DET bracis-28380 5 26 accuracy accuracy NOUN bracis-28380 5 27 of of ADP bracis-28380 5 28 up up ADP bracis-28380 5 29 to to PART bracis-28380 5 30 96 96 NUM bracis-28380 5 31 % % NOUN bracis-28380 5 32 . . PUNCT bracis-28380 6 1 in in ADP bracis-28380 6 2 addition addition NOUN bracis-28380 6 3 , , PUNCT bracis-28380 6 4 we we PRON bracis-28380 6 5 use use VERB bracis-28380 6 6 explainable explainable ADJ bracis-28380 6 7 artificial artificial ADJ bracis-28380 6 8 intelligence intelligence NOUN bracis-28380 6 9 ( ( PUNCT bracis-28380 6 10 xai xai PROPN bracis-28380 6 11 ) ) PUNCT bracis-28380 6 12 techniques technique NOUN bracis-28380 6 13 to to PART bracis-28380 6 14 visualize visualize VERB bracis-28380 6 15 and and CCONJ bracis-28380 6 16 interpret interpret VERB bracis-28380 6 17 the the DET bracis-28380 6 18 behavior behavior NOUN bracis-28380 6 19 of of ADP bracis-28380 6 20 cnn cnn PROPN bracis-28380 6 21 models model NOUN bracis-28380 6 22 . . PUNCT bracis-28380 7 1 our our PRON bracis-28380 7 2 results result NOUN bracis-28380 7 3 show show VERB bracis-28380 7 4 that that SCONJ bracis-28380 7 5 cnn cnn PROPN bracis-28380 7 6 models model NOUN bracis-28380 7 7 accurately accurately ADV bracis-28380 7 8 classified classify VERB bracis-28380 7 9 mri mri NOUN bracis-28380 7 10 images image NOUN bracis-28380 7 11 with with ADP bracis-28380 7 12 brain brain NOUN bracis-28380 7 13 tumors tumor NOUN bracis-28380 7 14 . . PUNCT bracis-28380 8 1 xai xai PROPN bracis-28380 8 2 techniques technique NOUN bracis-28380 8 3 helped help VERB bracis-28380 8 4 us we PRON bracis-28380 8 5 to to PART bracis-28380 8 6 identify identify VERB bracis-28380 8 7 the the DET bracis-28380 8 8 patterns pattern NOUN bracis-28380 8 9 and and CCONJ bracis-28380 8 10 features feature NOUN bracis-28380 8 11 used use VERB bracis-28380 8 12 by by ADP bracis-28380 8 13 the the DET bracis-28380 8 14 models model NOUN bracis-28380 8 15 to to PART bracis-28380 8 16 make make VERB bracis-28380 8 17 predictions prediction NOUN bracis-28380 8 18 . . PUNCT bracis-28380 9 1 this this DET bracis-28380 9 2 study study NOUN bracis-28380 9 3 supports support VERB bracis-28380 9 4 the the DET bracis-28380 9 5 development development NOUN bracis-28380 9 6 of of ADP bracis-28380 9 7 more more ADV bracis-28380 9 8 reliable reliable ADJ bracis-28380 9 9 medical medical ADJ bracis-28380 9 10 diagnoses diagnosis NOUN bracis-28380 9 11 for for ADP bracis-28380 9 12 brain brain NOUN bracis-28380 9 13 tumors tumor NOUN bracis-28380 9 14 using use VERB bracis-28380 9 15 cnn cnn PROPN bracis-28380 9 16 models model NOUN bracis-28380 9 17 and and CCONJ bracis-28380 9 18 xai xai PROPN bracis-28380 9 19 techniques technique NOUN bracis-28380 9 20 . . PUNCT bracis-28380 10 1 the the DET bracis-28380 10 2 source source NOUN bracis-28380 10 3 code code NOUN bracis-28380 10 4 is be AUX bracis-28380 10 5 available available ADJ bracis-28380 10 6 at at ADP bracis-28380 10 7 ( ( PUNCT bracis-28380 10 8 https://github.com/dieineb/bracis23 https://github.com/dieineb/bracis23 PROPN bracis-28380 10 9 ) ) PUNCT bracis-28380 10 10 . . PUNCT bracis-28380 11 1 the the DET bracis-28380 11 2 repository repository NOUN bracis-28380 11 3 also also ADV bracis-28380 11 4 has have VERB bracis-28380 11 5 images image NOUN bracis-28380 11 6 generated generate VERB bracis-28380 11 7 during during ADP bracis-28380 11 8 the the DET bracis-28380 11 9 experiments experiment NOUN bracis-28380 11 10 . . PUNCT bracis-28380 12 1 download download NOUN bracis-28380 12 2 conference conference NOUN bracis-28380 12 3 paper paper NOUN bracis-28380 12 4 pdf pdf NOUN bracis-28380 12 5 similar similar ADJ bracis-28380 12 6 content content NOUN bracis-28380 12 7 being be AUX bracis-28380 12 8 viewed view VERB bracis-28380 12 9 by by ADP bracis-28380 12 10 others other NOUN bracis-28380 12 11 interpreting interpret VERB bracis-28380 12 12 convolutional convolutional ADJ bracis-28380 12 13 neural neural ADJ bracis-28380 12 14 networks network NOUN bracis-28380 12 15 for for ADP bracis-28380 12 16 brain brain NOUN bracis-28380 12 17 tumors tumor NOUN bracis-28380 12 18 : : PUNCT bracis-28380 12 19 a a DET bracis-28380 12 20 transparent transparent ADJ bracis-28380 12 21 box box NOUN bracis-28380 12 22 approach approach NOUN bracis-28380 12 23 chapter chapter NOUN bracis-28380 12 24 © © PROPN bracis-28380 12 25 2025 2025 NUM bracis-28380 12 26 classification classification NOUN bracis-28380 12 27 of of ADP bracis-28380 12 28 brain brain NOUN bracis-28380 12 29 tumors tumor NOUN bracis-28380 12 30 from from ADP bracis-28380 12 31 mri mri NOUN bracis-28380 12 32 images image NOUN bracis-28380 12 33 utilizing utilize VERB bracis-28380 12 34 a a DET bracis-28380 12 35 convolutional convolutional ADJ bracis-28380 12 36 neural neural ADJ bracis-28380 12 37 network network NOUN bracis-28380 12 38 chapter chapter NOUN bracis-28380 12 39 © © PROPN bracis-28380 12 40 2024 2024 NUM bracis-28380 12 41 tumor tumor NOUN bracis-28380 12 42 sight sight NOUN bracis-28380 12 43 ai ai VERB bracis-28380 12 44 : : PUNCT bracis-28380 12 45 brain brain NOUN bracis-28380 12 46 tumor tumor NOUN bracis-28380 12 47 prediction prediction NOUN bracis-28380 12 48 system system NOUN bracis-28380 12 49 using use VERB bracis-28380 12 50 deep deep ADJ bracis-28380 12 51 learning learning NOUN bracis-28380 12 52 and and CCONJ bracis-28380 12 53 explainable explainable ADJ bracis-28380 12 54 artificial artificial ADJ bracis-28380 12 55 intelligence intelligence NOUN bracis-28380 12 56 ( ( PUNCT bracis-28380 12 57 xai xai PROPN bracis-28380 12 58 ) ) PUNCT bracis-28380 12 59 chapter chapter NOUN bracis-28380 12 60 © © PROPN bracis-28380 12 61 2026 2026 NUM bracis-28380 12 62 explore explore VERB bracis-28380 12 63 related relate VERB bracis-28380 12 64 subjects subject NOUN bracis-28380 12 65 discover discover VERB bracis-28380 12 66 the the DET bracis-28380 12 67 latest late ADJ bracis-28380 12 68 articles article NOUN bracis-28380 12 69 , , PUNCT bracis-28380 12 70 books book NOUN bracis-28380 12 71 and and CCONJ bracis-28380 12 72 news news NOUN bracis-28380 12 73 in in ADP bracis-28380 12 74 related related ADJ bracis-28380 12 75 subjects subject NOUN bracis-28380 12 76 , , PUNCT bracis-28380 12 77 suggested suggest VERB bracis-28380 12 78 using use VERB bracis-28380 12 79 machine machine NOUN bracis-28380 12 80 learning learning NOUN bracis-28380 12 81 . . PUNCT bracis-28380 13 1 brain brain NOUN bracis-28380 13 2 cancer cancer NOUN bracis-28380 13 3 computational computational ADJ bracis-28380 13 4 intelligence intelligence NOUN bracis-28380 13 5 learning learn VERB bracis-28380 13 6 algorithms algorithm NOUN bracis-28380 13 7 machine machine NOUN bracis-28380 13 8 learning learn VERB bracis-28380 13 9 neuroradiology neuroradiology NOUN bracis-28380 13 10 artificial artificial ADJ bracis-28380 13 11 intelligence intelligence NOUN bracis-28380 13 12 1 1 NUM bracis-28380 13 13 introduction introduction NOUN bracis-28380 13 14 the the DET bracis-28380 13 15 central central ADJ bracis-28380 13 16 nervous nervous ADJ bracis-28380 13 17 system system NOUN bracis-28380 13 18 ( ( PUNCT bracis-28380 13 19 cns cns NOUN bracis-28380 13 20 ) ) PUNCT bracis-28380 13 21 tumors tumor NOUN bracis-28380 13 22 are be AUX bracis-28380 13 23 due due ADJ bracis-28380 13 24 to to ADP bracis-28380 13 25 the the DET bracis-28380 13 26 abnormal abnormal ADJ bracis-28380 13 27 growth growth NOUN bracis-28380 13 28 of of ADP bracis-28380 13 29 cells cell NOUN bracis-28380 13 30 in in ADP bracis-28380 13 31 the the DET bracis-28380 13 32 tissues tissue NOUN bracis-28380 13 33 of of ADP bracis-28380 13 34 these these DET bracis-28380 13 35 locations location NOUN bracis-28380 13 36 . . PUNCT bracis-28380 14 1 according accord VERB bracis-28380 14 2 to to ADP bracis-28380 14 3 the the DET bracis-28380 14 4 national national PROPN bracis-28380 14 5 cancer cancer PROPN bracis-28380 14 6 institute institute PROPN bracis-28380 14 7 , , PUNCT bracis-28380 14 8 cns cns PROPN bracis-28380 14 9 cancer cancer NOUN bracis-28380 14 10 represents represent VERB bracis-28380 14 11 1.4 1.4 NUM bracis-28380 14 12 % % NOUN bracis-28380 14 13 to to ADP bracis-28380 14 14 1.8 1.8 NUM bracis-28380 14 15 % % NOUN bracis-28380 14 16 of of ADP bracis-28380 14 17 all all DET bracis-28380 14 18 malignant malignant ADJ bracis-28380 14 19 tumors tumor NOUN bracis-28380 14 20 worldwide worldwide ADV bracis-28380 14 21 . . PUNCT bracis-28380 15 1 about about ADV bracis-28380 15 2 88 88 NUM bracis-28380 15 3 % % NOUN bracis-28380 15 4 of of ADP bracis-28380 15 5 cns cns PROPN bracis-28380 15 6 tumors tumor NOUN bracis-28380 15 7 are be AUX bracis-28380 15 8 in in ADP bracis-28380 15 9 the the DET bracis-28380 15 10 brain brain NOUN bracis-28380 16 1 [ [ X bracis-28380 16 2 1 1 NUM bracis-28380 16 3 ] ] PUNCT bracis-28380 16 4 . . PUNCT bracis-28380 17 1 these these DET bracis-28380 17 2 tumors tumor NOUN bracis-28380 17 3 cause cause VERB bracis-28380 17 4 profound profound ADJ bracis-28380 17 5 deficits deficit NOUN bracis-28380 17 6 and and CCONJ bracis-28380 17 7 are be AUX bracis-28380 17 8 fatal fatal ADJ bracis-28380 17 9 in in ADP bracis-28380 17 10 patients patient NOUN bracis-28380 17 11 whose whose DET bracis-28380 17 12 initial initial ADJ bracis-28380 17 13 symptoms symptom NOUN bracis-28380 17 14 are be AUX bracis-28380 17 15 often often ADV bracis-28380 17 16 not not PART bracis-28380 17 17 perceived perceive VERB bracis-28380 17 18 . . PUNCT bracis-28380 18 1 malignant malignant ADJ bracis-28380 18 2 - - PUNCT bracis-28380 18 3 based base VERB bracis-28380 18 4 tumors tumor NOUN bracis-28380 18 5 are be AUX bracis-28380 18 6 classified classify VERB bracis-28380 18 7 by by ADP bracis-28380 18 8 the the DET bracis-28380 18 9 world world PROPN bracis-28380 18 10 health health PROPN bracis-28380 18 11 organization organization NOUN bracis-28380 18 12 ( ( PUNCT bracis-28380 18 13 who who PRON bracis-28380 18 14 ) ) PUNCT bracis-28380 19 1 [ [ X bracis-28380 19 2 2 2 X bracis-28380 19 3 ] ] PUNCT bracis-28380 19 4 into into ADP bracis-28380 19 5 four four NUM bracis-28380 19 6 stages stage NOUN bracis-28380 19 7 ( ( PUNCT bracis-28380 19 8 i i NOUN bracis-28380 19 9 - - PUNCT bracis-28380 19 10 iv iv NOUN bracis-28380 19 11 ) ) PUNCT bracis-28380 19 12 according accord VERB bracis-28380 19 13 to to ADP bracis-28380 19 14 aggressiveness aggressiveness NOUN bracis-28380 19 15 . . PUNCT bracis-28380 20 1 tumors tumor NOUN bracis-28380 20 2 classified classify VERB bracis-28380 20 3 in in ADP bracis-28380 20 4 stages stage NOUN bracis-28380 20 5 iii iii NOUN bracis-28380 20 6 and and CCONJ bracis-28380 20 7 iv iv NUM bracis-28380 20 8 are be AUX bracis-28380 20 9 more more ADV bracis-28380 20 10 aggressive aggressive ADJ bracis-28380 20 11 and and CCONJ bracis-28380 20 12 need need VERB bracis-28380 20 13 immediate immediate ADJ bracis-28380 20 14 treatment treatment NOUN bracis-28380 20 15 and and CCONJ bracis-28380 20 16 can can AUX bracis-28380 20 17 lead lead VERB bracis-28380 20 18 the the DET bracis-28380 20 19 individual individual NOUN bracis-28380 20 20 to to ADP bracis-28380 20 21 death death NOUN bracis-28380 20 22 in in ADP bracis-28380 20 23 less less ADJ bracis-28380 20 24 than than ADP bracis-28380 20 25 two two NUM bracis-28380 20 26 years year NOUN bracis-28380 20 27 . . PUNCT bracis-28380 21 1 the the DET bracis-28380 21 2 classification classification NOUN bracis-28380 21 3 of of ADP bracis-28380 21 4 magnetic magnetic ADJ bracis-28380 21 5 resonance resonance NOUN bracis-28380 21 6 imaging imaging NOUN bracis-28380 21 7 ( ( PUNCT bracis-28380 21 8 mri mri NOUN bracis-28380 21 9 ) ) PUNCT bracis-28380 21 10 is be AUX bracis-28380 21 11 an an DET bracis-28380 21 12 important important ADJ bracis-28380 21 13 part part NOUN bracis-28380 21 14 of of ADP bracis-28380 21 15 the the DET bracis-28380 21 16 processing processing NOUN bracis-28380 21 17 of of ADP bracis-28380 21 18 medical medical ADJ bracis-28380 21 19 imaging imaging NOUN bracis-28380 21 20 and and CCONJ bracis-28380 21 21 assists assist VERB bracis-28380 21 22 physicians physician NOUN bracis-28380 21 23 in in ADP bracis-28380 21 24 making make VERB bracis-28380 21 25 diagnoses diagnosis NOUN bracis-28380 21 26 and and CCONJ bracis-28380 21 27 more more ADV bracis-28380 21 28 accurate accurate ADJ bracis-28380 21 29 treatments treatment NOUN bracis-28380 21 30 . . PUNCT bracis-28380 22 1 the the DET bracis-28380 22 2 mri mri NOUN bracis-28380 22 3 is be AUX bracis-28380 22 4 one one NUM bracis-28380 22 5 of of ADP bracis-28380 22 6 the the DET bracis-28380 22 7 leading lead VERB bracis-28380 22 8 imaging imaging NOUN bracis-28380 22 9 tools tool NOUN bracis-28380 22 10 for for ADP bracis-28380 22 11 studying study VERB bracis-28380 22 12 brain brain NOUN bracis-28380 22 13 issues issue NOUN bracis-28380 22 14 . . PUNCT bracis-28380 23 1 it it PRON bracis-28380 23 2 is be AUX bracis-28380 23 3 often often ADV bracis-28380 23 4 the the DET bracis-28380 23 5 modality modality NOUN bracis-28380 23 6 of of ADP bracis-28380 23 7 choice choice NOUN bracis-28380 23 8 for for ADP bracis-28380 23 9 diagnosis diagnosis NOUN bracis-28380 23 10 and and CCONJ bracis-28380 23 11 treatment treatment NOUN bracis-28380 23 12 planning planning NOUN bracis-28380 23 13 for for ADP bracis-28380 23 14 tumors tumor NOUN bracis-28380 23 15 brain brain VERB bracis-28380 23 16 due due ADP bracis-28380 23 17 to to ADP bracis-28380 23 18 the the DET bracis-28380 23 19 high high ADJ bracis-28380 23 20 resolution resolution NOUN bracis-28380 23 21 and and CCONJ bracis-28380 23 22 significant significant ADJ bracis-28380 23 23 contrast contrast NOUN bracis-28380 23 24 of of ADP bracis-28380 23 25 soft soft ADJ bracis-28380 23 26 tissues tissue NOUN bracis-28380 23 27 [ [ X bracis-28380 23 28 3 3 NUM bracis-28380 23 29 ] ] PUNCT bracis-28380 23 30 . . PUNCT bracis-28380 24 1 the the DET bracis-28380 24 2 first first ADJ bracis-28380 24 3 methods method NOUN bracis-28380 24 4 of of ADP bracis-28380 24 5 characterization characterization NOUN bracis-28380 24 6 of of ADP bracis-28380 24 7 cancer cancer NOUN bracis-28380 24 8 brains brain NOUN bracis-28380 24 9 were be AUX bracis-28380 24 10 predominantly predominantly ADV bracis-28380 24 11 based base VERB bracis-28380 24 12 on on ADP bracis-28380 24 13 modeling model VERB bracis-28380 24 14 neuroimaging neuroimage VERB bracis-28380 24 15 data datum NOUN bracis-28380 24 16 statistics statistic NOUN bracis-28380 24 17 . . PUNCT bracis-28380 25 1 driven drive VERB bracis-28380 25 2 by by ADP bracis-28380 25 3 advances advance NOUN bracis-28380 25 4 in in ADP bracis-28380 25 5 computing compute VERB bracis-28380 25 6 , , PUNCT bracis-28380 25 7 deep deep ADJ bracis-28380 25 8 learning learning NOUN bracis-28380 25 9 has have AUX bracis-28380 25 10 become become VERB bracis-28380 25 11 standard standard ADJ bracis-28380 25 12 in in ADP bracis-28380 25 13 image image NOUN bracis-28380 25 14 classification classification NOUN bracis-28380 25 15 medical medical ADJ bracis-28380 25 16 . . PUNCT bracis-28380 26 1 integrated integrate VERB bracis-28380 26 2 statistics statistic NOUN bracis-28380 26 3 and and CCONJ bracis-28380 26 4 deep deep ADJ bracis-28380 26 5 learning learning NOUN bracis-28380 26 6 methods method NOUN bracis-28380 26 7 have have AUX bracis-28380 26 8 recently recently ADV bracis-28380 26 9 emerged emerge VERB bracis-28380 26 10 as as ADP bracis-28380 26 11 a a DET bracis-28380 26 12 new new ADJ bracis-28380 26 13 direction direction NOUN bracis-28380 26 14 in in ADP bracis-28380 26 15 the the DET bracis-28380 26 16 automation automation NOUN bracis-28380 26 17 of of ADP bracis-28380 26 18 medical medical ADJ bracis-28380 26 19 practice practice NOUN bracis-28380 26 20 , , PUNCT bracis-28380 26 21 unifying unify VERB bracis-28380 26 22 knowledge knowledge NOUN bracis-28380 26 23 multidisciplinary multidisciplinary NOUN bracis-28380 26 24 in in ADP bracis-28380 26 25 medicine medicine NOUN bracis-28380 26 26 , , PUNCT bracis-28380 26 27 statistics statistic NOUN bracis-28380 26 28 , , PUNCT bracis-28380 26 29 and and CCONJ bracis-28380 26 30 artificial artificial ADJ bracis-28380 26 31 intelligence intelligence NOUN bracis-28380 26 32 [ [ X bracis-28380 26 33 4 4 NUM bracis-28380 26 34 ] ] PUNCT bracis-28380 26 35 . . PUNCT bracis-28380 27 1 convolutional convolutional ADJ bracis-28380 27 2 neural neural ADJ bracis-28380 27 3 networks network NOUN bracis-28380 27 4 ( ( PUNCT bracis-28380 27 5 cnns cnns PROPN bracis-28380 27 6 ) ) PUNCT bracis-28380 27 7 are be AUX bracis-28380 27 8 deep deep ADJ bracis-28380 27 9 learning learn VERB bracis-28380 27 10 algorithms algorithm NOUN bracis-28380 27 11 , , PUNCT bracis-28380 27 12 very very ADV bracis-28380 27 13 efficient efficient ADJ bracis-28380 27 14 for for ADP bracis-28380 27 15 image image NOUN bracis-28380 27 16 analysis analysis NOUN bracis-28380 27 17 [ [ X bracis-28380 27 18 5 5 NUM bracis-28380 27 19 ] ] PUNCT bracis-28380 27 20 . . PUNCT bracis-28380 28 1 they they PRON bracis-28380 28 2 are be AUX bracis-28380 28 3 frequently frequently ADV bracis-28380 28 4 used use VERB bracis-28380 28 5 in in ADP bracis-28380 28 6 health health NOUN bracis-28380 28 7 , , PUNCT bracis-28380 28 8 mainly mainly ADV bracis-28380 28 9 for for ADP bracis-28380 28 10 classifying classify VERB bracis-28380 28 11 , , PUNCT bracis-28380 28 12 locating locating NOUN bracis-28380 28 13 , , PUNCT bracis-28380 28 14 detecting detecting NOUN bracis-28380 28 15 , , PUNCT bracis-28380 28 16 and and CCONJ bracis-28380 28 17 segmenting segment VERB bracis-28380 28 18 tumors tumor NOUN bracis-28380 28 19 on on ADP bracis-28380 28 20 magnetic magnetic ADJ bracis-28380 28 21 resonance resonance NOUN bracis-28380 28 22 imaging imaging NOUN bracis-28380 28 23 . . PUNCT bracis-28380 29 1 the the DET bracis-28380 29 2 architecture architecture NOUN bracis-28380 29 3 typical typical ADJ bracis-28380 29 4 of of ADP bracis-28380 29 5 cnn cnn PROPN bracis-28380 29 6 for for ADP bracis-28380 29 7 image image NOUN bracis-28380 29 8 processing processing NOUN bracis-28380 29 9 consists consist NOUN bracis-28380 29 10 of of ADP bracis-28380 29 11 a a DET bracis-28380 29 12 series series NOUN bracis-28380 29 13 of of ADP bracis-28380 29 14 layers layer NOUN bracis-28380 29 15 of of ADP bracis-28380 29 16 convolution convolution NOUN bracis-28380 29 17 filters filter NOUN bracis-28380 29 18 interspersed intersperse VERB bracis-28380 29 19 with with ADP bracis-28380 29 20 a a DET bracis-28380 29 21 series series NOUN bracis-28380 29 22 of of ADP bracis-28380 29 23 reduction reduction NOUN bracis-28380 29 24 layers layer NOUN bracis-28380 29 25 or or CCONJ bracis-28380 29 26 data datum NOUN bracis-28380 29 27 grouping grouping NOUN bracis-28380 29 28 . . PUNCT bracis-28380 30 1 convolutional convolutional ADJ bracis-28380 30 2 layers layer NOUN bracis-28380 30 3 in in ADP bracis-28380 30 4 cnns cnn NOUN bracis-28380 30 5 serve serve VERB bracis-28380 30 6 as as ADP bracis-28380 30 7 feature feature NOUN bracis-28380 30 8 extractors extractor NOUN bracis-28380 30 9 , , PUNCT bracis-28380 30 10 learning learn VERB bracis-28380 30 11 feature feature NOUN bracis-28380 30 12 representations representation NOUN bracis-28380 30 13 from from ADP bracis-28380 30 14 input input NOUN bracis-28380 30 15 images image NOUN bracis-28380 30 16 . . PUNCT bracis-28380 31 1 neurons neuron NOUN bracis-28380 31 2 in in ADP bracis-28380 31 3 these these DET bracis-28380 31 4 layers layer NOUN bracis-28380 31 5 are be AUX bracis-28380 31 6 organized organize VERB bracis-28380 31 7 into into ADP bracis-28380 31 8 feature feature NOUN bracis-28380 31 9 maps map NOUN bracis-28380 31 10 , , PUNCT bracis-28380 31 11 and and CCONJ bracis-28380 31 12 filters filter NOUN bracis-28380 31 13 connect connect VERB bracis-28380 31 14 them they PRON bracis-28380 31 15 to to ADP bracis-28380 31 16 neurons neuron NOUN bracis-28380 31 17 in in ADP bracis-28380 31 18 the the DET bracis-28380 31 19 previous previous ADJ bracis-28380 31 20 layer layer NOUN bracis-28380 31 21 . . PUNCT bracis-28380 32 1 the the DET bracis-28380 32 2 convolution convolution NOUN bracis-28380 32 3 filters filter NOUN bracis-28380 32 4 detect detect VERB bracis-28380 32 5 relevant relevant ADJ bracis-28380 32 6 image image NOUN bracis-28380 32 7 features feature NOUN bracis-28380 32 8 , , PUNCT bracis-28380 32 9 progressively progressively ADV bracis-28380 32 10 capturing capture VERB bracis-28380 32 11 more more ADJ bracis-28380 32 12 complex complex ADJ bracis-28380 32 13 patterns pattern NOUN bracis-28380 32 14 . . PUNCT bracis-28380 33 1 additionally additionally ADV bracis-28380 33 2 , , PUNCT bracis-28380 33 3 pooling pool VERB bracis-28380 33 4 layers layer NOUN bracis-28380 33 5 are be AUX bracis-28380 33 6 commonly commonly ADV bracis-28380 33 7 employed employ VERB bracis-28380 33 8 after after ADP bracis-28380 33 9 convolutional convolutional ADJ bracis-28380 33 10 layers layer NOUN bracis-28380 33 11 to to PART bracis-28380 33 12 reduce reduce VERB bracis-28380 33 13 image image NOUN bracis-28380 33 14 dimensions dimension NOUN bracis-28380 33 15 . . PUNCT bracis-28380 34 1 multiple multiple ADJ bracis-28380 34 2 convolutional convolutional ADJ bracis-28380 34 3 layers layer NOUN bracis-28380 34 4 are be AUX bracis-28380 34 5 often often ADV bracis-28380 34 6 stacked stack VERB bracis-28380 34 7 together together ADV bracis-28380 34 8 to to PART bracis-28380 34 9 create create VERB bracis-28380 34 10 a a DET bracis-28380 34 11 deep deep ADJ bracis-28380 34 12 model model NOUN bracis-28380 34 13 , , PUNCT bracis-28380 34 14 enabling enable VERB bracis-28380 34 15 the the DET bracis-28380 34 16 extraction extraction NOUN bracis-28380 34 17 of of ADP bracis-28380 34 18 abstract abstract ADJ bracis-28380 34 19 feature feature NOUN bracis-28380 34 20 representations representation NOUN bracis-28380 34 21 [ [ X bracis-28380 34 22 6 6 NUM bracis-28380 34 23 ] ] PUNCT bracis-28380 34 24 . . PUNCT bracis-28380 35 1 the the DET bracis-28380 35 2 fully fully ADV bracis-28380 35 3 connected connected ADJ bracis-28380 35 4 layers layer NOUN bracis-28380 35 5 interpret interpret VERB bracis-28380 35 6 these these DET bracis-28380 35 7 extracted extract VERB bracis-28380 35 8 features feature NOUN bracis-28380 35 9 and and CCONJ bracis-28380 35 10 perform perform VERB bracis-28380 35 11 high high ADJ bracis-28380 35 12 - - PUNCT bracis-28380 35 13 level level NOUN bracis-28380 35 14 reasoning reasoning NOUN bracis-28380 35 15 tasks task NOUN bracis-28380 35 16 . . PUNCT bracis-28380 36 1 deep deep ADJ bracis-28380 36 2 learning learning NOUN bracis-28380 36 3 algorithms algorithm NOUN bracis-28380 36 4 are be AUX bracis-28380 36 5 having have VERB bracis-28380 36 6 an an DET bracis-28380 36 7 increasing increase VERB bracis-28380 36 8 impact impact NOUN bracis-28380 36 9 on on ADP bracis-28380 36 10 our our PRON bracis-28380 36 11 everyday everyday ADJ bracis-28380 36 12 lives life NOUN bracis-28380 36 13 . . PUNCT bracis-28380 37 1 as as SCONJ bracis-28380 37 2 machine machine NOUN bracis-28380 37 3 learning learn VERB bracis-28380 37 4 algorithms algorithm NOUN bracis-28380 37 5 are be AUX bracis-28380 37 6 increasingly increasingly ADV bracis-28380 37 7 applied apply VERB bracis-28380 37 8 to to ADP bracis-28380 37 9 high high ADJ bracis-28380 37 10 - - PUNCT bracis-28380 37 11 impact impact NOUN bracis-28380 37 12 , , PUNCT bracis-28380 37 13 high high ADJ bracis-28380 37 14 - - PUNCT bracis-28380 37 15 risk risk NOUN bracis-28380 37 16 tasks task NOUN bracis-28380 37 17 like like ADP bracis-28380 37 18 medical medical ADJ bracis-28380 37 19 diagnoses diagnosis NOUN bracis-28380 37 20 , , PUNCT bracis-28380 37 21 it it PRON bracis-28380 37 22 is be AUX bracis-28380 37 23 critical critical ADJ bracis-28380 37 24 that that SCONJ bracis-28380 37 25 researchers researcher NOUN bracis-28380 37 26 can can AUX bracis-28380 37 27 explain explain VERB bracis-28380 37 28 how how SCONJ bracis-28380 37 29 these these DET bracis-28380 37 30 algorithms algorithm NOUN bracis-28380 37 31 arrived arrive VERB bracis-28380 37 32 at at ADP bracis-28380 37 33 their their PRON bracis-28380 37 34 predictions prediction NOUN bracis-28380 37 35 [ [ X bracis-28380 37 36 7 7 NUM bracis-28380 37 37 ] ] PUNCT bracis-28380 37 38 . . PUNCT bracis-28380 38 1 however however ADV bracis-28380 38 2 , , PUNCT bracis-28380 38 3 neural neural ADJ bracis-28380 38 4 network network NOUN bracis-28380 38 5 architectures architecture NOUN bracis-28380 38 6 have have VERB bracis-28380 38 7 a a DET bracis-28380 38 8 natural natural ADJ bracis-28380 38 9 propensity propensity NOUN bracis-28380 38 10 for for ADP bracis-28380 38 11 opacity opacity NOUN bracis-28380 38 12 in in ADP bracis-28380 38 13 understanding understand VERB bracis-28380 38 14 data datum NOUN bracis-28380 38 15 processing processing NOUN bracis-28380 38 16 and and CCONJ bracis-28380 38 17 prediction prediction NOUN bracis-28380 38 18 . . PUNCT bracis-28380 39 1 explainable explainable ADJ bracis-28380 39 2 artificial artificial ADJ bracis-28380 39 3 intelligence intelligence NOUN bracis-28380 39 4 ( ( PUNCT bracis-28380 39 5 xai xai PROPN bracis-28380 39 6 ) ) PUNCT bracis-28380 39 7 , , PUNCT bracis-28380 39 8 used use VERB bracis-28380 39 9 in in ADP bracis-28380 39 10 machine machine NOUN bracis-28380 39 11 learning learning NOUN bracis-28380 39 12 , , PUNCT bracis-28380 39 13 is be AUX bracis-28380 39 14 an an DET bracis-28380 39 15 artificial artificial ADJ bracis-28380 39 16 intelligence intelligence NOUN bracis-28380 39 17 approach approach NOUN bracis-28380 39 18 that that PRON bracis-28380 39 19 aims aim VERB bracis-28380 39 20 to to PART bracis-28380 39 21 create create VERB bracis-28380 39 22 transparent transparent ADJ bracis-28380 39 23 and and CCONJ bracis-28380 39 24 interpretable interpretable ADJ bracis-28380 39 25 models model NOUN bracis-28380 39 26 providing provide VERB bracis-28380 39 27 human human ADJ bracis-28380 39 28 - - PUNCT bracis-28380 39 29 understandable understandable ADJ bracis-28380 39 30 explanations explanation NOUN bracis-28380 39 31 for for ADP bracis-28380 39 32 their their PRON bracis-28380 39 33 decisions decision NOUN bracis-28380 39 34 or or CCONJ bracis-28380 39 35 predictions prediction NOUN bracis-28380 39 36 . . PUNCT bracis-28380 40 1 xai xai PROPN bracis-28380 40 2 aims aim VERB bracis-28380 40 3 to to PART bracis-28380 40 4 improve improve VERB bracis-28380 40 5 the the DET bracis-28380 40 6 reliability reliability NOUN bracis-28380 40 7 of of ADP bracis-28380 40 8 ai ai PROPN bracis-28380 40 9 systems system NOUN bracis-28380 40 10 and and CCONJ bracis-28380 40 11 facilitate facilitate NOUN bracis-28380 40 12 collaboration collaboration NOUN bracis-28380 40 13 between between ADP bracis-28380 40 14 humans human NOUN bracis-28380 40 15 and and CCONJ bracis-28380 40 16 machines machine NOUN bracis-28380 40 17 . . PUNCT bracis-28380 41 1 these these DET bracis-28380 41 2 techniques technique NOUN bracis-28380 41 3 include include VERB bracis-28380 41 4 methods method NOUN bracis-28380 41 5 for for ADP bracis-28380 41 6 visualizing visualize VERB bracis-28380 41 7 model model NOUN bracis-28380 41 8 internals internal NOUN bracis-28380 41 9 , , PUNCT bracis-28380 41 10 feature feature VERB bracis-28380 41 11 importance importance NOUN bracis-28380 41 12 analysis analysis NOUN bracis-28380 41 13 , , PUNCT bracis-28380 41 14 and and CCONJ bracis-28380 41 15 rule rule NOUN bracis-28380 41 16 extraction extraction NOUN bracis-28380 41 17 , , PUNCT bracis-28380 41 18 among among ADP bracis-28380 41 19 other other ADJ bracis-28380 41 20 methods method NOUN bracis-28380 41 21 . . PUNCT bracis-28380 42 1 this this DET bracis-28380 42 2 paper paper NOUN bracis-28380 42 3 focuses focus VERB bracis-28380 42 4 on on ADP bracis-28380 42 5 the the DET bracis-28380 42 6 model model NOUN bracis-28380 42 7 visualization visualization NOUN bracis-28380 42 8 technique technique NOUN bracis-28380 42 9 : : PUNCT bracis-28380 42 10 feature feature NOUN bracis-28380 42 11 visualization visualization NOUN bracis-28380 42 12 by by ADP bracis-28380 42 13 layers layer NOUN bracis-28380 42 14 and and CCONJ bracis-28380 42 15 class class NOUN bracis-28380 42 16 activation activation NOUN bracis-28380 42 17 maps map NOUN bracis-28380 42 18 ( ( PUNCT bracis-28380 42 19 cams cam NOUN bracis-28380 42 20 ) ) PUNCT bracis-28380 42 21 . . PUNCT bracis-28380 43 1 visualization visualization NOUN bracis-28380 43 2 and and CCONJ bracis-28380 43 3 cams cam NOUN bracis-28380 43 4 techniques technique NOUN bracis-28380 43 5 are be AUX bracis-28380 43 6 used use VERB bracis-28380 43 7 to to PART bracis-28380 43 8 interpret interpret VERB bracis-28380 43 9 and and CCONJ bracis-28380 43 10 understand understand VERB bracis-28380 43 11 the the DET bracis-28380 43 12 behavior behavior NOUN bracis-28380 43 13 of of ADP bracis-28380 43 14 machine machine NOUN bracis-28380 43 15 learning learning NOUN bracis-28380 43 16 models model NOUN bracis-28380 43 17 . . PUNCT bracis-28380 44 1 feature feature NOUN bracis-28380 44 2 visualization visualization NOUN bracis-28380 44 3 generates generate VERB bracis-28380 44 4 images image NOUN bracis-28380 44 5 that that PRON bracis-28380 44 6 activate activate VERB bracis-28380 44 7 specific specific ADJ bracis-28380 44 8 neurons neuron NOUN bracis-28380 44 9 in in ADP bracis-28380 44 10 a a DET bracis-28380 44 11 neural neural ADJ bracis-28380 44 12 network network NOUN bracis-28380 44 13 , , PUNCT bracis-28380 44 14 allowing allow VERB bracis-28380 44 15 researchers researcher NOUN bracis-28380 44 16 to to PART bracis-28380 44 17 understand understand VERB bracis-28380 44 18 the the DET bracis-28380 44 19 patterns pattern NOUN bracis-28380 44 20 and and CCONJ bracis-28380 44 21 features feature VERB bracis-28380 44 22 that that SCONJ bracis-28380 44 23 the the DET bracis-28380 44 24 network network NOUN bracis-28380 44 25 uses use VERB bracis-28380 44 26 to to PART bracis-28380 44 27 make make VERB bracis-28380 44 28 predictions prediction NOUN bracis-28380 44 29 . . PUNCT bracis-28380 45 1 cams cam NOUN bracis-28380 45 2 identify identify VERB bracis-28380 45 3 the the DET bracis-28380 45 4 important important ADJ bracis-28380 45 5 regions region NOUN bracis-28380 45 6 of of ADP bracis-28380 45 7 an an DET bracis-28380 45 8 input input NOUN bracis-28380 45 9 image image NOUN bracis-28380 45 10 that that PRON bracis-28380 45 11 contribute contribute VERB bracis-28380 45 12 the the DET bracis-28380 45 13 most most ADJ bracis-28380 45 14 to to ADP bracis-28380 45 15 the the DET bracis-28380 45 16 output output NOUN bracis-28380 45 17 of of ADP bracis-28380 45 18 a a DET bracis-28380 45 19 model model NOUN bracis-28380 45 20 , , PUNCT bracis-28380 45 21 which which PRON bracis-28380 45 22 helps help VERB bracis-28380 45 23 understand understand VERB bracis-28380 45 24 which which PRON bracis-28380 45 25 features feature VERB bracis-28380 45 26 the the DET bracis-28380 45 27 model model NOUN bracis-28380 45 28 uses use VERB bracis-28380 45 29 to to PART bracis-28380 45 30 make make VERB bracis-28380 45 31 predictions prediction NOUN bracis-28380 45 32 and and CCONJ bracis-28380 45 33 diagnose diagnose NOUN bracis-28380 45 34 performance performance NOUN bracis-28380 45 35 issues issue NOUN bracis-28380 45 36 . . PUNCT bracis-28380 46 1 both both DET bracis-28380 46 2 techniques technique NOUN bracis-28380 46 3 are be AUX bracis-28380 46 4 valuable valuable ADJ bracis-28380 46 5 tools tool NOUN bracis-28380 46 6 for for ADP bracis-28380 46 7 identifying identify VERB bracis-28380 46 8 areas area NOUN bracis-28380 46 9 for for ADP bracis-28380 46 10 improvement improvement NOUN bracis-28380 46 11 in in ADP bracis-28380 46 12 a a DET bracis-28380 46 13 model model NOUN bracis-28380 46 14 ’s ’s PART bracis-28380 46 15 performance performance NOUN bracis-28380 46 16 . . PUNCT bracis-28380 47 1 by by ADP bracis-28380 47 2 explaining explain VERB bracis-28380 47 3 the the DET bracis-28380 47 4 predictions prediction NOUN bracis-28380 47 5 made make VERB bracis-28380 47 6 by by ADP bracis-28380 47 7 the the DET bracis-28380 47 8 model model NOUN bracis-28380 47 9 , , PUNCT bracis-28380 47 10 xai xai PROPN bracis-28380 47 11 techniques technique NOUN bracis-28380 47 12 increase increase VERB bracis-28380 47 13 transparency transparency NOUN bracis-28380 47 14 and and CCONJ bracis-28380 47 15 reliability reliability NOUN bracis-28380 47 16 , , PUNCT bracis-28380 47 17 allowing allow VERB bracis-28380 47 18 medical medical ADJ bracis-28380 47 19 professionals professional NOUN bracis-28380 47 20 to to PART bracis-28380 47 21 understand understand VERB bracis-28380 47 22 the the DET bracis-28380 47 23 reasoning reasoning NOUN bracis-28380 47 24 behind behind ADP bracis-28380 47 25 the the DET bracis-28380 47 26 diagnosis diagnosis NOUN bracis-28380 47 27 better well ADV bracis-28380 47 28 . . PUNCT bracis-28380 48 1 this this DET bracis-28380 48 2 study study NOUN bracis-28380 48 3 aims aim VERB bracis-28380 48 4 to to PART bracis-28380 48 5 evaluate evaluate VERB bracis-28380 48 6 two two NUM bracis-28380 48 7 cnn cnn PROPN bracis-28380 48 8 models model NOUN bracis-28380 48 9 that that SCONJ bracis-28380 48 10 , , PUNCT bracis-28380 48 11 after after ADP bracis-28380 48 12 using use VERB bracis-28380 48 13 hyperparameter hyperparameter NOUN bracis-28380 48 14 optimization optimization NOUN bracis-28380 48 15 techniques technique NOUN bracis-28380 48 16 and and CCONJ bracis-28380 48 17 data datum NOUN bracis-28380 48 18 augmentation augmentation NOUN bracis-28380 48 19 , , PUNCT bracis-28380 48 20 presented present VERB bracis-28380 48 21 accuracy accuracy NOUN bracis-28380 48 22 results result VERB bracis-28380 48 23 greater great ADJ bracis-28380 48 24 than than ADP bracis-28380 48 25 96 96 NUM bracis-28380 48 26 % % NOUN bracis-28380 48 27 in in ADP bracis-28380 48 28 all all DET bracis-28380 48 29 stages stage NOUN bracis-28380 48 30 ( ( PUNCT bracis-28380 48 31 training training NOUN bracis-28380 48 32 , , PUNCT bracis-28380 48 33 validation validation NOUN bracis-28380 48 34 , , PUNCT bracis-28380 48 35 and and CCONJ bracis-28380 48 36 testing testing NOUN bracis-28380 48 37 ) ) PUNCT bracis-28380 48 38 to to PART bracis-28380 48 39 classify classify VERB bracis-28380 48 40 resonance resonance NOUN bracis-28380 48 41 images image NOUN bracis-28380 48 42 with with ADP bracis-28380 48 43 brain brain NOUN bracis-28380 48 44 tumors tumor NOUN bracis-28380 48 45 and and CCONJ bracis-28380 48 46 demonstrate demonstrate VERB bracis-28380 48 47 the the DET bracis-28380 48 48 characteristics characteristic NOUN bracis-28380 48 49 of of ADP bracis-28380 48 50 the the DET bracis-28380 48 51 networks network NOUN bracis-28380 48 52 considered consider VERB bracis-28380 48 53 for for ADP bracis-28380 48 54 image image NOUN bracis-28380 48 55 classification classification NOUN bracis-28380 48 56 using use VERB bracis-28380 48 57 the the DET bracis-28380 48 58 xai xai PROPN bracis-28380 48 59 technique technique NOUN bracis-28380 48 60 . . PUNCT bracis-28380 49 1 for for ADP bracis-28380 49 2 this this DET bracis-28380 49 3 purpose purpose NOUN bracis-28380 49 4 , , PUNCT bracis-28380 49 5 the the DET bracis-28380 49 6 databases database NOUN bracis-28380 49 7 br35h br35h VERB bracis-28380 49 8 brain brain NOUN bracis-28380 49 9 tumor tumor NOUN bracis-28380 49 10 detection detection NOUN bracis-28380 49 11 [ [ X bracis-28380 49 12 8 8 NUM bracis-28380 49 13 ] ] PUNCT bracis-28380 49 14 and and CCONJ bracis-28380 49 15 the the DET bracis-28380 49 16 brain brain NOUN bracis-28380 49 17 tumor tumor NOUN bracis-28380 49 18 classification classification NOUN bracis-28380 49 19 ( ( PUNCT bracis-28380 49 20 mri mri NOUN bracis-28380 49 21 ) ) PUNCT bracis-28380 49 22 by by ADP bracis-28380 49 23 sartaj sartaj NOUN bracis-28380 49 24 [ [ X bracis-28380 49 25 9 9 NUM bracis-28380 49 26 ] ] PUNCT bracis-28380 49 27 , , PUNCT bracis-28380 49 28 both both CCONJ bracis-28380 49 29 available available ADJ bracis-28380 49 30 in in ADP bracis-28380 49 31 kaggle kaggle PROPN bracis-28380 49 32 , , PUNCT bracis-28380 49 33 will will AUX bracis-28380 49 34 be be AUX bracis-28380 49 35 used use VERB bracis-28380 49 36 . . PUNCT bracis-28380 50 1 this this DET bracis-28380 50 2 paper paper NOUN bracis-28380 50 3 is be AUX bracis-28380 50 4 structured structure VERB bracis-28380 50 5 as as SCONJ bracis-28380 50 6 follows follow VERB bracis-28380 50 7 . . PUNCT bracis-28380 51 1 section section NOUN bracis-28380 51 2     SPACE bracis-28380 51 3 2 2 NUM bracis-28380 51 4 presents present NOUN bracis-28380 51 5 related relate VERB bracis-28380 51 6 works work NOUN bracis-28380 51 7 in in ADP bracis-28380 51 8 deep deep ADJ bracis-28380 51 9 learning learning NOUN bracis-28380 51 10 on on ADP bracis-28380 51 11 medical medical ADJ bracis-28380 51 12 image image NOUN bracis-28380 51 13 analysis analysis NOUN bracis-28380 51 14 and and CCONJ bracis-28380 51 15 explainable explainable ADJ bracis-28380 51 16 artificial artificial ADJ bracis-28380 51 17 intelligence intelligence NOUN bracis-28380 51 18 . . PUNCT bracis-28380 52 1 then then ADV bracis-28380 52 2 , , PUNCT bracis-28380 52 3 sect sect NOUN bracis-28380 52 4 . . PUNCT bracis-28380 52 5     SPACE bracis-28380 52 6 3 3 NUM bracis-28380 52 7 presents present VERB bracis-28380 52 8 the the DET bracis-28380 52 9 dataset dataset NOUN bracis-28380 52 10 , , PUNCT bracis-28380 52 11 pre pre ADJ bracis-28380 52 12 - - ADJ bracis-28380 52 13 processing processing ADJ bracis-28380 52 14 , , PUNCT bracis-28380 52 15 data datum NOUN bracis-28380 52 16 augmentation augmentation NOUN bracis-28380 52 17 , , PUNCT bracis-28380 52 18 and and CCONJ bracis-28380 52 19 the the DET bracis-28380 52 20 networks network NOUN bracis-28380 52 21 ’ ' PUNCT bracis-28380 52 22 structure structure NOUN bracis-28380 52 23 used use VERB bracis-28380 52 24 for for ADP bracis-28380 52 25 training training NOUN bracis-28380 52 26 . . PUNCT bracis-28380 53 1 section section NOUN bracis-28380 53 2     SPACE bracis-28380 53 3 4 4 NUM bracis-28380 53 4 presents present VERB bracis-28380 53 5 the the DET bracis-28380 53 6 results result NOUN bracis-28380 53 7 , , PUNCT bracis-28380 53 8 the the DET bracis-28380 53 9 discussion discussion NOUN bracis-28380 53 10 , , PUNCT bracis-28380 53 11 and and CCONJ bracis-28380 53 12 the the DET bracis-28380 53 13 extracted extract VERB bracis-28380 53 14 xai xai PROPN bracis-28380 53 15 features feature VERB bracis-28380 53 16 for for ADP bracis-28380 53 17 enhancing enhance VERB bracis-28380 53 18 interpretability interpretability NOUN bracis-28380 53 19 . . PUNCT bracis-28380 54 1 section section NOUN bracis-28380 54 2     SPACE bracis-28380 54 3 5 5 NUM bracis-28380 54 4 concludes conclude VERB bracis-28380 54 5 the the DET bracis-28380 54 6 paper paper NOUN bracis-28380 54 7 . . PUNCT bracis-28380 55 1 2 2 NUM bracis-28380 55 2 related relate VERB bracis-28380 55 3 works work NOUN bracis-28380 55 4 explainable explainable ADJ bracis-28380 55 5 artificial artificial ADJ bracis-28380 55 6 intelligence intelligence NOUN bracis-28380 55 7 is be AUX bracis-28380 55 8 a a DET bracis-28380 55 9 technique technique NOUN bracis-28380 55 10 used use VERB bracis-28380 55 11 to to PART bracis-28380 55 12 understand understand VERB bracis-28380 55 13 an an DET bracis-28380 55 14 ai ai NOUN bracis-28380 55 15 model model NOUN bracis-28380 55 16 and and CCONJ bracis-28380 55 17 explain explain VERB bracis-28380 55 18 how how SCONJ bracis-28380 55 19 it it PRON bracis-28380 55 20 arrived arrive VERB bracis-28380 55 21 at at ADP bracis-28380 55 22 its its PRON bracis-28380 55 23 results result NOUN bracis-28380 55 24 or or CCONJ bracis-28380 55 25 decisions decision NOUN bracis-28380 55 26 . . PUNCT bracis-28380 56 1 xai xai PROPN bracis-28380 56 2 is be AUX bracis-28380 56 3 particularly particularly ADV bracis-28380 56 4 important important ADJ bracis-28380 56 5 in in ADP bracis-28380 56 6 deep deep ADJ bracis-28380 56 7 learning learning NOUN bracis-28380 56 8 , , PUNCT bracis-28380 56 9 as as SCONJ bracis-28380 56 10 deep deep ADJ bracis-28380 56 11 neural neural ADJ bracis-28380 56 12 networks network NOUN bracis-28380 56 13 can can AUX bracis-28380 56 14 be be AUX bracis-28380 56 15 very very ADV bracis-28380 56 16 complex complex ADJ bracis-28380 56 17 , , PUNCT bracis-28380 56 18 making make VERB bracis-28380 56 19 it it PRON bracis-28380 56 20 difficult difficult ADJ bracis-28380 56 21 to to PART bracis-28380 56 22 understand understand VERB bracis-28380 56 23 how how SCONJ bracis-28380 56 24 they they PRON bracis-28380 56 25 arrived arrive VERB bracis-28380 56 26 at at ADP bracis-28380 56 27 a a DET bracis-28380 56 28 particular particular ADJ bracis-28380 56 29 decision decision NOUN bracis-28380 56 30 or or CCONJ bracis-28380 56 31 prediction prediction NOUN bracis-28380 56 32 . . PUNCT bracis-28380 57 1 in in ADP bracis-28380 57 2 deep deep ADJ bracis-28380 57 3 learning learning NOUN bracis-28380 57 4 , , PUNCT bracis-28380 57 5 xai xai PROPN bracis-28380 57 6 provides provide VERB bracis-28380 57 7 insight insight NOUN bracis-28380 57 8 into into ADP bracis-28380 57 9 a a DET bracis-28380 57 10 neural neural ADJ bracis-28380 57 11 network network NOUN bracis-28380 57 12 ’s ’s PART bracis-28380 57 13 internal internal ADJ bracis-28380 57 14 workings working NOUN bracis-28380 57 15 . . PUNCT bracis-28380 58 1 this this PRON bracis-28380 58 2 can can AUX bracis-28380 58 3 be be AUX bracis-28380 58 4 achieved achieve VERB bracis-28380 58 5 through through ADP bracis-28380 58 6 various various ADJ bracis-28380 58 7 techniques technique NOUN bracis-28380 58 8 , , PUNCT bracis-28380 58 9 such such ADJ bracis-28380 58 10 as as ADP bracis-28380 58 11 visualizing visualize VERB bracis-28380 58 12 the the DET bracis-28380 58 13 intermediate intermediate ADJ bracis-28380 58 14 layers layer NOUN bracis-28380 58 15 of of ADP bracis-28380 58 16 a a DET bracis-28380 58 17 network network NOUN bracis-28380 58 18 or or CCONJ bracis-28380 58 19 generating generate VERB bracis-28380 58 20 heatmaps heatmap NOUN bracis-28380 58 21 to to PART bracis-28380 58 22 show show VERB bracis-28380 58 23 which which PRON bracis-28380 58 24 parts part NOUN bracis-28380 58 25 of of ADP bracis-28380 58 26 an an DET bracis-28380 58 27 input input NOUN bracis-28380 58 28 image image NOUN bracis-28380 58 29 were be AUX bracis-28380 58 30 most most ADV bracis-28380 58 31 important important ADJ bracis-28380 58 32 in in ADP bracis-28380 58 33 making make VERB bracis-28380 58 34 a a DET bracis-28380 58 35 particular particular ADJ bracis-28380 58 36 prediction prediction NOUN bracis-28380 58 37 . . PUNCT bracis-28380 59 1 xai xai PROPN bracis-28380 59 2 techniques technique NOUN bracis-28380 59 3 are be AUX bracis-28380 59 4 becoming become VERB bracis-28380 59 5 increasingly increasingly ADV bracis-28380 59 6 important important ADJ bracis-28380 59 7 in in ADP bracis-28380 59 8 medical medical ADJ bracis-28380 59 9 diagnosis diagnosis NOUN bracis-28380 59 10 using use VERB bracis-28380 59 11 deep deep ADJ bracis-28380 59 12 learning learning NOUN bracis-28380 59 13 models model NOUN bracis-28380 59 14 , , PUNCT bracis-28380 59 15 as as SCONJ bracis-28380 59 16 they they PRON bracis-28380 59 17 can can AUX bracis-28380 59 18 help help VERB bracis-28380 59 19 improve improve VERB bracis-28380 59 20 the the DET bracis-28380 59 21 interpretability interpretability NOUN bracis-28380 59 22 and and CCONJ bracis-28380 59 23 trustworthiness trustworthiness NOUN bracis-28380 59 24 of of ADP bracis-28380 59 25 these these DET bracis-28380 59 26 models model NOUN bracis-28380 59 27 and and CCONJ bracis-28380 59 28 improve improve VERB bracis-28380 59 29 patient patient ADJ bracis-28380 59 30 outcomes outcome NOUN bracis-28380 59 31 . . PUNCT bracis-28380 60 1 medical medical ADJ bracis-28380 60 2 diagnostic diagnostic ADJ bracis-28380 60 3 support support NOUN bracis-28380 60 4 has have AUX bracis-28380 60 5 shown show VERB bracis-28380 60 6 great great ADJ bracis-28380 60 7 promise promise NOUN bracis-28380 60 8 with with ADP bracis-28380 60 9 them they PRON bracis-28380 60 10 in in ADP bracis-28380 60 11 identifying identify VERB bracis-28380 60 12 and and CCONJ bracis-28380 60 13 classifying classify VERB bracis-28380 60 14 various various ADJ bracis-28380 60 15 medical medical ADJ bracis-28380 60 16 conditions condition NOUN bracis-28380 60 17 based base VERB bracis-28380 60 18 on on ADP bracis-28380 60 19 medical medical ADJ bracis-28380 60 20 imaging imaging NOUN bracis-28380 60 21 data datum NOUN bracis-28380 60 22 , , PUNCT bracis-28380 60 23 such such ADJ bracis-28380 60 24 as as ADP bracis-28380 60 25 x x NOUN bracis-28380 60 26 - - NOUN bracis-28380 60 27 rays ray NOUN bracis-28380 60 28 , , PUNCT bracis-28380 60 29 mri mri NOUN bracis-28380 60 30 , , PUNCT bracis-28380 60 31 and and CCONJ bracis-28380 60 32 computed compute VERB bracis-28380 60 33 tomography tomography NOUN bracis-28380 60 34 scans scan NOUN bracis-28380 60 35 . . PUNCT bracis-28380 61 1 however however ADV bracis-28380 61 2 , , PUNCT bracis-28380 61 3 they they PRON bracis-28380 61 4 can can AUX bracis-28380 61 5 be be AUX bracis-28380 61 6 highly highly ADV bracis-28380 61 7 complex complex ADJ bracis-28380 61 8 . . PUNCT bracis-28380 62 1 this this DET bracis-28380 62 2 section section NOUN bracis-28380 62 3 reports report VERB bracis-28380 62 4 some some DET bracis-28380 62 5 published publish VERB bracis-28380 62 6 research research NOUN bracis-28380 62 7 on on ADP bracis-28380 62 8 the the DET bracis-28380 62 9 question question NOUN bracis-28380 62 10 of of ADP bracis-28380 62 11 this this DET bracis-28380 62 12 article article NOUN bracis-28380 62 13 : : PUNCT bracis-28380 62 14 are be AUX bracis-28380 62 15 the the DET bracis-28380 62 16 results result NOUN bracis-28380 62 17 obtained obtain VERB bracis-28380 62 18 through through ADP bracis-28380 62 19 cnns cnn NOUN bracis-28380 62 20 reliable reliable ADJ bracis-28380 62 21 for for ADP bracis-28380 62 22 supporting support VERB bracis-28380 62 23 the the DET bracis-28380 62 24 medical medical ADJ bracis-28380 62 25 diagnosis diagnosis NOUN bracis-28380 62 26 ? ? PUNCT bracis-28380 63 1 the the DET bracis-28380 63 2 paper paper NOUN bracis-28380 63 3 [ [ X bracis-28380 63 4 10 10 NUM bracis-28380 63 5 ] ] PUNCT bracis-28380 63 6 discusses discuss VERB bracis-28380 63 7 the the DET bracis-28380 63 8 need need NOUN bracis-28380 63 9 for for ADP bracis-28380 63 10 a a DET bracis-28380 63 11 more more ADV bracis-28380 63 12 scientific scientific ADJ bracis-28380 63 13 approach approach NOUN bracis-28380 63 14 to to ADP bracis-28380 63 15 interpretability interpretability NOUN bracis-28380 63 16 in in ADP bracis-28380 63 17 machine machine NOUN bracis-28380 63 18 learning learning NOUN bracis-28380 63 19 . . PUNCT bracis-28380 64 1 they they PRON bracis-28380 64 2 propose propose VERB bracis-28380 64 3 a a DET bracis-28380 64 4 framework framework NOUN bracis-28380 64 5 for for ADP bracis-28380 64 6 characterizing characterize VERB bracis-28380 64 7 different different ADJ bracis-28380 64 8 types type NOUN bracis-28380 64 9 of of ADP bracis-28380 64 10 interpretability interpretability NOUN bracis-28380 64 11 and and CCONJ bracis-28380 64 12 identify identify VERB bracis-28380 64 13 research research NOUN bracis-28380 64 14 challenges challenge NOUN bracis-28380 64 15 in in ADP bracis-28380 64 16 developing develop VERB bracis-28380 64 17 a a DET bracis-28380 64 18 rigorous rigorous ADJ bracis-28380 64 19 science science NOUN bracis-28380 64 20 of of ADP bracis-28380 64 21 interpretable interpretable ADJ bracis-28380 64 22 machine machine NOUN bracis-28380 64 23 learning learning NOUN bracis-28380 64 24 . . PUNCT bracis-28380 65 1 the the DET bracis-28380 65 2 authors author NOUN bracis-28380 65 3 discuss discuss VERB bracis-28380 65 4 potential potential ADJ bracis-28380 65 5 future future ADJ bracis-28380 65 6 directions direction NOUN bracis-28380 65 7 for for ADP bracis-28380 65 8 research research NOUN bracis-28380 65 9 in in ADP bracis-28380 65 10 this this DET bracis-28380 65 11 area area NOUN bracis-28380 65 12 , , PUNCT bracis-28380 65 13 such such ADJ bracis-28380 65 14 as as ADP bracis-28380 65 15 developing develop VERB bracis-28380 65 16 new new ADJ bracis-28380 65 17 evaluation evaluation NOUN bracis-28380 65 18 metrics metric NOUN bracis-28380 65 19 and and CCONJ bracis-28380 65 20 exploring explore VERB bracis-28380 65 21 the the DET bracis-28380 65 22 ethical ethical ADJ bracis-28380 65 23 implications implication NOUN bracis-28380 65 24 of of ADP bracis-28380 65 25 interpretability interpretability NOUN bracis-28380 65 26 . . PUNCT bracis-28380 66 1 the the DET bracis-28380 66 2 paper paper NOUN bracis-28380 66 3 emphasizes emphasize VERB bracis-28380 66 4 the the DET bracis-28380 66 5 importance importance NOUN bracis-28380 66 6 of of ADP bracis-28380 66 7 interpretability interpretability NOUN bracis-28380 66 8 as as ADP bracis-28380 66 9 a a DET bracis-28380 66 10 critical critical ADJ bracis-28380 66 11 component component NOUN bracis-28380 66 12 of of ADP bracis-28380 66 13 responsible responsible ADJ bracis-28380 66 14 and and CCONJ bracis-28380 66 15 trustworthy trustworthy ADJ bracis-28380 66 16 machine machine NOUN bracis-28380 66 17 learning learning NOUN bracis-28380 66 18 systems system NOUN bracis-28380 66 19 . . PUNCT bracis-28380 67 1 in in ADP bracis-28380 67 2 [ [ X bracis-28380 67 3 11 11 NUM bracis-28380 67 4 ] ] PUNCT bracis-28380 67 5 , , PUNCT bracis-28380 67 6 the the DET bracis-28380 67 7 authors author NOUN bracis-28380 67 8 propose propose VERB bracis-28380 67 9 a a DET bracis-28380 67 10 method method NOUN bracis-28380 67 11 that that PRON bracis-28380 67 12 uses use VERB bracis-28380 67 13 a a DET bracis-28380 67 14 combination combination NOUN bracis-28380 67 15 of of ADP bracis-28380 67 16 deep deep ADJ bracis-28380 67 17 neural neural ADJ bracis-28380 67 18 networks network NOUN bracis-28380 67 19 ( ( PUNCT bracis-28380 67 20 dnns dnn NOUN bracis-28380 67 21 ) ) PUNCT bracis-28380 67 22 and and CCONJ bracis-28380 67 23 multiclass multiclass ADJ bracis-28380 67 24 support support NOUN bracis-28380 67 25 vector vector NOUN bracis-28380 67 26 machines machine NOUN bracis-28380 67 27 ( ( PUNCT bracis-28380 67 28 svms svms NOUN bracis-28380 67 29 ) ) PUNCT bracis-28380 67 30 to to PART bracis-28380 67 31 improve improve VERB bracis-28380 67 32 the the DET bracis-28380 67 33 accuracy accuracy NOUN bracis-28380 67 34 and and CCONJ bracis-28380 67 35 interpretability interpretability NOUN bracis-28380 67 36 of of ADP bracis-28380 67 37 brain brain NOUN bracis-28380 67 38 tumor tumor NOUN bracis-28380 67 39 diagnosis diagnosis NOUN bracis-28380 67 40 . . PUNCT bracis-28380 68 1 to to PART bracis-28380 68 2 improve improve VERB bracis-28380 68 3 the the DET bracis-28380 68 4 interpretability interpretability NOUN bracis-28380 68 5 of of ADP bracis-28380 68 6 their their PRON bracis-28380 68 7 model model NOUN bracis-28380 68 8 , , PUNCT bracis-28380 68 9 the the DET bracis-28380 68 10 authors author NOUN bracis-28380 68 11 use use VERB bracis-28380 68 12 explainable explainable ADJ bracis-28380 68 13 artificial artificial ADJ bracis-28380 68 14 intelligence intelligence NOUN bracis-28380 68 15 ( ( PUNCT bracis-28380 68 16 xai xai PROPN bracis-28380 68 17 ) ) PUNCT bracis-28380 68 18 techniques technique NOUN bracis-28380 68 19 , , PUNCT bracis-28380 68 20 including include VERB bracis-28380 68 21 feature feature NOUN bracis-28380 68 22 visualization visualization NOUN bracis-28380 68 23 and and CCONJ bracis-28380 68 24 salience salience NOUN bracis-28380 68 25 maps map NOUN bracis-28380 68 26 , , PUNCT bracis-28380 68 27 to to PART bracis-28380 68 28 identify identify VERB bracis-28380 68 29 the the DET bracis-28380 68 30 key key ADJ bracis-28380 68 31 features feature NOUN bracis-28380 68 32 and and CCONJ bracis-28380 68 33 regions region NOUN bracis-28380 68 34 of of ADP bracis-28380 68 35 the the DET bracis-28380 68 36 input input NOUN bracis-28380 68 37 images image NOUN bracis-28380 68 38 most most ADV bracis-28380 68 39 relevant relevant ADJ bracis-28380 68 40 for for ADP bracis-28380 68 41 tumor tumor NOUN bracis-28380 68 42 detection detection NOUN bracis-28380 68 43 . . PUNCT bracis-28380 69 1 they they PRON bracis-28380 69 2 also also ADV bracis-28380 69 3 compare compare VERB bracis-28380 69 4 the the DET bracis-28380 69 5 performance performance NOUN bracis-28380 69 6 of of ADP bracis-28380 69 7 their their PRON bracis-28380 69 8 model model NOUN bracis-28380 69 9 to to ADP bracis-28380 69 10 other other ADJ bracis-28380 69 11 state state NOUN bracis-28380 69 12 - - PUNCT bracis-28380 69 13 of of ADP bracis-28380 69 14 - - PUNCT bracis-28380 69 15 the the DET bracis-28380 69 16 - - PUNCT bracis-28380 69 17 art art NOUN bracis-28380 69 18 brain brain NOUN bracis-28380 69 19 tumor tumor NOUN bracis-28380 69 20 detection detection NOUN bracis-28380 69 21 methods method NOUN bracis-28380 69 22 . . PUNCT bracis-28380 70 1 the the DET bracis-28380 70 2 authors author NOUN bracis-28380 70 3 demonstrate demonstrate VERB bracis-28380 70 4 that that SCONJ bracis-28380 70 5 the the DET bracis-28380 70 6 proposed propose VERB bracis-28380 70 7 method method NOUN bracis-28380 70 8 achieves achieve VERB bracis-28380 70 9 high high ADJ bracis-28380 70 10 accuracy accuracy NOUN bracis-28380 70 11 in in ADP bracis-28380 70 12 detecting detect VERB bracis-28380 70 13 brain brain NOUN bracis-28380 70 14 tumors tumor NOUN bracis-28380 70 15 greater great ADJ bracis-28380 70 16 than than ADP bracis-28380 70 17 97 97 NUM bracis-28380 70 18 % % NOUN bracis-28380 70 19 and and CCONJ bracis-28380 70 20 that that SCONJ bracis-28380 70 21 using use VERB bracis-28380 70 22 xai xai PROPN bracis-28380 70 23 techniques technique NOUN bracis-28380 70 24 improves improve VERB bracis-28380 70 25 the the DET bracis-28380 70 26 interpretability interpretability NOUN bracis-28380 70 27 of of ADP bracis-28380 70 28 their their PRON bracis-28380 70 29 model model NOUN bracis-28380 70 30 and and CCONJ bracis-28380 70 31 provides provide VERB bracis-28380 70 32 information information NOUN bracis-28380 70 33 about about ADP bracis-28380 70 34 the the DET bracis-28380 70 35 underlying underlie VERB bracis-28380 70 36 characteristics characteristic NOUN bracis-28380 70 37 of of ADP bracis-28380 70 38 brain brain NOUN bracis-28380 70 39 tumors tumor NOUN bracis-28380 70 40 . . PUNCT bracis-28380 71 1 the the DET bracis-28380 71 2 authors author NOUN bracis-28380 71 3 tested test VERB bracis-28380 71 4 the the DET bracis-28380 71 5 model model NOUN bracis-28380 71 6 using use VERB bracis-28380 71 7 two two NUM bracis-28380 71 8 databases database NOUN bracis-28380 71 9 ( ( PUNCT bracis-28380 71 10 figshare figshare NOUN bracis-28380 71 11 and and CCONJ bracis-28380 71 12 brats brat NOUN bracis-28380 71 13 2018 2018 NUM bracis-28380 71 14 ) ) PUNCT bracis-28380 71 15 . . PUNCT bracis-28380 72 1 the the DET bracis-28380 72 2 paper paper NOUN bracis-28380 72 3 highlights highlight VERB bracis-28380 72 4 the the DET bracis-28380 72 5 potential potential NOUN bracis-28380 72 6 of of ADP bracis-28380 72 7 combining combine VERB bracis-28380 72 8 deep deep ADJ bracis-28380 72 9 learning learning NOUN bracis-28380 72 10 , , PUNCT bracis-28380 72 11 machine machine NOUN bracis-28380 72 12 learning learning NOUN bracis-28380 72 13 , , PUNCT bracis-28380 72 14 and and CCONJ bracis-28380 72 15 xai xai PROPN bracis-28380 72 16 techniques technique NOUN bracis-28380 72 17 to to PART bracis-28380 72 18 improve improve VERB bracis-28380 72 19 the the DET bracis-28380 72 20 accuracy accuracy NOUN bracis-28380 72 21 and and CCONJ bracis-28380 72 22 interpretability interpretability NOUN bracis-28380 72 23 of of ADP bracis-28380 72 24 brain brain NOUN bracis-28380 72 25 tumor tumor NOUN bracis-28380 72 26 diagnosis diagnosis NOUN bracis-28380 72 27 . . PUNCT bracis-28380 73 1 in in ADP bracis-28380 73 2 the the DET bracis-28380 73 3 study study NOUN bracis-28380 73 4 [ [ X bracis-28380 73 5 12 12 NUM bracis-28380 73 6 ] ] PUNCT bracis-28380 73 7 , , PUNCT bracis-28380 73 8 seven seven NUM bracis-28380 73 9 cnn cnn PROPN bracis-28380 73 10 models model NOUN bracis-28380 73 11 are be AUX bracis-28380 73 12 evaluated evaluate VERB bracis-28380 73 13 for for ADP bracis-28380 73 14 the the DET bracis-28380 73 15 classification classification NOUN bracis-28380 73 16 of of ADP bracis-28380 73 17 brain brain NOUN bracis-28380 73 18 tumors tumor NOUN bracis-28380 73 19 . . PUNCT bracis-28380 74 1 the the DET bracis-28380 74 2 models model NOUN bracis-28380 74 3 considered consider VERB bracis-28380 74 4 include include VERB bracis-28380 74 5 a a DET bracis-28380 74 6 generic generic ADJ bracis-28380 74 7 cnn cnn NOUN bracis-28380 74 8 model model NOUN bracis-28380 74 9 and and CCONJ bracis-28380 74 10 six six NUM bracis-28380 74 11 pre pre ADJ bracis-28380 74 12 - - ADJ bracis-28380 74 13 trained train VERB bracis-28380 74 14 cnn cnn PROPN bracis-28380 74 15 models model NOUN bracis-28380 74 16 . . PUNCT bracis-28380 75 1 the the DET bracis-28380 75 2 dataset dataset NOUN bracis-28380 75 3 used use VERB bracis-28380 75 4 for for ADP bracis-28380 75 5 this this DET bracis-28380 75 6 research research NOUN bracis-28380 75 7 is be AUX bracis-28380 75 8 called call VERB bracis-28380 75 9 msoud msoud ADJ bracis-28380 75 10 , , PUNCT bracis-28380 75 11 which which PRON bracis-28380 75 12 comprises comprise VERB bracis-28380 75 13 three three NUM bracis-28380 75 14 existing exist VERB bracis-28380 75 15 datasets dataset NOUN bracis-28380 75 16 : : PUNCT bracis-28380 75 17 fighshare fighshare NOUN bracis-28380 75 18 , , PUNCT bracis-28380 75 19 sartaj sartaj NOUN bracis-28380 75 20 , , PUNCT bracis-28380 75 21 and and CCONJ bracis-28380 75 22 br35h br35h VERB bracis-28380 75 23 . . PUNCT bracis-28380 76 1 this this DET bracis-28380 76 2 combined combine VERB bracis-28380 76 3 dataset dataset NOUN bracis-28380 76 4 consists consist NOUN bracis-28380 76 5 of of ADP bracis-28380 76 6 7023 7023 NUM bracis-28380 76 7 mri mri NOUN bracis-28380 76 8 images image NOUN bracis-28380 76 9 . . PUNCT bracis-28380 77 1 the the DET bracis-28380 77 2 study study NOUN bracis-28380 77 3 results result NOUN bracis-28380 77 4 indicate indicate VERB bracis-28380 77 5 that that SCONJ bracis-28380 77 6 the the DET bracis-28380 77 7 inceptionv3 inceptionv3 NOUN bracis-28380 77 8 model model NOUN bracis-28380 77 9 achieves achieve VERB bracis-28380 77 10 the the DET bracis-28380 77 11 highest high ADJ bracis-28380 77 12 accuracy accuracy NOUN bracis-28380 77 13 among among ADP bracis-28380 77 14 all all DET bracis-28380 77 15 evaluated evaluate VERB bracis-28380 77 16 cnn cnn PROPN bracis-28380 77 17 models model NOUN bracis-28380 77 18 . . PUNCT bracis-28380 78 1 it it PRON bracis-28380 78 2 achieves achieve VERB bracis-28380 78 3 an an DET bracis-28380 78 4 average average ADJ bracis-28380 78 5 accuracy accuracy NOUN bracis-28380 78 6 of of ADP bracis-28380 78 7 97.12 97.12 NUM bracis-28380 78 8 % % NOUN bracis-28380 78 9 on on ADP bracis-28380 78 10 the the DET bracis-28380 78 11 msoud msoud ADJ bracis-28380 78 12 dataset dataset NOUN bracis-28380 78 13 . . PUNCT bracis-28380 79 1 the the DET bracis-28380 79 2 authors author NOUN bracis-28380 79 3 used use VERB bracis-28380 79 4 the the DET bracis-28380 79 5 three three NUM bracis-28380 79 6 banks bank NOUN bracis-28380 79 7 together together ADV bracis-28380 79 8 ( ( PUNCT bracis-28380 79 9 80 80 NUM bracis-28380 79 10 % % NOUN bracis-28380 79 11 for for ADP bracis-28380 79 12 training training NOUN bracis-28380 79 13 and and CCONJ bracis-28380 79 14 20 20 NUM bracis-28380 79 15 % % NOUN bracis-28380 79 16 for for ADP bracis-28380 79 17 testing testing NOUN bracis-28380 79 18 ) ) PUNCT bracis-28380 79 19 . . PUNCT bracis-28380 80 1 the the DET bracis-28380 80 2 paper paper NOUN bracis-28380 80 3 focused focus VERB bracis-28380 80 4 on on ADP bracis-28380 80 5 showing show VERB bracis-28380 80 6 the the DET bracis-28380 80 7 performance performance NOUN bracis-28380 80 8 of of ADP bracis-28380 80 9 the the DET bracis-28380 80 10 models model NOUN bracis-28380 80 11 through through ADP bracis-28380 80 12 metrics metric NOUN bracis-28380 80 13 based base VERB bracis-28380 80 14 on on ADP bracis-28380 80 15 values value NOUN bracis-28380 80 16 , , PUNCT bracis-28380 80 17 even even ADV bracis-28380 80 18 though though SCONJ bracis-28380 80 19 it it PRON bracis-28380 80 20 is be AUX bracis-28380 80 21 a a DET bracis-28380 80 22 recent recent ADJ bracis-28380 80 23 study study NOUN bracis-28380 80 24 published publish VERB bracis-28380 80 25 in in ADP bracis-28380 80 26 2023 2023 NUM bracis-28380 80 27 . . PUNCT bracis-28380 81 1 however however ADV bracis-28380 81 2 , , PUNCT bracis-28380 81 3 there there PRON bracis-28380 81 4 has have AUX bracis-28380 81 5 been be AUX bracis-28380 81 6 a a DET bracis-28380 81 7 growing grow VERB bracis-28380 81 8 tendency tendency NOUN bracis-28380 81 9 to to PART bracis-28380 81 10 incorporate incorporate VERB bracis-28380 81 11 explainability explainability NOUN bracis-28380 81 12 techniques technique NOUN bracis-28380 81 13 for for ADP bracis-28380 81 14 greater great ADJ bracis-28380 81 15 transparency transparency NOUN bracis-28380 81 16 of of ADP bracis-28380 81 17 the the DET bracis-28380 81 18 models model NOUN bracis-28380 81 19 , , PUNCT bracis-28380 81 20 in in ADP bracis-28380 81 21 addition addition NOUN bracis-28380 81 22 to to ADP bracis-28380 81 23 the the DET bracis-28380 81 24 traditional traditional ADJ bracis-28380 81 25 metrics metric NOUN bracis-28380 81 26 in in ADP bracis-28380 81 27 the the DET bracis-28380 81 28 last last ADJ bracis-28380 81 29 few few ADJ bracis-28380 81 30 years year NOUN bracis-28380 81 31 . . PUNCT bracis-28380 82 1 in in ADP bracis-28380 82 2 [ [ X bracis-28380 82 3 13 13 NUM bracis-28380 82 4 ] ] PUNCT bracis-28380 82 5 , , PUNCT bracis-28380 82 6 the the DET bracis-28380 82 7 authors author NOUN bracis-28380 82 8 propose propose VERB bracis-28380 82 9 an an DET bracis-28380 82 10 explanation explanation NOUN bracis-28380 82 11 - - PUNCT bracis-28380 82 12 driven drive VERB bracis-28380 82 13 dl dl PROPN bracis-28380 82 14 model model NOUN bracis-28380 82 15 that that PRON bracis-28380 82 16 utilizes utilize VERB bracis-28380 82 17 a a DET bracis-28380 82 18 convolutional convolutional ADJ bracis-28380 82 19 neural neural ADJ bracis-28380 82 20 network network NOUN bracis-28380 82 21 , , PUNCT bracis-28380 82 22 local local ADJ bracis-28380 82 23 interpretable interpretable ADJ bracis-28380 82 24 model model ADJ bracis-28380 82 25 - - ADJ bracis-28380 82 26 agnostic agnostic ADJ bracis-28380 82 27 explanation explanation NOUN bracis-28380 82 28 ( ( PUNCT bracis-28380 82 29 lime lime NOUN bracis-28380 82 30 ) ) PUNCT bracis-28380 82 31 , , PUNCT bracis-28380 82 32 and and CCONJ bracis-28380 82 33 shap shap NOUN bracis-28380 82 34 for for ADP bracis-28380 82 35 predicting predict VERB bracis-28380 82 36 discrete discrete ADJ bracis-28380 82 37 subtypes subtype NOUN bracis-28380 82 38 of of ADP bracis-28380 82 39 brain brain NOUN bracis-28380 82 40 tumors tumor NOUN bracis-28380 82 41 ( ( PUNCT bracis-28380 82 42 meningioma meningioma NOUN bracis-28380 82 43 , , PUNCT bracis-28380 82 44 glioma glioma NOUN bracis-28380 82 45 , , PUNCT bracis-28380 82 46 and and CCONJ bracis-28380 82 47 pituitary pituitary NOUN bracis-28380 82 48 ) ) PUNCT bracis-28380 82 49 using use VERB bracis-28380 82 50 an an DET bracis-28380 82 51 mri mri NOUN bracis-28380 82 52 image image NOUN bracis-28380 82 53 dataset dataset NOUN bracis-28380 82 54 . . PUNCT bracis-28380 83 1 unlike unlike ADP bracis-28380 83 2 previous previous ADJ bracis-28380 83 3 models model NOUN bracis-28380 83 4 , , PUNCT bracis-28380 83 5 the the DET bracis-28380 83 6 proposed propose VERB bracis-28380 83 7 model model NOUN bracis-28380 83 8 uses use VERB bracis-28380 83 9 a a DET bracis-28380 83 10 dual dual ADJ bracis-28380 83 11 - - PUNCT bracis-28380 83 12 input input NOUN bracis-28380 83 13 cnn cnn NOUN bracis-28380 83 14 approach approach NOUN bracis-28380 83 15 to to PART bracis-28380 83 16 overcome overcome VERB bracis-28380 83 17 classification classification NOUN bracis-28380 83 18 challenges challenge NOUN bracis-28380 83 19 posed pose VERB bracis-28380 83 20 by by ADP bracis-28380 83 21 low low ADJ bracis-28380 83 22 - - PUNCT bracis-28380 83 23 quality quality NOUN bracis-28380 83 24 images image NOUN bracis-28380 83 25 with with ADP bracis-28380 83 26 noise noise NOUN bracis-28380 83 27 and and CCONJ bracis-28380 83 28 metal metal NOUN bracis-28380 83 29 artifacts artifact NOUN bracis-28380 83 30 by by ADP bracis-28380 83 31 adding add VERB bracis-28380 83 32 gaussian gaussian ADJ bracis-28380 83 33 noise noise NOUN bracis-28380 83 34 . . PUNCT bracis-28380 84 1 the the DET bracis-28380 84 2 cnn cnn PROPN bracis-28380 84 3 training training NOUN bracis-28380 84 4 results result NOUN bracis-28380 84 5 demonstrate demonstrate VERB bracis-28380 84 6 94.64 94.64 NUM bracis-28380 84 7 % % NOUN bracis-28380 84 8 accuracy accuracy NOUN bracis-28380 84 9 , , PUNCT bracis-28380 84 10 outperforming outperform VERB bracis-28380 84 11 other other ADJ bracis-28380 84 12 state state NOUN bracis-28380 84 13 - - PUNCT bracis-28380 84 14 of of ADP bracis-28380 84 15 - - PUNCT bracis-28380 84 16 the the DET bracis-28380 84 17 - - PUNCT bracis-28380 84 18 art art NOUN bracis-28380 84 19 methods method NOUN bracis-28380 84 20 . . PUNCT bracis-28380 85 1 to to PART bracis-28380 85 2 ensure ensure VERB bracis-28380 85 3 consistency consistency NOUN bracis-28380 85 4 and and CCONJ bracis-28380 85 5 local local ADJ bracis-28380 85 6 accuracy accuracy NOUN bracis-28380 85 7 for for ADP bracis-28380 85 8 interpretation interpretation NOUN bracis-28380 85 9 , , PUNCT bracis-28380 85 10 the the DET bracis-28380 85 11 authors author NOUN bracis-28380 85 12 employ employ VERB bracis-28380 85 13 shapley shapley ADJ bracis-28380 85 14 additive additive ADJ bracis-28380 85 15 explanations explanation NOUN bracis-28380 85 16 ( ( PUNCT bracis-28380 85 17 shap shap NOUN bracis-28380 85 18 ) ) PUNCT bracis-28380 85 19 , , PUNCT bracis-28380 85 20 which which PRON bracis-28380 85 21 examines examine VERB bracis-28380 85 22 all all DET bracis-28380 85 23 future future ADJ bracis-28380 85 24 predictions prediction NOUN bracis-28380 85 25 applying apply VERB bracis-28380 85 26 all all DET bracis-28380 85 27 possible possible ADJ bracis-28380 85 28 combinations combination NOUN bracis-28380 85 29 of of ADP bracis-28380 85 30 inputs input NOUN bracis-28380 85 31 , , PUNCT bracis-28380 85 32 and and CCONJ bracis-28380 85 33 lime lime NOUN bracis-28380 85 34 , , PUNCT bracis-28380 85 35 which which PRON bracis-28380 85 36 constructs construct VERB bracis-28380 85 37 sparse sparse ADJ bracis-28380 85 38 linear linear ADJ bracis-28380 85 39 models model NOUN bracis-28380 85 40 around around ADP bracis-28380 85 41 each each DET bracis-28380 85 42 prediction prediction NOUN bracis-28380 85 43 to to PART bracis-28380 85 44 illustrate illustrate VERB bracis-28380 85 45 how how SCONJ bracis-28380 85 46 the the DET bracis-28380 85 47 model model NOUN bracis-28380 85 48 operates operate VERB bracis-28380 85 49 in in ADP bracis-28380 85 50 the the DET bracis-28380 85 51 immediate immediate ADJ bracis-28380 85 52 area area NOUN bracis-28380 85 53 . . PUNCT bracis-28380 86 1 the the DET bracis-28380 86 2 study study NOUN bracis-28380 86 3 [ [ X bracis-28380 86 4 14 14 NUM bracis-28380 86 5 ] ] PUNCT bracis-28380 86 6 proposes propose VERB bracis-28380 86 7 a a DET bracis-28380 86 8 convolutional convolutional ADJ bracis-28380 86 9 neural neural ADJ bracis-28380 86 10 network network NOUN bracis-28380 86 11 ( ( PUNCT bracis-28380 86 12 cnn cnn PROPN bracis-28380 86 13 ) ) PUNCT bracis-28380 86 14 model model NOUN bracis-28380 86 15 with with ADP bracis-28380 86 16 a a DET bracis-28380 86 17 new new ADJ bracis-28380 86 18 form form NOUN bracis-28380 86 19 of of ADP bracis-28380 86 20 grad grad NOUN bracis-28380 86 21 - - PUNCT bracis-28380 86 22 cam cam NOUN bracis-28380 86 23 called call VERB bracis-28380 86 24 numerical numerical PROPN bracis-28380 86 25 grad grad PROPN bracis-28380 86 26 - - PUNCT bracis-28380 86 27 cam cam PROPN bracis-28380 86 28 ( ( PUNCT bracis-28380 86 29 numgrad numgrad NOUN bracis-28380 86 30 - - PUNCT bracis-28380 86 31 cam cam NOUN bracis-28380 86 32 ) ) PUNCT bracis-28380 86 33 to to PART bracis-28380 86 34 provide provide VERB bracis-28380 86 35 a a DET bracis-28380 86 36 user user NOUN bracis-28380 86 37 - - PUNCT bracis-28380 86 38 friendly friendly ADJ bracis-28380 86 39 explainability explainability NOUN bracis-28380 86 40 interface interface NOUN bracis-28380 86 41 for for ADP bracis-28380 86 42 brain brain NOUN bracis-28380 86 43 tumor tumor NOUN bracis-28380 86 44 diagnosis diagnosis NOUN bracis-28380 86 45 . . PUNCT bracis-28380 87 1 the the DET bracis-28380 87 2 numgrad numgrad NOUN bracis-28380 87 3 - - PUNCT bracis-28380 87 4 cam cam NOUN bracis-28380 87 5 - - PUNCT bracis-28380 87 6 cnn cnn PROPN bracis-28380 87 7 model model NOUN bracis-28380 87 8 was be AUX bracis-28380 87 9 evaluated evaluate VERB bracis-28380 87 10 using use VERB bracis-28380 87 11 technical technical ADJ bracis-28380 87 12 and and CCONJ bracis-28380 87 13 physician physician NOUN bracis-28380 87 14 - - PUNCT bracis-28380 87 15 oriented orient VERB bracis-28380 87 16 ( ( PUNCT bracis-28380 87 17 human human ADJ bracis-28380 87 18 - - PUNCT bracis-28380 87 19 side side NOUN bracis-28380 87 20 ) ) PUNCT bracis-28380 87 21 evaluations evaluation NOUN bracis-28380 87 22 and and CCONJ bracis-28380 87 23 achieved achieve VERB bracis-28380 87 24 an an DET bracis-28380 87 25 average average ADJ bracis-28380 87 26 accuracy accuracy NOUN bracis-28380 87 27 of of ADP bracis-28380 87 28 97.11 97.11 NUM bracis-28380 87 29 % % NOUN bracis-28380 87 30 a a DET bracis-28380 87 31 sensitivity sensitivity NOUN bracis-28380 87 32 of of ADP bracis-28380 87 33 95.58 95.58 NUM bracis-28380 87 34 % % NOUN bracis-28380 87 35 and and CCONJ bracis-28380 87 36 a a DET bracis-28380 87 37 specificity specificity NOUN bracis-28380 87 38 of of ADP bracis-28380 87 39 96.81 96.81 NUM bracis-28380 87 40 % % NOUN bracis-28380 87 41 for for ADP bracis-28380 87 42 the the DET bracis-28380 87 43 targeted target VERB bracis-28380 87 44 brain brain NOUN bracis-28380 87 45 tumor tumor NOUN bracis-28380 87 46 diagnosis diagnosis NOUN bracis-28380 87 47 setup setup NOUN bracis-28380 87 48 . . PUNCT bracis-28380 88 1 moreover moreover ADV bracis-28380 88 2 , , PUNCT bracis-28380 88 3 the the DET bracis-28380 88 4 integration integration NOUN bracis-28380 88 5 of of ADP bracis-28380 88 6 numgrad numgrad NOUN bracis-28380 88 7 - - PUNCT bracis-28380 88 8 cam cam NOUN bracis-28380 88 9 provided provide VERB bracis-28380 88 10 an an DET bracis-28380 88 11 accuracy accuracy NOUN bracis-28380 88 12 of of ADP bracis-28380 88 13 90.11 90.11 NUM bracis-28380 88 14 % % NOUN bracis-28380 88 15 compared compare VERB bracis-28380 88 16 to to ADP bracis-28380 88 17 other other ADJ bracis-28380 88 18 cam cam NOUN bracis-28380 88 19 variations variation NOUN bracis-28380 88 20 in in ADP bracis-28380 88 21 the the DET bracis-28380 88 22 same same ADJ bracis-28380 88 23 cnn cnn PROPN bracis-28380 88 24 model model NOUN bracis-28380 88 25 . . PUNCT bracis-28380 89 1 physicians physician NOUN bracis-28380 89 2 who who PRON bracis-28380 89 3 used use VERB bracis-28380 89 4 the the DET bracis-28380 89 5 numgrad numgrad NOUN bracis-28380 89 6 - - PUNCT bracis-28380 89 7 cam cam NOUN bracis-28380 89 8 - - PUNCT bracis-28380 89 9 cnn cnn PROPN bracis-28380 89 10 model model NOUN bracis-28380 89 11 provided provide VERB bracis-28380 89 12 positive positive ADJ bracis-28380 89 13 feedback feedback NOUN bracis-28380 89 14 on on ADP bracis-28380 89 15 its its PRON bracis-28380 89 16 usefulness usefulness NOUN bracis-28380 89 17 in in ADP bracis-28380 89 18 providing provide VERB bracis-28380 89 19 an an DET bracis-28380 89 20 explainable explainable ADJ bracis-28380 89 21 and and CCONJ bracis-28380 89 22 safe safe ADJ bracis-28380 89 23 diagnosis diagnosis NOUN bracis-28380 89 24 decision decision NOUN bracis-28380 89 25 - - PUNCT bracis-28380 89 26 making make VERB bracis-28380 89 27 perspective perspective NOUN bracis-28380 89 28 for for ADP bracis-28380 89 29 brain brain NOUN bracis-28380 89 30 tumors tumor NOUN bracis-28380 89 31 . . PUNCT bracis-28380 90 1 this this DET bracis-28380 90 2 study study NOUN bracis-28380 90 3 demonstrates demonstrate VERB bracis-28380 90 4 that that SCONJ bracis-28380 90 5 the the DET bracis-28380 90 6 proposed propose VERB bracis-28380 90 7 method method NOUN bracis-28380 90 8 can can AUX bracis-28380 90 9 effectively effectively ADV bracis-28380 90 10 enhance enhance VERB bracis-28380 90 11 the the DET bracis-28380 90 12 interpretability interpretability NOUN bracis-28380 90 13 of of ADP bracis-28380 90 14 deep deep ADJ bracis-28380 90 15 learning learning NOUN bracis-28380 90 16 models model NOUN bracis-28380 90 17 for for ADP bracis-28380 90 18 brain brain NOUN bracis-28380 90 19 tumor tumor NOUN bracis-28380 90 20 diagnosis diagnosis NOUN bracis-28380 90 21 , , PUNCT bracis-28380 90 22 making make VERB bracis-28380 90 23 them they PRON bracis-28380 90 24 more more ADV bracis-28380 90 25 trustworthy trustworthy ADJ bracis-28380 90 26 and and CCONJ bracis-28380 90 27 accessible accessible ADJ bracis-28380 90 28 to to ADP bracis-28380 90 29 healthcare healthcare PROPN bracis-28380 90 30 professionals professional NOUN bracis-28380 90 31 . . PUNCT bracis-28380 91 1 the the DET bracis-28380 91 2 study study NOUN bracis-28380 91 3 [ [ X bracis-28380 91 4 15 15 NUM bracis-28380 91 5 ] ] PUNCT bracis-28380 91 6 demonstrated demonstrate VERB bracis-28380 91 7 the the DET bracis-28380 91 8 performance performance NOUN bracis-28380 91 9 of of ADP bracis-28380 91 10 the the DET bracis-28380 91 11 commonly commonly ADV bracis-28380 91 12 used use VERB bracis-28380 91 13 convolutional convolutional ADJ bracis-28380 91 14 neural neural ADJ bracis-28380 91 15 network network NOUN bracis-28380 91 16 ( ( PUNCT bracis-28380 91 17 cnn cnn PROPN bracis-28380 91 18 ) ) PUNCT bracis-28380 91 19 models model NOUN bracis-28380 91 20 : : PUNCT bracis-28380 91 21 vgg16 vgg16 NOUN bracis-28380 91 22 , , PUNCT bracis-28380 91 23 resnet50 resnet50 NOUN bracis-28380 91 24 , , PUNCT bracis-28380 91 25 and and CCONJ bracis-28380 91 26 mobilenet mobilenet NOUN bracis-28380 91 27 . . PUNCT bracis-28380 92 1 these these DET bracis-28380 92 2 models model NOUN bracis-28380 92 3 were be AUX bracis-28380 92 4 applied apply VERB bracis-28380 92 5 to to ADP bracis-28380 92 6 brain brain NOUN bracis-28380 92 7 magnetic magnetic ADJ bracis-28380 92 8 resonance resonance NOUN bracis-28380 92 9 imaging imaging NOUN bracis-28380 92 10 to to PART bracis-28380 92 11 identify identify VERB bracis-28380 92 12 tumor tumor NOUN bracis-28380 92 13 cells cell NOUN bracis-28380 92 14 . . PUNCT bracis-28380 93 1 the the DET bracis-28380 93 2 reported report VERB bracis-28380 93 3 accuracy accuracy NOUN bracis-28380 93 4 of of ADP bracis-28380 93 5 these these DET bracis-28380 93 6 models model NOUN bracis-28380 93 7 was be AUX bracis-28380 93 8 97 97 NUM bracis-28380 93 9 % % NOUN bracis-28380 93 10 , , PUNCT bracis-28380 93 11 94.5 94.5 NUM bracis-28380 93 12 % % NOUN bracis-28380 93 13 , , PUNCT bracis-28380 93 14 and and CCONJ bracis-28380 93 15 99 99 NUM bracis-28380 93 16 % % NOUN bracis-28380 93 17 , , PUNCT bracis-28380 93 18 respectively respectively ADV bracis-28380 93 19 . . PUNCT bracis-28380 94 1 the the DET bracis-28380 94 2 authors author NOUN bracis-28380 94 3 aimed aim VERB bracis-28380 94 4 to to PART bracis-28380 94 5 develop develop VERB bracis-28380 94 6 a a DET bracis-28380 94 7 modified modify VERB bracis-28380 94 8 cnn cnn PROPN bracis-28380 94 9 model model NOUN bracis-28380 94 10 comparing compare VERB bracis-28380 94 11 the the DET bracis-28380 94 12 proposed propose VERB bracis-28380 94 13 model model NOUN bracis-28380 94 14 with with ADP bracis-28380 94 15 the the DET bracis-28380 94 16 three three NUM bracis-28380 94 17 known know VERB bracis-28380 94 18 networks network NOUN bracis-28380 94 19 . . PUNCT bracis-28380 95 1 the the DET bracis-28380 95 2 model model NOUN bracis-28380 95 3 proposed propose VERB bracis-28380 95 4 in in ADP bracis-28380 95 5 the the DET bracis-28380 95 6 study study NOUN bracis-28380 95 7 achieved achieve VERB bracis-28380 95 8 an an DET bracis-28380 95 9 overall overall ADJ bracis-28380 95 10 classification classification NOUN bracis-28380 95 11 accuracy accuracy NOUN bracis-28380 95 12 of of ADP bracis-28380 95 13 98.5 98.5 NUM bracis-28380 95 14 % % NOUN bracis-28380 95 15 . . PUNCT bracis-28380 96 1 all all DET bracis-28380 96 2 models model NOUN bracis-28380 96 3 were be AUX bracis-28380 96 4 trained train VERB bracis-28380 96 5 and and CCONJ bracis-28380 96 6 tested test VERB bracis-28380 96 7 using use VERB bracis-28380 96 8 only only ADV bracis-28380 96 9 one one NUM bracis-28380 96 10 database database NOUN bracis-28380 96 11 , , PUNCT bracis-28380 96 12 the the DET bracis-28380 96 13 br35h br35h NOUN bracis-28380 96 14 . . PUNCT bracis-28380 97 1 the the DET bracis-28380 97 2 authors author NOUN bracis-28380 97 3 did do AUX bracis-28380 97 4 not not PART bracis-28380 97 5 explicitly explicitly ADV bracis-28380 97 6 mention mention VERB bracis-28380 97 7 the the DET bracis-28380 97 8 use use NOUN bracis-28380 97 9 of of ADP bracis-28380 97 10 explainable explainable ADJ bracis-28380 97 11 techniques technique NOUN bracis-28380 97 12 in in ADP bracis-28380 97 13 their their PRON bracis-28380 97 14 study study NOUN bracis-28380 97 15 . . PUNCT bracis-28380 98 1 however however ADV bracis-28380 98 2 , , PUNCT bracis-28380 98 3 they they PRON bracis-28380 98 4 expressed express VERB bracis-28380 98 5 their their PRON bracis-28380 98 6 intention intention NOUN bracis-28380 98 7 to to PART bracis-28380 98 8 increase increase VERB bracis-28380 98 9 the the DET bracis-28380 98 10 model model NOUN bracis-28380 98 11 ’s ’s PART bracis-28380 98 12 reliability reliability NOUN bracis-28380 98 13 by by ADP bracis-28380 98 14 implementing implement VERB bracis-28380 98 15 similar similar ADJ bracis-28380 98 16 medical medical ADJ bracis-28380 98 17 imaging imaging NOUN bracis-28380 98 18 datasets dataset NOUN bracis-28380 98 19 . . PUNCT bracis-28380 99 1 using use VERB bracis-28380 99 2 a a DET bracis-28380 99 3 single single ADJ bracis-28380 99 4 dataset dataset NOUN bracis-28380 99 5 in in ADP bracis-28380 99 6 a a DET bracis-28380 99 7 study study NOUN bracis-28380 99 8 , , PUNCT bracis-28380 99 9 as as SCONJ bracis-28380 99 10 mentioned mention VERB bracis-28380 99 11 in in ADP bracis-28380 99 12 some some DET bracis-28380 99 13 papers paper NOUN bracis-28380 99 14 presented present VERB bracis-28380 99 15 in in ADP bracis-28380 99 16 this this DET bracis-28380 99 17 section section NOUN bracis-28380 99 18 , , PUNCT bracis-28380 99 19 may may AUX bracis-28380 99 20 limit limit VERB bracis-28380 99 21 the the DET bracis-28380 99 22 generalizability generalizability NOUN bracis-28380 99 23 of of ADP bracis-28380 99 24 the the DET bracis-28380 99 25 results result NOUN bracis-28380 99 26 . . PUNCT bracis-28380 100 1 robust robust ADJ bracis-28380 100 2 conclusions conclusion NOUN bracis-28380 100 3 typically typically ADV bracis-28380 100 4 require require VERB bracis-28380 100 5 validation validation NOUN bracis-28380 100 6 across across ADP bracis-28380 100 7 multiple multiple ADJ bracis-28380 100 8 independent independent ADJ bracis-28380 100 9 datasets dataset NOUN bracis-28380 100 10 . . PUNCT bracis-28380 101 1 while while SCONJ bracis-28380 101 2 the the DET bracis-28380 101 3 studies study NOUN bracis-28380 101 4 presented present VERB bracis-28380 101 5 provide provide VERB bracis-28380 101 6 insights insight NOUN bracis-28380 101 7 into into ADP bracis-28380 101 8 the the DET bracis-28380 101 9 performance performance NOUN bracis-28380 101 10 of of ADP bracis-28380 101 11 specific specific ADJ bracis-28380 101 12 cnn cnn PROPN bracis-28380 101 13 models model NOUN bracis-28380 101 14 for for ADP bracis-28380 101 15 medical medical ADJ bracis-28380 101 16 diagnosis diagnosis NOUN bracis-28380 101 17 , , PUNCT bracis-28380 101 18 further further ADJ bracis-28380 101 19 research research NOUN bracis-28380 101 20 , , PUNCT bracis-28380 101 21 and and CCONJ bracis-28380 101 22 validation validation NOUN bracis-28380 101 23 on on ADP bracis-28380 101 24 diverse diverse ADJ bracis-28380 101 25 datasets dataset NOUN bracis-28380 101 26 are be AUX bracis-28380 101 27 essential essential ADJ bracis-28380 101 28 to to PART bracis-28380 101 29 establish establish VERB bracis-28380 101 30 the the DET bracis-28380 101 31 reliability reliability NOUN bracis-28380 101 32 and and CCONJ bracis-28380 101 33 generalizability generalizability NOUN bracis-28380 101 34 of of ADP bracis-28380 101 35 these these DET bracis-28380 101 36 models model NOUN bracis-28380 101 37 . . PUNCT bracis-28380 102 1 in in ADP bracis-28380 102 2 addition addition NOUN bracis-28380 102 3 to to ADP bracis-28380 102 4 evaluating evaluate VERB bracis-28380 102 5 the the DET bracis-28380 102 6 performance performance NOUN bracis-28380 102 7 of of ADP bracis-28380 102 8 the the DET bracis-28380 102 9 models model NOUN bracis-28380 102 10 with with ADP bracis-28380 102 11 statistical statistical ADJ bracis-28380 102 12 metrics metric NOUN bracis-28380 102 13 and and CCONJ bracis-28380 102 14 testing testing NOUN bracis-28380 102 15 in in ADP bracis-28380 102 16 more more ADJ bracis-28380 102 17 than than ADP bracis-28380 102 18 one one NUM bracis-28380 102 19 database database NOUN bracis-28380 102 20 , , PUNCT bracis-28380 102 21 an an DET bracis-28380 102 22 excellent excellent ADJ bracis-28380 102 23 current current ADJ bracis-28380 102 24 practice practice NOUN bracis-28380 102 25 is be AUX bracis-28380 102 26 to to PART bracis-28380 102 27 evaluate evaluate VERB bracis-28380 102 28 through through ADP bracis-28380 102 29 interpretability interpretability NOUN bracis-28380 102 30 techniques technique NOUN bracis-28380 102 31 so so SCONJ bracis-28380 102 32 that that SCONJ bracis-28380 102 33 the the DET bracis-28380 102 34 models model NOUN bracis-28380 102 35 are be AUX bracis-28380 102 36 not not PART bracis-28380 102 37 seen see VERB bracis-28380 102 38 as as ADP bracis-28380 102 39 black black ADJ bracis-28380 102 40 boxes box NOUN bracis-28380 102 41 . . PUNCT bracis-28380 103 1 these these DET bracis-28380 103 2 xai xai PROPN bracis-28380 103 3 techniques technique NOUN bracis-28380 103 4 offer offer VERB bracis-28380 103 5 different different ADJ bracis-28380 103 6 approaches approach NOUN bracis-28380 103 7 to to PART bracis-28380 103 8 explain explain VERB bracis-28380 103 9 the the DET bracis-28380 103 10 decision decision NOUN bracis-28380 103 11 - - PUNCT bracis-28380 103 12 making make VERB bracis-28380 103 13 processes process NOUN bracis-28380 103 14 of of ADP bracis-28380 103 15 deep deep ADJ bracis-28380 103 16 learning learning NOUN bracis-28380 103 17 models model NOUN bracis-28380 103 18 . . PUNCT bracis-28380 104 1 visualizing visualize VERB bracis-28380 104 2 important important ADJ bracis-28380 104 3 regions region NOUN bracis-28380 104 4 , , PUNCT bracis-28380 104 5 assigning assign VERB bracis-28380 104 6 feature feature NOUN bracis-28380 104 7 importance importance NOUN bracis-28380 104 8 values value NOUN bracis-28380 104 9 , , PUNCT bracis-28380 104 10 or or CCONJ bracis-28380 104 11 calculating calculate VERB bracis-28380 104 12 gradients gradient NOUN bracis-28380 104 13 provide provide VERB bracis-28380 104 14 insight insight NOUN bracis-28380 104 15 into into ADP bracis-28380 104 16 the the DET bracis-28380 104 17 model model NOUN bracis-28380 104 18 ’s ’s PART bracis-28380 104 19 inner inner ADJ bracis-28380 104 20 workings working NOUN bracis-28380 104 21 and and CCONJ bracis-28380 104 22 help help VERB bracis-28380 104 23 users user NOUN bracis-28380 104 24 understand understand VERB bracis-28380 104 25 why why SCONJ bracis-28380 104 26 specific specific ADJ bracis-28380 104 27 decisions decision NOUN bracis-28380 104 28 or or CCONJ bracis-28380 104 29 predictions prediction NOUN bracis-28380 104 30 were be AUX bracis-28380 104 31 made make VERB bracis-28380 104 32 . . PUNCT bracis-28380 105 1 these these DET bracis-28380 105 2 techniques technique NOUN bracis-28380 105 3 contribute contribute VERB bracis-28380 105 4 to to ADP bracis-28380 105 5 the the DET bracis-28380 105 6 reliability reliability NOUN bracis-28380 105 7 , , PUNCT bracis-28380 105 8 acceptance acceptance NOUN bracis-28380 105 9 , , PUNCT bracis-28380 105 10 and and CCONJ bracis-28380 105 11 responsible responsible ADJ bracis-28380 105 12 application application NOUN bracis-28380 105 13 of of ADP bracis-28380 105 14 deep deep ADJ bracis-28380 105 15 learning learning NOUN bracis-28380 105 16 models model NOUN bracis-28380 105 17 in in ADP bracis-28380 105 18 various various ADJ bracis-28380 105 19 domains domain NOUN bracis-28380 105 20 , , PUNCT bracis-28380 105 21 including include VERB bracis-28380 105 22 medical medical ADJ bracis-28380 105 23 diagnosis diagnosis NOUN bracis-28380 105 24 . . PUNCT bracis-28380 106 1 3 3 NUM bracis-28380 106 2 experiments experiment NOUN bracis-28380 106 3 for for ADP bracis-28380 106 4 the the DET bracis-28380 106 5 development development NOUN bracis-28380 106 6 and and CCONJ bracis-28380 106 7 execution execution NOUN bracis-28380 106 8 of of ADP bracis-28380 106 9 codes code NOUN bracis-28380 106 10 necessary necessary ADJ bracis-28380 106 11 for for ADP bracis-28380 106 12 manipulating manipulate VERB bracis-28380 106 13 data datum NOUN bracis-28380 106 14 and and CCONJ bracis-28380 106 15 implementing implement VERB bracis-28380 106 16 the the DET bracis-28380 106 17 model model NOUN bracis-28380 106 18 , , PUNCT bracis-28380 106 19 was be AUX bracis-28380 106 20 used use VERB bracis-28380 106 21 google google PROPN bracis-28380 106 22 collaboratory collaboratory NOUN bracis-28380 106 23 pro pro NOUN bracis-28380 106 24 . . PUNCT bracis-28380 107 1 the the DET bracis-28380 107 2 machine machine NOUN bracis-28380 107 3 learning learn VERB bracis-28380 107 4 library library NOUN bracis-28380 107 5 used use VERB bracis-28380 107 6 was be AUX bracis-28380 107 7 keras keras PROPN bracis-28380 107 8 via via ADP bracis-28380 107 9 the the DET bracis-28380 107 10 tensorflow tensorflow NOUN bracis-28380 107 11 platform platform NOUN bracis-28380 107 12 . . PUNCT bracis-28380 108 1 two two NUM bracis-28380 108 2 cnn cnn PROPN bracis-28380 108 3 architectures architecture NOUN bracis-28380 108 4 were be AUX bracis-28380 108 5 used use VERB bracis-28380 108 6 for for ADP bracis-28380 108 7 the the DET bracis-28380 108 8 experiments experiment NOUN bracis-28380 108 9 , , PUNCT bracis-28380 108 10 a a DET bracis-28380 108 11 standard standard ADJ bracis-28380 108 12 multi multi ADJ bracis-28380 108 13 - - ADJ bracis-28380 108 14 layered layered ADJ bracis-28380 108 15 deep deep ADJ bracis-28380 108 16 cnn cnn NOUN bracis-28380 108 17 ( ( PUNCT bracis-28380 108 18 fig fig NOUN bracis-28380 108 19 . . PUNCT bracis-28380 108 20     SPACE bracis-28380 108 21 1 1 NUM bracis-28380 108 22 ) ) PUNCT bracis-28380 108 23 , , PUNCT bracis-28380 108 24 conv2d conv2d VERB bracis-28380 108 25 developed develop VERB bracis-28380 108 26 with with ADP bracis-28380 108 27 three three NUM bracis-28380 108 28 convolutional convolutional ADJ bracis-28380 108 29 layers layer NOUN bracis-28380 108 30 , , PUNCT bracis-28380 108 31 and and CCONJ bracis-28380 108 32 a a DET bracis-28380 108 33 deeper deep ADJ bracis-28380 108 34 one one NOUN bracis-28380 108 35 , , PUNCT bracis-28380 108 36 the the DET bracis-28380 108 37 xception xception PROPN bracis-28380 108 38 model model NOUN bracis-28380 108 39 ( ( PUNCT bracis-28380 108 40 fig fig NOUN bracis-28380 108 41 . . PUNCT bracis-28380 108 42     SPACE bracis-28380 108 43 2 2 X bracis-28380 108 44 ) ) PUNCT bracis-28380 108 45 with with ADP bracis-28380 108 46 36 36 NUM bracis-28380 108 47 and and CCONJ bracis-28380 108 48 separable separable ADJ bracis-28380 108 49 convolutional convolutional ADJ bracis-28380 108 50 layers layer NOUN bracis-28380 108 51 . . PUNCT bracis-28380 109 1 fig fig NOUN bracis-28380 109 2 . . PUNCT bracis-28380 110 1 1 1 X bracis-28380 110 2 . . X bracis-28380 110 3 the the DET bracis-28380 110 4 basic basic ADJ bracis-28380 110 5 model model NOUN bracis-28380 110 6 layout layout NOUN bracis-28380 110 7 [ [ X bracis-28380 110 8 own own ADJ bracis-28380 110 9 author author NOUN bracis-28380 110 10 ] ] X bracis-28380 110 11 . . PUNCT bracis-28380 111 1 full full ADJ bracis-28380 111 2 size size NOUN bracis-28380 111 3 image image NOUN bracis-28380 111 4 fig fig NOUN bracis-28380 111 5 . . PUNCT bracis-28380 112 1 2 2 X bracis-28380 112 2 . . X bracis-28380 112 3 xception xception PROPN bracis-28380 112 4 model model PROPN bracis-28380 112 5 layout layout PROPN bracis-28380 113 1 [ [ X bracis-28380 113 2 16 16 NUM bracis-28380 113 3 ] ] X bracis-28380 113 4 full full ADJ bracis-28380 113 5 size size NOUN bracis-28380 113 6 image image NOUN bracis-28380 113 7 datasets dataset NOUN bracis-28380 113 8 . . PUNCT bracis-28380 114 1 this this DET bracis-28380 114 2 study study NOUN bracis-28380 114 3 utilized utilize VERB bracis-28380 114 4 two two NUM bracis-28380 114 5 databases database NOUN bracis-28380 114 6 for for ADP bracis-28380 114 7 brain brain NOUN bracis-28380 114 8 tumor tumor NOUN bracis-28380 114 9 detection detection NOUN bracis-28380 114 10 and and CCONJ bracis-28380 114 11 classification classification NOUN bracis-28380 114 12 . . PUNCT bracis-28380 115 1 the the DET bracis-28380 115 2 first first ADJ bracis-28380 115 3 database database NOUN bracis-28380 115 4 used use VERB bracis-28380 115 5 was be AUX bracis-28380 115 6 the the DET bracis-28380 115 7 br35h br35h ADJ bracis-28380 115 8 [ [ X bracis-28380 115 9 8 8 NUM bracis-28380 115 10 ] ] PUNCT bracis-28380 115 11 brain brain NOUN bracis-28380 115 12 tumor tumor NOUN bracis-28380 115 13 detection detection NOUN bracis-28380 115 14 database database NOUN bracis-28380 115 15 , , PUNCT bracis-28380 115 16 which which PRON bracis-28380 115 17 consists consist VERB bracis-28380 115 18 of of ADP bracis-28380 115 19 3000 3000 NUM bracis-28380 115 20 magnetic magnetic ADJ bracis-28380 115 21 resonance resonance NOUN bracis-28380 115 22 images image NOUN bracis-28380 115 23 ( ( PUNCT bracis-28380 115 24 mri mri NOUN bracis-28380 115 25 ) ) PUNCT bracis-28380 115 26 representing represent VERB bracis-28380 115 27 both both DET bracis-28380 115 28 images image NOUN bracis-28380 115 29 with with ADP bracis-28380 115 30 brain brain NOUN bracis-28380 115 31 tumors tumor NOUN bracis-28380 115 32 and and CCONJ bracis-28380 115 33 normal normal ADJ bracis-28380 115 34 images image NOUN bracis-28380 115 35 . . PUNCT bracis-28380 116 1 the the DET bracis-28380 116 2 database database NOUN bracis-28380 116 3 is be AUX bracis-28380 116 4 well well ADV bracis-28380 116 5 - - PUNCT bracis-28380 116 6 balanced balanced ADJ bracis-28380 116 7 , , PUNCT bracis-28380 116 8 containing contain VERB bracis-28380 116 9 1500 1500 NUM bracis-28380 116 10 images image NOUN bracis-28380 116 11 depicting depict VERB bracis-28380 116 12 brain brain NOUN bracis-28380 116 13 tumors tumor NOUN bracis-28380 116 14 and and CCONJ bracis-28380 116 15 an an DET bracis-28380 116 16 equal equal ADJ bracis-28380 116 17 number number NOUN bracis-28380 116 18 of of ADP bracis-28380 116 19 1500 1500 NUM bracis-28380 116 20 normal normal ADJ bracis-28380 116 21 images image NOUN bracis-28380 116 22 . . PUNCT bracis-28380 117 1 the the DET bracis-28380 117 2 available available ADJ bracis-28380 117 3 interpretation interpretation NOUN bracis-28380 117 4 categories category NOUN bracis-28380 117 5 for for ADP bracis-28380 117 6 this this DET bracis-28380 117 7 database database NOUN bracis-28380 117 8 are be AUX bracis-28380 117 9 “ " PUNCT bracis-28380 117 10 no no INTJ bracis-28380 117 11 ” " PUNCT bracis-28380 117 12 ( ( PUNCT bracis-28380 117 13 without without ADP bracis-28380 117 14 a a DET bracis-28380 117 15 brain brain NOUN bracis-28380 117 16 tumor tumor NOUN bracis-28380 117 17 ) ) PUNCT bracis-28380 117 18 and and CCONJ bracis-28380 117 19 “ " PUNCT bracis-28380 117 20 yes yes INTJ bracis-28380 117 21 ” " PUNCT bracis-28380 117 22 ( ( PUNCT bracis-28380 117 23 with with ADP bracis-28380 117 24 the the DET bracis-28380 117 25 presence presence NOUN bracis-28380 117 26 of of ADP bracis-28380 117 27 a a DET bracis-28380 117 28 brain brain NOUN bracis-28380 117 29 tumor tumor NOUN bracis-28380 117 30 ) ) PUNCT bracis-28380 117 31 . . PUNCT bracis-28380 118 1 the the DET bracis-28380 118 2 brain brain NOUN bracis-28380 118 3 tumor tumor NOUN bracis-28380 118 4 classification classification NOUN bracis-28380 118 5 ( ( PUNCT bracis-28380 118 6 mri mri NOUN bracis-28380 118 7 ) ) PUNCT bracis-28380 118 8 ( ( PUNCT bracis-28380 118 9 sartaj sartaj NOUN bracis-28380 118 10 ) ) PUNCT bracis-28380 119 1 [ [ X bracis-28380 119 2 9 9 NUM bracis-28380 119 3 ] ] PUNCT bracis-28380 119 4 database database NOUN bracis-28380 119 5 from from ADP bracis-28380 119 6 the the DET bracis-28380 119 7 kaggle kaggle ADJ bracis-28380 119 8 platform platform NOUN bracis-28380 119 9 was be AUX bracis-28380 119 10 employed employ VERB bracis-28380 119 11 to to PART bracis-28380 119 12 assess assess VERB bracis-28380 119 13 the the DET bracis-28380 119 14 generalization generalization NOUN bracis-28380 119 15 of of ADP bracis-28380 119 16 the the DET bracis-28380 119 17 models model NOUN bracis-28380 119 18 . . PUNCT bracis-28380 120 1 this this DET bracis-28380 120 2 database database NOUN bracis-28380 120 3 consists consist VERB bracis-28380 120 4 of of ADP bracis-28380 120 5 2870 2870 NUM bracis-28380 120 6 images image NOUN bracis-28380 120 7 for for ADP bracis-28380 120 8 training training NOUN bracis-28380 120 9 and and CCONJ bracis-28380 120 10 394 394 NUM bracis-28380 120 11 for for ADP bracis-28380 120 12 testing testing NOUN bracis-28380 120 13 , , PUNCT bracis-28380 120 14 divided divide VERB bracis-28380 120 15 into into ADP bracis-28380 120 16 four four NUM bracis-28380 120 17 classes class NOUN bracis-28380 120 18 : : PUNCT bracis-28380 120 19 without without ADP bracis-28380 120 20 tumor tumor NOUN bracis-28380 120 21 , , PUNCT bracis-28380 120 22 meningioma meningioma NOUN bracis-28380 120 23 , , PUNCT bracis-28380 120 24 pituitary pituitary ADJ bracis-28380 120 25 tumor tumor NOUN bracis-28380 120 26 , , PUNCT bracis-28380 120 27 and and CCONJ bracis-28380 120 28 glioma glioma NOUN bracis-28380 120 29 . . PUNCT bracis-28380 121 1 among among ADP bracis-28380 121 2 the the DET bracis-28380 121 3 test test NOUN bracis-28380 121 4 folder folder NOUN bracis-28380 121 5 images image NOUN bracis-28380 121 6 , , PUNCT bracis-28380 121 7 there there PRON bracis-28380 121 8 are be VERB bracis-28380 121 9 105 105 NUM bracis-28380 121 10 images image NOUN bracis-28380 121 11 without without ADP bracis-28380 121 12 tumors tumor NOUN bracis-28380 121 13 , , PUNCT bracis-28380 121 14 115 115 NUM bracis-28380 121 15 with with ADP bracis-28380 121 16 meningioma meningioma NOUN bracis-28380 121 17 , , PUNCT bracis-28380 121 18 100 100 NUM bracis-28380 121 19 with with ADP bracis-28380 121 20 glioma glioma NOUN bracis-28380 121 21 , , PUNCT bracis-28380 121 22 and and CCONJ bracis-28380 121 23 74 74 NUM bracis-28380 121 24 with with ADP bracis-28380 121 25 pituitary pituitary ADJ bracis-28380 121 26 tumors tumor NOUN bracis-28380 121 27 . . PUNCT bracis-28380 122 1 since since SCONJ bracis-28380 122 2 the the DET bracis-28380 122 3 research research NOUN bracis-28380 122 4 paper paper NOUN bracis-28380 122 5 focuses focus VERB bracis-28380 122 6 on on ADP bracis-28380 122 7 binary binary ADJ bracis-28380 122 8 classification classification NOUN bracis-28380 122 9 ( ( PUNCT bracis-28380 122 10 with with ADP bracis-28380 122 11 and and CCONJ bracis-28380 122 12 without without ADP bracis-28380 122 13 tumor tumor NOUN bracis-28380 122 14 ) ) PUNCT bracis-28380 122 15 , , PUNCT bracis-28380 122 16 without without ADP bracis-28380 122 17 classifying classify VERB bracis-28380 122 18 the the DET bracis-28380 122 19 tumor tumor NOUN bracis-28380 122 20 types type NOUN bracis-28380 122 21 , , PUNCT bracis-28380 122 22 the the DET bracis-28380 122 23 authors author NOUN bracis-28380 122 24 decided decide VERB bracis-28380 122 25 to to PART bracis-28380 122 26 conduct conduct VERB bracis-28380 122 27 an an DET bracis-28380 122 28 experiment experiment NOUN bracis-28380 122 29 using use VERB bracis-28380 122 30 the the DET bracis-28380 122 31 meningioma meningioma NOUN bracis-28380 122 32 class class NOUN bracis-28380 122 33 and and CCONJ bracis-28380 122 34 the the DET bracis-28380 122 35 no no DET bracis-28380 122 36 - - PUNCT bracis-28380 122 37 tumor tumor NOUN bracis-28380 122 38 class class NOUN bracis-28380 122 39 due due ADP bracis-28380 122 40 to to ADP bracis-28380 122 41 their their PRON bracis-28380 122 42 balanced balanced ADJ bracis-28380 122 43 distribution distribution NOUN bracis-28380 122 44 . . PUNCT bracis-28380 123 1 as as ADP bracis-28380 123 2 a a DET bracis-28380 123 3 result result NOUN bracis-28380 123 4 , , PUNCT bracis-28380 123 5 only only ADV bracis-28380 123 6 220 220 NUM bracis-28380 123 7 images image NOUN bracis-28380 123 8 were be AUX bracis-28380 123 9 selected select VERB bracis-28380 123 10 for for ADP bracis-28380 123 11 this this DET bracis-28380 123 12 experiment experiment NOUN bracis-28380 123 13 : : PUNCT bracis-28380 123 14 105 105 NUM bracis-28380 123 15 files file NOUN bracis-28380 123 16 representing represent VERB bracis-28380 123 17 “ " PUNCT bracis-28380 123 18 no no DET bracis-28380 123 19 tumor tumor NOUN bracis-28380 123 20 ” " PUNCT bracis-28380 123 21 and and CCONJ bracis-28380 123 22 115 115 NUM bracis-28380 123 23 files file NOUN bracis-28380 123 24 representing represent VERB bracis-28380 123 25 “ " PUNCT bracis-28380 123 26 meningioma meningioma ADJ bracis-28380 123 27 tumor tumor NOUN bracis-28380 123 28 ” " PUNCT bracis-28380 123 29 . . PUNCT bracis-28380 124 1 data datum NOUN bracis-28380 124 2 preprocessing preprocessing NOUN bracis-28380 124 3 and and CCONJ bracis-28380 124 4 data datum NOUN bracis-28380 124 5 augmentation augmentation NOUN bracis-28380 124 6 . . PUNCT bracis-28380 125 1 the the DET bracis-28380 125 2 images image NOUN bracis-28380 125 3 were be AUX bracis-28380 125 4 resized resize VERB bracis-28380 125 5 to to ADP bracis-28380 125 6 ( ( PUNCT bracis-28380 125 7 224 224 NUM bracis-28380 125 8 , , PUNCT bracis-28380 125 9 224 224 NUM bracis-28380 125 10 ) ) PUNCT bracis-28380 125 11 for for ADP bracis-28380 125 12 the the DET bracis-28380 125 13 first first ADJ bracis-28380 125 14 model model NOUN bracis-28380 125 15 and and CCONJ bracis-28380 125 16 ( ( PUNCT bracis-28380 125 17 299,299 299,299 NUM bracis-28380 125 18 ) ) PUNCT bracis-28380 125 19 for for ADP bracis-28380 125 20 the the DET bracis-28380 125 21 xception xception NOUN bracis-28380 125 22 pre pre ADJ bracis-28380 125 23 - - ADJ bracis-28380 125 24 trained train VERB bracis-28380 125 25 model model NOUN bracis-28380 125 26 . . PUNCT bracis-28380 126 1 data datum NOUN bracis-28380 126 2 augmentation augmentation NOUN bracis-28380 126 3 generates generate VERB bracis-28380 126 4 more more ADJ bracis-28380 126 5 training training NOUN bracis-28380 126 6 samples sample NOUN bracis-28380 126 7 by by ADP bracis-28380 126 8 augmenting augment VERB bracis-28380 126 9 the the DET bracis-28380 126 10 samples sample NOUN bracis-28380 126 11 through through ADP bracis-28380 126 12 some some DET bracis-28380 126 13 random random ADJ bracis-28380 126 14 transformations transformation NOUN bracis-28380 126 15 that that PRON bracis-28380 126 16 produce produce VERB bracis-28380 126 17 believable believable ADJ bracis-28380 126 18 - - PUNCT bracis-28380 126 19 looking look VERB bracis-28380 126 20 images image NOUN bracis-28380 126 21 . . PUNCT bracis-28380 127 1 the the DET bracis-28380 127 2 goal goal NOUN bracis-28380 127 3 is be AUX bracis-28380 127 4 for for SCONJ bracis-28380 127 5 the the DET bracis-28380 127 6 model model NOUN bracis-28380 127 7 to to PART bracis-28380 127 8 see see VERB bracis-28380 127 9 the the DET bracis-28380 127 10 same same ADJ bracis-28380 127 11 image image NOUN bracis-28380 127 12 only only ADV bracis-28380 127 13 once once ADV bracis-28380 127 14 during during ADP bracis-28380 127 15 training training NOUN bracis-28380 127 16 . . PUNCT bracis-28380 128 1 the the DET bracis-28380 128 2 following follow VERB bracis-28380 128 3 transformations transformation NOUN bracis-28380 128 4 were be AUX bracis-28380 128 5 applied apply VERB bracis-28380 128 6 in in ADP bracis-28380 128 7 the the DET bracis-28380 128 8 training training NOUN bracis-28380 128 9 of of ADP bracis-28380 128 10 both both DET bracis-28380 128 11 models model NOUN bracis-28380 128 12 : : PUNCT bracis-28380 128 13 randomflip(“horizontal randomflip(“horizontal ADJ bracis-28380 128 14 ” " PUNCT bracis-28380 128 15 ): ): PUNCT bracis-28380 128 16 this this DET bracis-28380 128 17 layer layer NOUN bracis-28380 128 18 performs perform VERB bracis-28380 128 19 random random ADJ bracis-28380 128 20 horizontal horizontal ADJ bracis-28380 128 21 flips flip NOUN bracis-28380 128 22 on on ADP bracis-28380 128 23 the the DET bracis-28380 128 24 input input NOUN bracis-28380 128 25 images image NOUN bracis-28380 128 26 . . PUNCT bracis-28380 129 1 it it PRON bracis-28380 129 2 randomly randomly ADV bracis-28380 129 3 mirrors mirror NOUN bracis-28380 129 4 images image NOUN bracis-28380 129 5 horizontally horizontally ADV bracis-28380 129 6 , , PUNCT bracis-28380 129 7 which which PRON bracis-28380 129 8 helps help VERB bracis-28380 129 9 the the DET bracis-28380 129 10 model model NOUN bracis-28380 129 11 learn learn VERB bracis-28380 129 12 invariant invariant ADJ bracis-28380 129 13 features feature NOUN bracis-28380 129 14 and and CCONJ bracis-28380 129 15 improve improve VERB bracis-28380 129 16 its its PRON bracis-28380 129 17 generalizability generalizability NOUN bracis-28380 129 18 . . PUNCT bracis-28380 130 1 randomrotation(0.1 randomrotation(0.1 PROPN bracis-28380 130 2 ): ): PUNCT bracis-28380 130 3 this this DET bracis-28380 130 4 layer layer NOUN bracis-28380 130 5 applies apply VERB bracis-28380 130 6 a a DET bracis-28380 130 7 random random ADJ bracis-28380 130 8 rotation rotation NOUN bracis-28380 130 9 to to ADP bracis-28380 130 10 the the DET bracis-28380 130 11 input input NOUN bracis-28380 130 12 images image NOUN bracis-28380 130 13 . . PUNCT bracis-28380 131 1 parameter parameter NOUN bracis-28380 131 2 0.1 0.1 NUM bracis-28380 131 3 indicates indicate VERB bracis-28380 131 4 the the DET bracis-28380 131 5 maximum maximum ADJ bracis-28380 131 6 angle angle NOUN bracis-28380 131 7 of of ADP bracis-28380 131 8 rotation rotation NOUN bracis-28380 131 9 in in ADP bracis-28380 131 10 radians radian NOUN bracis-28380 131 11 . . PUNCT bracis-28380 132 1 in in ADP bracis-28380 132 2 this this DET bracis-28380 132 3 case case NOUN bracis-28380 132 4 , , PUNCT bracis-28380 132 5 the the DET bracis-28380 132 6 images image NOUN bracis-28380 132 7 will will AUX bracis-28380 132 8 be be AUX bracis-28380 132 9 rotated rotate VERB bracis-28380 132 10 by by ADP bracis-28380 132 11 a a DET bracis-28380 132 12 random random ADJ bracis-28380 132 13 angle angle NOUN bracis-28380 132 14 within within ADP bracis-28380 132 15 the the DET bracis-28380 132 16 -0.1 -0.1 NOUN bracis-28380 132 17 to to PART bracis-28380 132 18 0.1 0.1 NUM bracis-28380 132 19 radians radian NOUN bracis-28380 132 20 range range NOUN bracis-28380 132 21 . . PUNCT bracis-28380 133 1 randomzoom(0.2 randomzoom(0.2 NOUN bracis-28380 133 2 ): ): PUNCT bracis-28380 133 3 this this DET bracis-28380 133 4 layer layer NOUN bracis-28380 133 5 randomly randomly ADV bracis-28380 133 6 zooms zoom VERB bracis-28380 133 7 in in ADP bracis-28380 133 8 on on ADP bracis-28380 133 9 the the DET bracis-28380 133 10 input input NOUN bracis-28380 133 11 images image NOUN bracis-28380 133 12 . . PUNCT bracis-28380 134 1 parameter parameter NOUN bracis-28380 134 2 0.2 0.2 NUM bracis-28380 134 3 specifies specifie NOUN bracis-28380 134 4 the the DET bracis-28380 134 5 maximum maximum ADJ bracis-28380 134 6 zoom zoom NOUN bracis-28380 134 7 range range NOUN bracis-28380 134 8 . . PUNCT bracis-28380 135 1 this this PRON bracis-28380 135 2 means mean VERB bracis-28380 135 3 that that SCONJ bracis-28380 135 4 images image NOUN bracis-28380 135 5 can can AUX bracis-28380 135 6 be be AUX bracis-28380 135 7 enlarged enlarge VERB bracis-28380 135 8 or or CCONJ bracis-28380 135 9 reduced reduce VERB bracis-28380 135 10 by by ADP bracis-28380 135 11 a a DET bracis-28380 135 12 random random ADJ bracis-28380 135 13 factor factor NOUN bracis-28380 135 14 within within ADP bracis-28380 135 15 0.8 0.8 NUM bracis-28380 135 16 to to PART bracis-28380 135 17 1.2 1.2 NUM bracis-28380 135 18 . . PUNCT bracis-28380 136 1 architectures architecture NOUN bracis-28380 136 2 . . PUNCT bracis-28380 137 1 below below ADP bracis-28380 137 2 we we PRON bracis-28380 137 3 present present VERB bracis-28380 137 4 the the DET bracis-28380 137 5 details detail NOUN bracis-28380 137 6 of of ADP bracis-28380 137 7 the the DET bracis-28380 137 8 two two NUM bracis-28380 137 9 architectures architecture NOUN bracis-28380 137 10 used use VERB bracis-28380 137 11 in in ADP bracis-28380 137 12 the the DET bracis-28380 137 13 experiments experiment NOUN bracis-28380 137 14 . . PUNCT bracis-28380 138 1 basic basic ADJ bracis-28380 138 2 model model NOUN bracis-28380 138 3 . . PUNCT bracis-28380 139 1 the the DET bracis-28380 139 2 basic basic ADJ bracis-28380 139 3 model model NOUN bracis-28380 139 4 was be AUX bracis-28380 139 5 developed develop VERB bracis-28380 139 6 with with ADP bracis-28380 139 7 a a DET bracis-28380 139 8 lean lean ADJ bracis-28380 139 9 architecture architecture NOUN bracis-28380 139 10 , , PUNCT bracis-28380 139 11 with with ADP bracis-28380 139 12 three three NUM bracis-28380 139 13 convolutional convolutional ADJ bracis-28380 139 14 layers layer NOUN bracis-28380 139 15 , , PUNCT bracis-28380 139 16 three three NUM bracis-28380 139 17 max max PROPN bracis-28380 139 18 pooling pool VERB bracis-28380 139 19 layers layer NOUN bracis-28380 139 20 , , PUNCT bracis-28380 139 21 a a DET bracis-28380 139 22 dense dense ADJ bracis-28380 139 23 layer layer NOUN bracis-28380 139 24 with with ADP bracis-28380 139 25 128 128 NUM bracis-28380 139 26 neurons neuron NOUN bracis-28380 139 27 , , PUNCT bracis-28380 139 28 and and CCONJ bracis-28380 139 29 the the DET bracis-28380 139 30 final final ADJ bracis-28380 139 31 layer layer NOUN bracis-28380 139 32 with with ADP bracis-28380 139 33 one one NUM bracis-28380 139 34 neuron neuron NOUN bracis-28380 139 35 for for ADP bracis-28380 139 36 the the DET bracis-28380 139 37 binary binary ADJ bracis-28380 139 38 classification classification NOUN bracis-28380 139 39 problem problem NOUN bracis-28380 139 40 . . PUNCT bracis-28380 140 1 a a DET bracis-28380 140 2 dropout dropout NOUN bracis-28380 140 3 of of ADP bracis-28380 140 4 40 40 NUM bracis-28380 140 5 % % NOUN bracis-28380 140 6 was be AUX bracis-28380 140 7 added add VERB bracis-28380 140 8 after after ADP bracis-28380 140 9 the the DET bracis-28380 140 10 flatten flatten NOUN bracis-28380 140 11 and and CCONJ bracis-28380 140 12 dense(128 dense(128 PROPN bracis-28380 140 13 ) ) PUNCT bracis-28380 140 14 layers layer NOUN bracis-28380 140 15 to to PART bracis-28380 140 16 avoid avoid VERB bracis-28380 140 17 overfitting overfitte VERB bracis-28380 140 18 . . PUNCT bracis-28380 141 1 xception xception PROPN bracis-28380 141 2 . . PUNCT bracis-28380 142 1 inspired inspire VERB bracis-28380 142 2 by by ADP bracis-28380 142 3 the the DET bracis-28380 142 4 assumptions assumption NOUN bracis-28380 142 5 of of ADP bracis-28380 142 6 inception inception ADJ bracis-28380 142 7 architecture architecture NOUN bracis-28380 142 8 , , PUNCT bracis-28380 142 9 the the DET bracis-28380 142 10 xception xception PROPN bracis-28380 142 11 model model NOUN bracis-28380 142 12 , , PUNCT bracis-28380 142 13 which which PRON bracis-28380 142 14 means mean VERB bracis-28380 142 15 “ " PUNCT bracis-28380 142 16 extreme extreme ADJ bracis-28380 142 17 inception inception NOUN bracis-28380 142 18 , , PUNCT bracis-28380 142 19 ” " PUNCT bracis-28380 142 20 was be AUX bracis-28380 142 21 named name VERB bracis-28380 142 22 because because SCONJ bracis-28380 142 23 , , PUNCT bracis-28380 142 24 according accord VERB bracis-28380 142 25 to to ADP bracis-28380 142 26 the the DET bracis-28380 142 27 author author NOUN bracis-28380 142 28 [ [ X bracis-28380 142 29 17 17 NUM bracis-28380 142 30 ] ] PUNCT bracis-28380 142 31 , , PUNCT bracis-28380 142 32 this this DET bracis-28380 142 33 hypothesis hypothesis NOUN bracis-28380 142 34 is be AUX bracis-28380 142 35 a a DET bracis-28380 142 36 stronger strong ADJ bracis-28380 142 37 version version NOUN bracis-28380 142 38 of of ADP bracis-28380 142 39 the the DET bracis-28380 142 40 hypothesis hypothesis NOUN bracis-28380 142 41 underlying underlie VERB bracis-28380 142 42 inception inception NOUN bracis-28380 142 43 architecture architecture NOUN bracis-28380 142 44 . . PUNCT bracis-28380 143 1 the the DET bracis-28380 143 2 author author NOUN bracis-28380 143 3 proposed propose VERB bracis-28380 143 4 a a DET bracis-28380 143 5 convolutional convolutional ADJ bracis-28380 143 6 neural neural ADJ bracis-28380 143 7 network network NOUN bracis-28380 143 8 architecture architecture NOUN bracis-28380 143 9 based base VERB bracis-28380 143 10 entirely entirely ADV bracis-28380 143 11 on on ADP bracis-28380 143 12 depthwise depthwise NOUN bracis-28380 143 13 separable separable ADJ bracis-28380 143 14 convolution convolution NOUN bracis-28380 143 15 layers layer NOUN bracis-28380 143 16 . . PUNCT bracis-28380 144 1 in in ADP bracis-28380 144 2 effect effect NOUN bracis-28380 144 3 , , PUNCT bracis-28380 144 4 they they PRON bracis-28380 144 5 make make VERB bracis-28380 144 6 the the DET bracis-28380 144 7 following follow VERB bracis-28380 144 8 hypothesis hypothesis NOUN bracis-28380 144 9 : : PUNCT bracis-28380 144 10 mapping map VERB bracis-28380 144 11 cross cross ADJ bracis-28380 144 12 - - ADJ bracis-28380 144 13 channel channel ADJ bracis-28380 144 14 and and CCONJ bracis-28380 144 15 spatial spatial ADJ bracis-28380 144 16 correlations correlation NOUN bracis-28380 144 17 in in ADP bracis-28380 144 18 the the DET bracis-28380 144 19 feature feature NOUN bracis-28380 144 20 maps map NOUN bracis-28380 144 21 of of ADP bracis-28380 144 22 convolutional convolutional ADJ bracis-28380 144 23 neural neural ADJ bracis-28380 144 24 networks network NOUN bracis-28380 144 25 can can AUX bracis-28380 144 26 be be AUX bracis-28380 144 27 entirely entirely ADV bracis-28380 144 28 decoupled decouple VERB bracis-28380 144 29 . . PUNCT bracis-28380 145 1 the the DET bracis-28380 145 2 xception xception PROPN bracis-28380 145 3 architecture architecture NOUN bracis-28380 145 4 has have VERB bracis-28380 145 5 36 36 NUM bracis-28380 145 6 convolutional convolutional ADJ bracis-28380 145 7 layers layer NOUN bracis-28380 145 8 forming form VERB bracis-28380 145 9 the the DET bracis-28380 145 10 feature feature NOUN bracis-28380 145 11 extraction extraction NOUN bracis-28380 145 12 base base NOUN bracis-28380 145 13 of of ADP bracis-28380 145 14 the the DET bracis-28380 145 15 network network NOUN bracis-28380 145 16 . . PUNCT bracis-28380 146 1 the the DET bracis-28380 146 2 convolutional convolutional ADJ bracis-28380 146 3 layers layer NOUN bracis-28380 146 4 are be AUX bracis-28380 146 5 structured structure VERB bracis-28380 146 6 into into ADP bracis-28380 146 7 14 14 NUM bracis-28380 146 8 modules module NOUN bracis-28380 146 9 with with ADP bracis-28380 146 10 linear linear ADJ bracis-28380 146 11 residual residual ADJ bracis-28380 146 12 connections connection NOUN bracis-28380 146 13 around around ADP bracis-28380 146 14 them they PRON bracis-28380 146 15 , , PUNCT bracis-28380 146 16 except except SCONJ bracis-28380 146 17 for for ADP bracis-28380 146 18 the the DET bracis-28380 146 19 first first ADJ bracis-28380 146 20 and and CCONJ bracis-28380 146 21 last last ADJ bracis-28380 146 22 modules module NOUN bracis-28380 146 23 [ [ X bracis-28380 146 24 17 17 NUM bracis-28380 146 25 ] ] PUNCT bracis-28380 146 26 . . PUNCT bracis-28380 147 1 metrics metric NOUN bracis-28380 147 2 . . PUNCT bracis-28380 148 1 the the DET bracis-28380 148 2 statistical statistical ADJ bracis-28380 148 3 metrics metric NOUN bracis-28380 148 4 used use VERB bracis-28380 148 5 to to PART bracis-28380 148 6 evaluate evaluate VERB bracis-28380 148 7 the the DET bracis-28380 148 8 models model NOUN bracis-28380 148 9 were be AUX bracis-28380 148 10 : : PUNCT bracis-28380 148 11 accuracy accuracy NOUN bracis-28380 148 12 , , PUNCT bracis-28380 148 13 precision precision NOUN bracis-28380 148 14 , , PUNCT bracis-28380 148 15 recall recall NOUN bracis-28380 148 16 , , PUNCT bracis-28380 148 17 specificity specificity NOUN bracis-28380 148 18 , , PUNCT bracis-28380 148 19 and and CCONJ bracis-28380 148 20 receiving receive VERB bracis-28380 148 21 operational operational ADJ bracis-28380 148 22 characteristics characteristic NOUN bracis-28380 148 23 ( ( PUNCT bracis-28380 148 24 roc roc PROPN bracis-28380 148 25 curve curve NOUN bracis-28380 148 26 ) ) PUNCT bracis-28380 148 27 . . PUNCT bracis-28380 149 1 these these DET bracis-28380 149 2 metrics metric NOUN bracis-28380 149 3 are be AUX bracis-28380 149 4 often often ADV bracis-28380 149 5 employed employ VERB bracis-28380 149 6 in in ADP bracis-28380 149 7 similar similar ADJ bracis-28380 149 8 works work NOUN bracis-28380 149 9 [ [ X bracis-28380 149 10 11 11 NUM bracis-28380 149 11 , , PUNCT bracis-28380 149 12 14 14 NUM bracis-28380 149 13 ] ] PUNCT bracis-28380 149 14 . . PUNCT bracis-28380 150 1 xai xai PROPN bracis-28380 150 2 techniques techniques PROPN bracis-28380 150 3 . . PUNCT bracis-28380 151 1 the the DET bracis-28380 151 2 feature feature NOUN bracis-28380 151 3 visualization visualization NOUN bracis-28380 151 4 technique technique NOUN bracis-28380 151 5 and and CCONJ bracis-28380 151 6 gradient gradient NOUN bracis-28380 151 7 - - PUNCT bracis-28380 151 8 weighted weight VERB bracis-28380 151 9 class class NOUN bracis-28380 151 10 activation activation NOUN bracis-28380 151 11 mapping mapping NOUN bracis-28380 151 12 ( ( PUNCT bracis-28380 151 13 grad grad NOUN bracis-28380 151 14 - - PUNCT bracis-28380 151 15 cam cam NOUN bracis-28380 151 16 ) ) PUNCT bracis-28380 151 17 were be AUX bracis-28380 151 18 used use VERB bracis-28380 151 19 in in ADP bracis-28380 151 20 the the DET bracis-28380 151 21 basic basic ADJ bracis-28380 151 22 cnn cnn PROPN bracis-28380 151 23 model model NOUN bracis-28380 151 24 and and CCONJ bracis-28380 151 25 the the DET bracis-28380 151 26 cam cam NOUN bracis-28380 151 27 with with ADP bracis-28380 151 28 lime lime NOUN bracis-28380 151 29 ( ( PUNCT bracis-28380 151 30 local local ADJ bracis-28380 151 31 interpretable interpretable ADJ bracis-28380 151 32 model model ADJ bracis-28380 151 33 - - ADJ bracis-28380 151 34 agnostic agnostic ADJ bracis-28380 151 35 explanation explanation NOUN bracis-28380 151 36 ) ) PUNCT bracis-28380 151 37 techniques technique NOUN bracis-28380 151 38 in in ADP bracis-28380 151 39 the the DET bracis-28380 151 40 xception xception PROPN bracis-28380 151 41 model model NOUN bracis-28380 151 42 . . PUNCT bracis-28380 152 1 the the DET bracis-28380 152 2 general general ADJ bracis-28380 152 3 idea idea NOUN bracis-28380 152 4 behind behind ADP bracis-28380 152 5 visualizing visualize VERB bracis-28380 152 6 cnn cnn PROPN bracis-28380 152 7 layers layer NOUN bracis-28380 152 8 is be AUX bracis-28380 152 9 to to PART bracis-28380 152 10 identify identify VERB bracis-28380 152 11 which which DET bracis-28380 152 12 parts part NOUN bracis-28380 152 13 of of ADP bracis-28380 152 14 the the DET bracis-28380 152 15 input input NOUN bracis-28380 152 16 image image NOUN bracis-28380 152 17 activate activate VERB bracis-28380 152 18 the the DET bracis-28380 152 19 different different ADJ bracis-28380 152 20 filters filter NOUN bracis-28380 152 21 at at ADP bracis-28380 152 22 each each DET bracis-28380 152 23 layer layer NOUN bracis-28380 152 24 and and CCONJ bracis-28380 152 25 how how SCONJ bracis-28380 152 26 these these DET bracis-28380 152 27 activations activation NOUN bracis-28380 152 28 are be AUX bracis-28380 152 29 combined combine VERB bracis-28380 152 30 to to PART bracis-28380 152 31 form form VERB bracis-28380 152 32 the the DET bracis-28380 152 33 final final ADJ bracis-28380 152 34 prediction prediction NOUN bracis-28380 152 35 [ [ X bracis-28380 152 36 18 18 NUM bracis-28380 152 37 ] ] PUNCT bracis-28380 152 38 . . PUNCT bracis-28380 153 1 this this DET bracis-28380 153 2 information information NOUN bracis-28380 153 3 can can AUX bracis-28380 153 4 be be AUX bracis-28380 153 5 used use VERB bracis-28380 153 6 to to PART bracis-28380 153 7 understand understand VERB bracis-28380 153 8 how how SCONJ bracis-28380 153 9 the the DET bracis-28380 153 10 network network NOUN bracis-28380 153 11 is be AUX bracis-28380 153 12 making make VERB bracis-28380 153 13 its its PRON bracis-28380 153 14 predictions prediction NOUN bracis-28380 153 15 and and CCONJ bracis-28380 153 16 identify identify VERB bracis-28380 153 17 potential potential ADJ bracis-28380 153 18 improvement improvement NOUN bracis-28380 153 19 areas area NOUN bracis-28380 153 20 . . PUNCT bracis-28380 154 1 the the DET bracis-28380 154 2 heatmap heatmap NOUN bracis-28380 154 3 technique technique NOUN bracis-28380 154 4 [ [ X bracis-28380 154 5 18 18 NUM bracis-28380 154 6 ] ] PUNCT bracis-28380 154 7 , , PUNCT bracis-28380 154 8 denominated denominate VERB bracis-28380 154 9 “ " PUNCT bracis-28380 154 10 class class NOUN bracis-28380 154 11 activation activation NOUN bracis-28380 154 12 maps map NOUN bracis-28380 154 13 ” " PUNCT bracis-28380 154 14 ( ( PUNCT bracis-28380 154 15 cams cam NOUN bracis-28380 154 16 ) ) PUNCT bracis-28380 154 17 , , PUNCT bracis-28380 154 18 involves involve VERB bracis-28380 154 19 taking take VERB bracis-28380 154 20 the the DET bracis-28380 154 21 output output NOUN bracis-28380 154 22 of of ADP bracis-28380 154 23 the the DET bracis-28380 154 24 last last ADJ bracis-28380 154 25 convolutional convolutional ADJ bracis-28380 154 26 layer layer NOUN bracis-28380 154 27 and and CCONJ bracis-28380 154 28 obtaining obtain VERB bracis-28380 154 29 a a DET bracis-28380 154 30 vector vector NOUN bracis-28380 154 31 feature feature NOUN bracis-28380 154 32 by by ADP bracis-28380 154 33 applying apply VERB bracis-28380 154 34 a a DET bracis-28380 154 35 global global ADJ bracis-28380 154 36 average average ADJ bracis-28380 154 37 pooling pool VERB bracis-28380 154 38 operation operation NOUN bracis-28380 154 39 . . PUNCT bracis-28380 155 1 the the DET bracis-28380 155 2 vector vector NOUN bracis-28380 155 3 feature feature NOUN bracis-28380 155 4 is be AUX bracis-28380 155 5 then then ADV bracis-28380 155 6 used use VERB bracis-28380 155 7 to to PART bracis-28380 155 8 compute compute VERB bracis-28380 155 9 the the DET bracis-28380 155 10 final final ADJ bracis-28380 155 11 prediction prediction NOUN bracis-28380 155 12 and and CCONJ bracis-28380 155 13 the the DET bracis-28380 155 14 class class NOUN bracis-28380 155 15 activation activation NOUN bracis-28380 155 16 map map NOUN bracis-28380 155 17 , , PUNCT bracis-28380 155 18 which which PRON bracis-28380 155 19 is be AUX bracis-28380 155 20 obtained obtain VERB bracis-28380 155 21 by by ADP bracis-28380 155 22 weighting weight VERB bracis-28380 155 23 the the DET bracis-28380 155 24 output output NOUN bracis-28380 155 25 of of ADP bracis-28380 155 26 the the DET bracis-28380 155 27 last last ADJ bracis-28380 155 28 convolutional convolutional ADJ bracis-28380 155 29 layer layer NOUN bracis-28380 155 30 with with ADP bracis-28380 155 31 the the DET bracis-28380 155 32 corresponding corresponding ADJ bracis-28380 155 33 weights weight NOUN bracis-28380 155 34 of of ADP bracis-28380 155 35 the the DET bracis-28380 155 36 final final ADJ bracis-28380 155 37 prediction prediction NOUN bracis-28380 155 38 ( ( PUNCT bracis-28380 155 39 eq eq NOUN bracis-28380 155 40 . . PUNCT bracis-28380 155 41     SPACE bracis-28380 155 42 1 1 NUM bracis-28380 155 43 ) ) PUNCT bracis-28380 155 44 . . PUNCT bracis-28380 156 1 $ $ SYM bracis-28380 156 2 $ $ SYM bracis-28380 156 3 \begin{aligned \begin{aligne VERB bracis-28380 156 4 } } PUNCT bracis-28380 156 5 m^{(c m^{(c NOUN bracis-28380 156 6 ) ) PUNCT bracis-28380 156 7 } } PUNCT bracis-28380 157 1 = = VERB bracis-28380 157 2 \sum \sum NOUN bracis-28380 157 3 _ _ PUNCT bracis-28380 157 4 k k PROPN bracis-28380 157 5 w_k^{(c w_k^{(c PROPN bracis-28380 157 6 ) ) PUNCT bracis-28380 157 7 } } PUNCT bracis-28380 158 1 \phi \phi PROPN bracis-28380 158 2 _ _ SYM bracis-28380 158 3 k(x k(x NOUN bracis-28380 158 4 ) ) PUNCT bracis-28380 158 5 \end{aligned}$$ \end{aligned}$$ X bracis-28380 158 6 ( ( PUNCT bracis-28380 158 7 1 1 NUM bracis-28380 158 8 ) ) PUNCT bracis-28380 158 9 where where SCONJ bracis-28380 158 10 \(w_{k}^{(c)}\ \(w_{k}^{(c)}\ NOUN bracis-28380 158 11 ) ) PUNCT bracis-28380 158 12 is be AUX bracis-28380 158 13 the the DET bracis-28380 158 14 weight weight NOUN bracis-28380 158 15 of of ADP bracis-28380 158 16 the the DET bracis-28380 158 17 k k NOUN bracis-28380 158 18 - - PUNCT bracis-28380 158 19 th th VERB bracis-28380 158 20 feature feature NOUN bracis-28380 158 21 map map NOUN bracis-28380 158 22 in in ADP bracis-28380 158 23 the the DET bracis-28380 158 24 last last ADJ bracis-28380 158 25 convolutional convolutional ADJ bracis-28380 158 26 layer layer NOUN bracis-28380 158 27 for for ADP bracis-28380 158 28 class class NOUN bracis-28380 158 29 c c NOUN bracis-28380 158 30 , , PUNCT bracis-28380 158 31 and and CCONJ bracis-28380 158 32 \(\phi \(\phi NOUN bracis-28380 158 33 _ _ PRON bracis-28380 158 34 k(x)\ k(x)\ PROPN bracis-28380 158 35 ) ) PUNCT bracis-28380 158 36 is be AUX bracis-28380 158 37 the the DET bracis-28380 158 38 activation activation NOUN bracis-28380 158 39 map map NOUN bracis-28380 158 40 of of ADP bracis-28380 158 41 the the DET bracis-28380 158 42 k k NOUN bracis-28380 158 43 - - PUNCT bracis-28380 158 44 th th VERB bracis-28380 158 45 feature feature NOUN bracis-28380 158 46 map map NOUN bracis-28380 158 47 in in ADP bracis-28380 158 48 the the DET bracis-28380 158 49 last last ADJ bracis-28380 158 50 convolutional convolutional ADJ bracis-28380 158 51 layer layer NOUN bracis-28380 158 52 for for ADP bracis-28380 158 53 the the DET bracis-28380 158 54 input input NOUN bracis-28380 158 55 image image NOUN bracis-28380 158 56 x. x. PROPN bracis-28380 158 57 grad grad PROPN bracis-28380 158 58 - - PUNCT bracis-28380 158 59 cam cam NOUN bracis-28380 159 1 [ [ X bracis-28380 159 2 19 19 NUM bracis-28380 159 3 ] ] PUNCT bracis-28380 159 4 is be AUX bracis-28380 159 5 an an DET bracis-28380 159 6 xai xai PROPN bracis-28380 159 7 technique technique NOUN bracis-28380 159 8 used use VERB bracis-28380 159 9 in in ADP bracis-28380 159 10 cnns cnn NOUN bracis-28380 159 11 to to PART bracis-28380 159 12 identify identify VERB bracis-28380 159 13 the the DET bracis-28380 159 14 most most ADV bracis-28380 159 15 important important ADJ bracis-28380 159 16 regions region NOUN bracis-28380 159 17 of of ADP bracis-28380 159 18 an an DET bracis-28380 159 19 image image NOUN bracis-28380 159 20 that that PRON bracis-28380 159 21 contributed contribute VERB bracis-28380 159 22 to to ADP bracis-28380 159 23 a a DET bracis-28380 159 24 particular particular ADJ bracis-28380 159 25 class class NOUN bracis-28380 159 26 prediction prediction NOUN bracis-28380 159 27 . . PUNCT bracis-28380 160 1 it it PRON bracis-28380 160 2 uses use VERB bracis-28380 160 3 the the DET bracis-28380 160 4 gradient gradient ADJ bracis-28380 160 5 information information NOUN bracis-28380 160 6 flowing flow VERB bracis-28380 160 7 into into ADP bracis-28380 160 8 the the DET bracis-28380 160 9 last last ADJ bracis-28380 160 10 convolutional convolutional ADJ bracis-28380 160 11 layer layer NOUN bracis-28380 160 12 of of ADP bracis-28380 160 13 the the DET bracis-28380 160 14 network network NOUN bracis-28380 160 15 to to PART bracis-28380 160 16 generate generate VERB bracis-28380 160 17 a a DET bracis-28380 160 18 heatmap heatmap NOUN bracis-28380 160 19 that that PRON bracis-28380 160 20 highlights highlight VERB bracis-28380 160 21 these these DET bracis-28380 160 22 regions region NOUN bracis-28380 160 23 . . PUNCT bracis-28380 161 1 grad grad NOUN bracis-28380 161 2 - - PUNCT bracis-28380 161 3 cam cam PROPN bracis-28380 161 4 is be AUX bracis-28380 161 5 an an DET bracis-28380 161 6 extension extension NOUN bracis-28380 161 7 of of ADP bracis-28380 161 8 cam cam NOUN bracis-28380 161 9 that that PRON bracis-28380 161 10 can can AUX bracis-28380 161 11 be be AUX bracis-28380 161 12 used use VERB bracis-28380 161 13 with with ADP bracis-28380 161 14 any any DET bracis-28380 161 15 cnn cnn NOUN bracis-28380 161 16 architecture architecture NOUN bracis-28380 161 17 and and CCONJ bracis-28380 161 18 produces produce VERB bracis-28380 161 19 more more ADV bracis-28380 161 20 accurate accurate ADJ bracis-28380 161 21 and and CCONJ bracis-28380 161 22 finer finer ADV bracis-28380 161 23 - - PUNCT bracis-28380 161 24 grained grain VERB bracis-28380 161 25 heatmaps heatmap NOUN bracis-28380 161 26 than than ADP bracis-28380 161 27 cam cam NOUN bracis-28380 161 28 . . PUNCT bracis-28380 162 1 these these DET bracis-28380 162 2 techniques technique NOUN bracis-28380 162 3 help help VERB bracis-28380 162 4 interpret interpret VERB bracis-28380 162 5 cnn cnn PROPN bracis-28380 162 6 predictions prediction NOUN bracis-28380 162 7 and and CCONJ bracis-28380 162 8 identify identify VERB bracis-28380 162 9 potential potential ADJ bracis-28380 162 10 biases bias NOUN bracis-28380 162 11 or or CCONJ bracis-28380 162 12 limitations limitation NOUN bracis-28380 162 13 in in ADP bracis-28380 162 14 the the DET bracis-28380 162 15 model model NOUN bracis-28380 162 16 . . PUNCT bracis-28380 163 1 3.1 3.1 NUM bracis-28380 163 2 training training NOUN bracis-28380 163 3 , , PUNCT bracis-28380 163 4 validation validation NOUN bracis-28380 163 5 , , PUNCT bracis-28380 163 6 and and CCONJ bracis-28380 163 7     SPACE bracis-28380 163 8 test test VERB bracis-28380 163 9 the the DET bracis-28380 163 10 br35h br35h NOUN bracis-28380 164 1 [ [ X bracis-28380 164 2 8 8 NUM bracis-28380 164 3 ] ] X bracis-28380 164 4 database database NOUN bracis-28380 164 5 was be AUX bracis-28380 164 6 used use VERB bracis-28380 164 7 in in ADP bracis-28380 164 8 both both DET bracis-28380 164 9 models model NOUN bracis-28380 164 10 for for ADP bracis-28380 164 11 training training NOUN bracis-28380 164 12 and and CCONJ bracis-28380 164 13 validation validation NOUN bracis-28380 164 14 using use VERB bracis-28380 164 15 the the DET bracis-28380 164 16 holdout holdout NOUN bracis-28380 164 17 technique technique NOUN bracis-28380 164 18 ( ( PUNCT bracis-28380 164 19 80 80 NUM bracis-28380 164 20 % % NOUN bracis-28380 164 21 for for ADP bracis-28380 164 22 training training NOUN bracis-28380 164 23 and and CCONJ bracis-28380 164 24 20 20 NUM bracis-28380 164 25 % % NOUN bracis-28380 164 26 for for ADP bracis-28380 164 27 validation validation NOUN bracis-28380 164 28 ) ) PUNCT bracis-28380 164 29 , , PUNCT bracis-28380 164 30 and and CCONJ bracis-28380 164 31 in in ADP bracis-28380 164 32 the the DET bracis-28380 164 33 first first ADJ bracis-28380 164 34 step step NOUN bracis-28380 164 35 , , PUNCT bracis-28380 164 36 10 10 NUM bracis-28380 164 37 % % NOUN bracis-28380 164 38 of of ADP bracis-28380 164 39 the the DET bracis-28380 164 40 entire entire ADJ bracis-28380 164 41 dataset dataset NOUN bracis-28380 164 42 , , PUNCT bracis-28380 164 43 300 300 NUM bracis-28380 164 44 samples sample NOUN bracis-28380 164 45 , , PUNCT bracis-28380 164 46 was be AUX bracis-28380 164 47 saved save VERB bracis-28380 164 48 for for ADP bracis-28380 164 49 testing testing NOUN bracis-28380 164 50 . . PUNCT bracis-28380 165 1 afterward afterward ADV bracis-28380 165 2 , , PUNCT bracis-28380 165 3 the the DET bracis-28380 165 4 models model NOUN bracis-28380 165 5 were be AUX bracis-28380 165 6 trained train VERB bracis-28380 165 7 for for ADP bracis-28380 165 8 60 60 NUM bracis-28380 165 9 epochs epoch NOUN bracis-28380 165 10 and and CCONJ bracis-28380 165 11 compiled compile VERB bracis-28380 165 12 using use VERB bracis-28380 165 13 the the DET bracis-28380 165 14 adam adam PROPN bracis-28380 165 15 learning learning NOUN bracis-28380 165 16 rate rate NOUN bracis-28380 165 17 optimizer optimizer NOUN bracis-28380 165 18 by by ADP bracis-28380 165 19 default default NOUN bracis-28380 165 20 0.001 0.001 NUM bracis-28380 165 21 . . PUNCT bracis-28380 166 1 the the DET bracis-28380 166 2 metrics metric NOUN bracis-28380 166 3 for for ADP bracis-28380 166 4 monitoring monitor VERB bracis-28380 166 5 the the DET bracis-28380 166 6 convergence convergence NOUN bracis-28380 166 7 of of ADP bracis-28380 166 8 the the DET bracis-28380 166 9 models model NOUN bracis-28380 166 10 were be AUX bracis-28380 166 11 accuracy accuracy NOUN bracis-28380 166 12 and and CCONJ bracis-28380 166 13 binary binary PROPN bracis-28380 166 14 cross cross NOUN bracis-28380 166 15 - - NOUN bracis-28380 166 16 entropy entropy NOUN bracis-28380 166 17 for for ADP bracis-28380 166 18 loss loss NOUN bracis-28380 166 19 . . PUNCT bracis-28380 167 1 in in ADP bracis-28380 167 2 the the DET bracis-28380 167 3 final final ADJ bracis-28380 167 4 phase phase NOUN bracis-28380 167 5 , , PUNCT bracis-28380 167 6 to to PART bracis-28380 167 7 test test VERB bracis-28380 167 8 the the DET bracis-28380 167 9 models model NOUN bracis-28380 167 10 and and CCONJ bracis-28380 167 11 generate generate VERB bracis-28380 167 12 the the DET bracis-28380 167 13 confusion confusion NOUN bracis-28380 167 14 matrix matrix NOUN bracis-28380 167 15 , , PUNCT bracis-28380 167 16 220 220 NUM bracis-28380 167 17 images image NOUN bracis-28380 167 18 from from ADP bracis-28380 167 19 the the DET bracis-28380 167 20 other other ADJ bracis-28380 167 21 database database NOUN bracis-28380 167 22 [ [ X bracis-28380 167 23 9 9 NUM bracis-28380 167 24 ] ] PUNCT bracis-28380 167 25 were be AUX bracis-28380 167 26 used use VERB bracis-28380 167 27 . . PUNCT bracis-28380 168 1 4 4 NUM bracis-28380 168 2 results result NOUN bracis-28380 168 3 and and CCONJ bracis-28380 168 4     SPACE bracis-28380 168 5 discussion discussion NOUN bracis-28380 168 6 this this DET bracis-28380 168 7 section section NOUN bracis-28380 168 8 will will AUX bracis-28380 168 9 present present VERB bracis-28380 168 10 and and CCONJ bracis-28380 168 11 discuss discuss VERB bracis-28380 168 12 the the DET bracis-28380 168 13 performance performance NOUN bracis-28380 168 14 of of ADP bracis-28380 168 15 the the DET bracis-28380 168 16 models model NOUN bracis-28380 168 17 in in ADP bracis-28380 168 18 the the DET bracis-28380 168 19 br35h br35h NOUN bracis-28380 168 20 and and CCONJ bracis-28380 168 21 sartaj sartaj VERB bracis-28380 168 22 datasets dataset NOUN bracis-28380 168 23 . . PUNCT bracis-28380 169 1 afterward afterward ADV bracis-28380 169 2 , , PUNCT bracis-28380 169 3 information information NOUN bracis-28380 169 4 regarding regard VERB bracis-28380 169 5 the the DET bracis-28380 169 6 xai xai PROPN bracis-28380 169 7 techniques technique NOUN bracis-28380 169 8 applied apply VERB bracis-28380 169 9 to to ADP bracis-28380 169 10 the the DET bracis-28380 169 11 models model NOUN bracis-28380 169 12 will will AUX bracis-28380 169 13 be be AUX bracis-28380 169 14 presented present VERB bracis-28380 169 15 . . PUNCT bracis-28380 170 1 the the DET bracis-28380 170 2 hyperparameters hyperparameter NOUN bracis-28380 170 3 used use VERB bracis-28380 170 4 were be AUX bracis-28380 170 5 selected select VERB bracis-28380 170 6 based base VERB bracis-28380 170 7 on on ADP bracis-28380 170 8 previous previous ADJ bracis-28380 170 9 research research NOUN bracis-28380 170 10 and and CCONJ bracis-28380 170 11 empirical empirical ADJ bracis-28380 170 12 evaluation evaluation NOUN bracis-28380 170 13 and and CCONJ bracis-28380 170 14 were be AUX bracis-28380 170 15 considered consider VERB bracis-28380 170 16 effective effective ADJ bracis-28380 170 17 for for ADP bracis-28380 170 18 the the DET bracis-28380 170 19 classification classification NOUN bracis-28380 170 20 task task NOUN bracis-28380 170 21 of of ADP bracis-28380 170 22 brain brain NOUN bracis-28380 170 23 tumor tumor NOUN bracis-28380 170 24 images image NOUN bracis-28380 170 25 . . PUNCT bracis-28380 171 1 the the DET bracis-28380 171 2 adam adam PROPN bracis-28380 171 3 optimizer optimizer NOUN bracis-28380 171 4 was be AUX bracis-28380 171 5 used use VERB bracis-28380 171 6 for for ADP bracis-28380 171 7 both both DET bracis-28380 171 8 models model NOUN bracis-28380 171 9 , , PUNCT bracis-28380 171 10 with with ADP bracis-28380 171 11 a a DET bracis-28380 171 12 learning learn VERB bracis-28380 171 13 rate rate NOUN bracis-28380 171 14 of of ADP bracis-28380 171 15 0.001 0.001 NUM bracis-28380 171 16 . . PUNCT bracis-28380 172 1 the the DET bracis-28380 172 2 adam adam PROPN bracis-28380 172 3 optimizer optimizer NOUN bracis-28380 172 4 is be AUX bracis-28380 172 5 a a DET bracis-28380 172 6 widely widely ADV bracis-28380 172 7 used use VERB bracis-28380 172 8 optimization optimization NOUN bracis-28380 172 9 algorithm algorithm NOUN bracis-28380 172 10 that that PRON bracis-28380 172 11 adapts adapt VERB bracis-28380 172 12 the the DET bracis-28380 172 13 learning learning NOUN bracis-28380 172 14 rate rate NOUN bracis-28380 172 15 for for ADP bracis-28380 172 16 each each DET bracis-28380 172 17 parameter parameter NOUN bracis-28380 172 18 during during ADP bracis-28380 172 19 training training NOUN bracis-28380 172 20 , , PUNCT bracis-28380 172 21 making make VERB bracis-28380 172 22 it it PRON bracis-28380 172 23 suitable suitable ADJ bracis-28380 172 24 for for ADP bracis-28380 172 25 various various ADJ bracis-28380 172 26 tasks task NOUN bracis-28380 172 27 . . PUNCT bracis-28380 173 1 the the DET bracis-28380 173 2 learning learning NOUN bracis-28380 173 3 rate rate NOUN bracis-28380 173 4 determines determine VERB bracis-28380 173 5 the the DET bracis-28380 173 6 step step NOUN bracis-28380 173 7 size size NOUN bracis-28380 173 8 in in ADP bracis-28380 173 9 each each DET bracis-28380 173 10 iteration iteration NOUN bracis-28380 173 11 during during ADP bracis-28380 173 12 optimization optimization NOUN bracis-28380 173 13 . . PUNCT bracis-28380 174 1 the the DET bracis-28380 174 2 batch batch NOUN bracis-28380 174 3 size size NOUN bracis-28380 174 4 was be AUX bracis-28380 174 5 32 32 NUM bracis-28380 174 6 for for ADP bracis-28380 174 7 the the DET bracis-28380 174 8 base base NOUN bracis-28380 174 9 model model NOUN bracis-28380 174 10 and and CCONJ bracis-28380 174 11 64 64 NUM bracis-28380 174 12 for for ADP bracis-28380 174 13 the the DET bracis-28380 174 14 xception xception PROPN bracis-28380 174 15 model model PROPN bracis-28380 174 16 . . PUNCT bracis-28380 175 1 batch batch NOUN bracis-28380 175 2 size size NOUN bracis-28380 175 3 affects affect VERB bracis-28380 175 4 the the DET bracis-28380 175 5 data datum NOUN bracis-28380 175 6 used use VERB bracis-28380 175 7 to to PART bracis-28380 175 8 update update VERB bracis-28380 175 9 model model NOUN bracis-28380 175 10 parameters parameter NOUN bracis-28380 175 11 on on ADP bracis-28380 175 12 each each DET bracis-28380 175 13 iteration iteration NOUN bracis-28380 175 14 . . PUNCT bracis-28380 176 1 larger large ADJ bracis-28380 176 2 batch batch NOUN bracis-28380 176 3 size size NOUN bracis-28380 176 4 can can AUX bracis-28380 176 5 lead lead VERB bracis-28380 176 6 to to ADP bracis-28380 176 7 more more ADV bracis-28380 176 8 stable stable ADJ bracis-28380 176 9 convergence convergence NOUN bracis-28380 176 10 and and CCONJ bracis-28380 176 11 increase increase VERB bracis-28380 176 12 memory memory NOUN bracis-28380 176 13 requirement requirement NOUN bracis-28380 176 14 . . PUNCT bracis-28380 177 1 finally finally ADV bracis-28380 177 2 , , PUNCT bracis-28380 177 3 both both DET bracis-28380 177 4 models model NOUN bracis-28380 177 5 were be AUX bracis-28380 177 6 trained train VERB bracis-28380 177 7 for for ADP bracis-28380 177 8 60 60 NUM bracis-28380 177 9 epochs epoch NOUN bracis-28380 177 10 . . PUNCT bracis-28380 178 1 the the DET bracis-28380 178 2 learning learn VERB bracis-28380 178 3 curves curve NOUN bracis-28380 178 4 shown show VERB bracis-28380 178 5 in in ADP bracis-28380 178 6 fig fig NOUN bracis-28380 178 7 . . PUNCT bracis-28380 178 8     SPACE bracis-28380 178 9 3 3 NUM bracis-28380 178 10 and and CCONJ bracis-28380 178 11 fig fig NOUN bracis-28380 178 12 . . PUNCT bracis-28380 178 13     SPACE bracis-28380 178 14 4 4 NUM bracis-28380 178 15 demonstrate demonstrate VERB bracis-28380 178 16 the the DET bracis-28380 178 17 training training NOUN bracis-28380 178 18 and and CCONJ bracis-28380 178 19 validation validation NOUN bracis-28380 178 20 accuracy accuracy NOUN bracis-28380 178 21 and and CCONJ bracis-28380 178 22 loss loss NOUN bracis-28380 178 23 over over ADP bracis-28380 178 24 the the DET bracis-28380 178 25 60 60 NUM bracis-28380 178 26 epochs epoch NOUN bracis-28380 178 27 for for ADP bracis-28380 178 28 the the DET bracis-28380 178 29 basic basic ADJ bracis-28380 178 30 model model NOUN bracis-28380 178 31 cnn cnn PROPN bracis-28380 178 32 and and CCONJ bracis-28380 178 33 xception xception PROPN bracis-28380 178 34 model model PROPN bracis-28380 178 35 cnn cnn PROPN bracis-28380 178 36 , , PUNCT bracis-28380 178 37 respectively respectively ADV bracis-28380 178 38 . . PUNCT bracis-28380 179 1 in in ADP bracis-28380 179 2 the the DET bracis-28380 179 3 case case NOUN bracis-28380 179 4 of of ADP bracis-28380 179 5 the the DET bracis-28380 179 6 basic basic ADJ bracis-28380 179 7 model model NOUN bracis-28380 179 8 cnn cnn PROPN bracis-28380 179 9 , , PUNCT bracis-28380 179 10 the the DET bracis-28380 179 11 training training NOUN bracis-28380 179 12 accuracy accuracy NOUN bracis-28380 179 13 gradually gradually ADV bracis-28380 179 14 increases increase VERB bracis-28380 179 15 and and CCONJ bracis-28380 179 16 plateaus plateaus NOUN bracis-28380 179 17 at at ADP bracis-28380 179 18 around around ADP bracis-28380 179 19 95 95 NUM bracis-28380 179 20 % % NOUN bracis-28380 179 21 after after ADP bracis-28380 179 22 about about ADV bracis-28380 179 23 20 20 NUM bracis-28380 179 24 epochs epoch NOUN bracis-28380 179 25 . . PUNCT bracis-28380 180 1 the the DET bracis-28380 180 2 validation validation NOUN bracis-28380 180 3 accuracy accuracy NOUN bracis-28380 180 4 also also ADV bracis-28380 180 5 increases increase VERB bracis-28380 180 6 but but CCONJ bracis-28380 180 7 plateaus plateaus NOUN bracis-28380 180 8 at at ADP bracis-28380 180 9 around around ADV bracis-28380 180 10 90 90 NUM bracis-28380 180 11 % % NOUN bracis-28380 180 12 after after ADP bracis-28380 180 13 about about ADV bracis-28380 180 14 15 15 NUM bracis-28380 180 15 epochs epoch NOUN bracis-28380 180 16 . . PUNCT bracis-28380 181 1 the the DET bracis-28380 181 2 loss loss NOUN bracis-28380 181 3 decreases decrease VERB bracis-28380 181 4 sharply sharply ADV bracis-28380 181 5 in in ADP bracis-28380 181 6 the the DET bracis-28380 181 7 first first ADJ bracis-28380 181 8 few few ADJ bracis-28380 181 9 epochs epoch NOUN bracis-28380 181 10 , , PUNCT bracis-28380 181 11 then then ADV bracis-28380 181 12 decreases decrease VERB bracis-28380 181 13 at at ADP bracis-28380 181 14 a a DET bracis-28380 181 15 slower slow ADJ bracis-28380 181 16 rate rate NOUN bracis-28380 181 17 until until SCONJ bracis-28380 181 18 it it PRON bracis-28380 181 19 plateaus plateaus ADV bracis-28380 181 20 after after ADP bracis-28380 181 21 about about ADV bracis-28380 181 22 30 30 NUM bracis-28380 181 23 epochs epoch NOUN bracis-28380 181 24 . . PUNCT bracis-28380 182 1 these these DET bracis-28380 182 2 results result NOUN bracis-28380 182 3 suggest suggest VERB bracis-28380 182 4 that that SCONJ bracis-28380 182 5 the the DET bracis-28380 182 6 basic basic ADJ bracis-28380 182 7 model model NOUN bracis-28380 182 8 cnn cnn PROPN bracis-28380 182 9 can can AUX bracis-28380 182 10 achieve achieve VERB bracis-28380 182 11 high high ADJ bracis-28380 182 12 accuracy accuracy NOUN bracis-28380 182 13 on on ADP bracis-28380 182 14 the the DET bracis-28380 182 15 training training NOUN bracis-28380 182 16 data datum NOUN bracis-28380 182 17 but but CCONJ bracis-28380 182 18 may may AUX bracis-28380 182 19 be be AUX bracis-28380 182 20 overfitting overfitte VERB bracis-28380 182 21 after after ADP bracis-28380 182 22 20 20 NUM bracis-28380 182 23 epochs epoch NOUN bracis-28380 182 24 , , PUNCT bracis-28380 182 25 as as SCONJ bracis-28380 182 26 the the DET bracis-28380 182 27 validation validation NOUN bracis-28380 182 28 accuracy accuracy NOUN bracis-28380 182 29 is be AUX bracis-28380 182 30 a a DET bracis-28380 182 31 bit bit NOUN bracis-28380 182 32 lower low ADJ bracis-28380 182 33 than than ADP bracis-28380 182 34 the the DET bracis-28380 182 35 training training NOUN bracis-28380 182 36 accuracy accuracy NOUN bracis-28380 182 37 . . PUNCT bracis-28380 183 1 for for ADP bracis-28380 183 2 the the DET bracis-28380 183 3 xception xception PROPN bracis-28380 183 4 model model PROPN bracis-28380 183 5 cnn cnn PROPN bracis-28380 183 6 , , PUNCT bracis-28380 183 7 the the DET bracis-28380 183 8 training training NOUN bracis-28380 183 9 accuracy accuracy NOUN bracis-28380 183 10 increases increase VERB bracis-28380 183 11 more more ADV bracis-28380 183 12 rapidly rapidly ADV bracis-28380 183 13 than than ADP bracis-28380 183 14 the the DET bracis-28380 183 15 basic basic ADJ bracis-28380 183 16 model model NOUN bracis-28380 183 17 cnn cnn PROPN bracis-28380 183 18 and and CCONJ bracis-28380 183 19 plateaus plateaus NOUN bracis-28380 183 20 at at ADP bracis-28380 183 21 around around ADP bracis-28380 183 22 98 98 NUM bracis-28380 183 23 % % NOUN bracis-28380 183 24 after after ADP bracis-28380 183 25 ten ten NUM bracis-28380 183 26 epochs epoch NOUN bracis-28380 183 27 . . PUNCT bracis-28380 184 1 the the DET bracis-28380 184 2 validation validation NOUN bracis-28380 184 3 accuracy accuracy NOUN bracis-28380 184 4 also also ADV bracis-28380 184 5 increases increase VERB bracis-28380 184 6 and and CCONJ bracis-28380 184 7 plateaus plateaus NOUN bracis-28380 184 8 at at ADP bracis-28380 184 9 around around ADV bracis-28380 184 10 96 96 NUM bracis-28380 184 11 % % NOUN bracis-28380 184 12 after after ADP bracis-28380 184 13 about about ADV bracis-28380 184 14 15 15 NUM bracis-28380 184 15 epochs epoch NOUN bracis-28380 184 16 . . PUNCT bracis-28380 185 1 the the DET bracis-28380 185 2 loss loss NOUN bracis-28380 185 3 decreases decrease VERB bracis-28380 185 4 more more ADV bracis-28380 185 5 smoothly smoothly ADV bracis-28380 185 6 and and CCONJ bracis-28380 185 7 plateaus plateaus NOUN bracis-28380 185 8 after after ADP bracis-28380 185 9 about about ADV bracis-28380 185 10 40 40 NUM bracis-28380 185 11 epochs epoch NOUN bracis-28380 185 12 . . PUNCT bracis-28380 186 1 these these DET bracis-28380 186 2 results result NOUN bracis-28380 186 3 suggest suggest VERB bracis-28380 186 4 that that SCONJ bracis-28380 186 5 the the DET bracis-28380 186 6 xception xception PROPN bracis-28380 186 7 model model NOUN bracis-28380 186 8 cnn cnn PROPN bracis-28380 186 9 can can AUX bracis-28380 186 10 achieve achieve VERB bracis-28380 186 11 even even ADV bracis-28380 186 12 higher high ADJ bracis-28380 186 13 accuracy accuracy NOUN bracis-28380 186 14 than than ADP bracis-28380 186 15 the the DET bracis-28380 186 16 basic basic ADJ bracis-28380 186 17 model model NOUN bracis-28380 186 18 cnn cnn PROPN bracis-28380 186 19 and and CCONJ bracis-28380 186 20 is be AUX bracis-28380 186 21 less less ADV bracis-28380 186 22 prone prone ADJ bracis-28380 186 23 to to ADP bracis-28380 186 24 overfitting overfitte VERB bracis-28380 186 25 . . PUNCT bracis-28380 187 1 the the DET bracis-28380 187 2 hyperparameters hyperparameter NOUN bracis-28380 187 3 used use VERB bracis-28380 187 4 appear appear VERB bracis-28380 187 5 to to PART bracis-28380 187 6 have have AUX bracis-28380 187 7 been be AUX bracis-28380 187 8 effectively effectively ADV bracis-28380 187 9 training train VERB bracis-28380 187 10 both both DET bracis-28380 187 11 models model NOUN bracis-28380 187 12 . . PUNCT bracis-28380 188 1 however however ADV bracis-28380 188 2 , , PUNCT bracis-28380 188 3 the the DET bracis-28380 188 4 xception xception PROPN bracis-28380 188 5 model model PROPN bracis-28380 188 6 cnn cnn PROPN bracis-28380 188 7 performs perform VERB bracis-28380 188 8 better well ADJ bracis-28380 188 9 than than ADP bracis-28380 188 10 the the DET bracis-28380 188 11 basic basic ADJ bracis-28380 188 12 model model NOUN bracis-28380 188 13 cnn cnn PROPN bracis-28380 188 14 , , PUNCT bracis-28380 188 15 likely likely ADJ bracis-28380 188 16 due due ADP bracis-28380 188 17 to to ADP bracis-28380 188 18 its its PRON bracis-28380 188 19 more more ADV bracis-28380 188 20 complex complex ADJ bracis-28380 188 21 architecture architecture NOUN bracis-28380 188 22 . . PUNCT bracis-28380 189 1 fig fig NOUN bracis-28380 189 2 . . PUNCT bracis-28380 190 1 3 3 X bracis-28380 190 2 . . X bracis-28380 190 3 learning learn VERB bracis-28380 190 4 curves curve NOUN bracis-28380 190 5 of of ADP bracis-28380 190 6 basic basic ADJ bracis-28380 190 7 model model NOUN bracis-28380 190 8 cnn cnn PROPN bracis-28380 190 9 full full ADJ bracis-28380 190 10 size size NOUN bracis-28380 190 11 image image NOUN bracis-28380 190 12 fig fig NOUN bracis-28380 190 13 . . PUNCT bracis-28380 191 1 4 4 X bracis-28380 191 2 . . X bracis-28380 191 3 learning learn VERB bracis-28380 191 4 curves curve NOUN bracis-28380 191 5 of of ADP bracis-28380 191 6 xception xception PROPN bracis-28380 191 7 model model PROPN bracis-28380 191 8 cnn cnn PROPN bracis-28380 191 9 full full ADJ bracis-28380 191 10 size size NOUN bracis-28380 191 11 image image NOUN bracis-28380 191 12 fig fig NOUN bracis-28380 191 13 . . PUNCT bracis-28380 192 1 5 5 X bracis-28380 192 2 . . X bracis-28380 192 3 roc roc PROPN bracis-28380 192 4 curve curve PROPN bracis-28380 192 5 basic basic ADJ bracis-28380 192 6 and and CCONJ bracis-28380 192 7 xception xception NOUN bracis-28380 192 8 models model NOUN bracis-28380 192 9 on on ADP bracis-28380 192 10 br35h br35h PROPN bracis-28380 192 11 dataset dataset PROPN bracis-28380 192 12 . . PUNCT bracis-28380 193 1 full full ADJ bracis-28380 193 2 size size NOUN bracis-28380 193 3 image image NOUN bracis-28380 193 4 the the DET bracis-28380 193 5 experimental experimental ADJ bracis-28380 193 6 results result NOUN bracis-28380 193 7 in in ADP bracis-28380 193 8 table table NOUN bracis-28380 193 9     SPACE bracis-28380 193 10 1 1 NUM bracis-28380 193 11 indicate indicate VERB bracis-28380 193 12 that that SCONJ bracis-28380 193 13 the the DET bracis-28380 193 14 basic basic ADJ bracis-28380 193 15 model model NOUN bracis-28380 193 16 and and CCONJ bracis-28380 193 17 xception xception NOUN bracis-28380 193 18 perform perform VERB bracis-28380 193 19 well well ADV bracis-28380 193 20 on on ADP bracis-28380 193 21 the the DET bracis-28380 193 22 br35h br35h ADJ bracis-28380 193 23 dataset dataset NOUN bracis-28380 193 24 , , PUNCT bracis-28380 193 25 achieving achieve VERB bracis-28380 193 26 high high ADJ bracis-28380 193 27 accuracy accuracy NOUN bracis-28380 193 28 and and CCONJ bracis-28380 193 29 auc auc NOUN bracis-28380 193 30 ( ( PUNCT bracis-28380 193 31 area area NOUN bracis-28380 193 32 under under ADP bracis-28380 193 33 the the DET bracis-28380 193 34 curve curve NOUN bracis-28380 193 35 ) ) PUNCT bracis-28380 193 36 values value NOUN bracis-28380 193 37 ( ( PUNCT bracis-28380 193 38 fig fig NOUN bracis-28380 193 39 . . PUNCT bracis-28380 193 40     SPACE bracis-28380 193 41 5 5 NUM bracis-28380 193 42 ) ) PUNCT bracis-28380 193 43 . . PUNCT bracis-28380 194 1 however however ADV bracis-28380 194 2 , , PUNCT bracis-28380 194 3 xception xception PROPN bracis-28380 194 4 outperforms outperform VERB bracis-28380 194 5 the the DET bracis-28380 194 6 basic basic ADJ bracis-28380 194 7 precision precision NOUN bracis-28380 194 8 , , PUNCT bracis-28380 194 9 recall recall NOUN bracis-28380 194 10 , , PUNCT bracis-28380 194 11 and and CCONJ bracis-28380 194 12 specificity specificity NOUN bracis-28380 194 13 model model NOUN bracis-28380 194 14 . . PUNCT bracis-28380 195 1 specifically specifically ADV bracis-28380 195 2 , , PUNCT bracis-28380 195 3 xception xception PROPN bracis-28380 195 4 achieves achieve VERB bracis-28380 195 5 a a DET bracis-28380 195 6 precision precision NOUN bracis-28380 195 7 of of ADP bracis-28380 195 8 96.71 96.71 NUM bracis-28380 195 9 % % NOUN bracis-28380 195 10 , , PUNCT bracis-28380 195 11 a a DET bracis-28380 195 12 recall recall NOUN bracis-28380 195 13 of of ADP bracis-28380 195 14 98 98 NUM bracis-28380 195 15 % % NOUN bracis-28380 195 16 , , PUNCT bracis-28380 195 17 and and CCONJ bracis-28380 195 18 a a DET bracis-28380 195 19 specificity specificity NOUN bracis-28380 195 20 of of ADP bracis-28380 195 21 96.67 96.67 NUM bracis-28380 195 22 % % NOUN bracis-28380 195 23 , , PUNCT bracis-28380 195 24 compared compare VERB bracis-28380 195 25 to to ADP bracis-28380 195 26 the the DET bracis-28380 195 27 basic basic ADJ bracis-28380 195 28 model model NOUN bracis-28380 195 29 ’s ’s PART bracis-28380 195 30 precision precision NOUN bracis-28380 195 31 of of ADP bracis-28380 195 32 96.03 96.03 NUM bracis-28380 195 33 % % NOUN bracis-28380 195 34 , , PUNCT bracis-28380 195 35 recall recall NOUN bracis-28380 195 36 of of ADP bracis-28380 195 37 96.67 96.67 NUM bracis-28380 195 38 % % NOUN bracis-28380 195 39 , , PUNCT bracis-28380 195 40 and and CCONJ bracis-28380 195 41 specificity specificity NOUN bracis-28380 195 42 of of ADP bracis-28380 195 43 96 96 NUM bracis-28380 195 44 % % NOUN bracis-28380 195 45 . . PUNCT bracis-28380 196 1 these these DET bracis-28380 196 2 results result NOUN bracis-28380 196 3 suggest suggest VERB bracis-28380 196 4 that that SCONJ bracis-28380 196 5 xception xception PROPN bracis-28380 196 6 can can AUX bracis-28380 196 7 better well ADV bracis-28380 196 8 distinguish distinguish VERB bracis-28380 196 9 between between ADP bracis-28380 196 10 positive positive ADJ bracis-28380 196 11 and and CCONJ bracis-28380 196 12 negative negative ADJ bracis-28380 196 13 instances instance NOUN bracis-28380 196 14 in in ADP bracis-28380 196 15 the the DET bracis-28380 196 16 br35h br35h ADJ bracis-28380 196 17 dataset dataset PROPN bracis-28380 196 18 . . PUNCT bracis-28380 197 1 fig fig NOUN bracis-28380 197 2 . . PUNCT bracis-28380 198 1 6 6 NUM bracis-28380 198 2 . . X bracis-28380 198 3 roc roc PROPN bracis-28380 198 4 curve curve PROPN bracis-28380 198 5 basic basic ADJ bracis-28380 198 6 and and CCONJ bracis-28380 198 7 xception xception NOUN bracis-28380 198 8 models model NOUN bracis-28380 198 9 on on ADP bracis-28380 198 10 sartaj sartaj NOUN bracis-28380 198 11 dataset dataset VERB bracis-28380 198 12 . . PUNCT bracis-28380 199 1 full full ADJ bracis-28380 199 2 size size NOUN bracis-28380 199 3 image image NOUN bracis-28380 199 4 on on ADP bracis-28380 199 5 the the DET bracis-28380 199 6 other other ADJ bracis-28380 199 7 hand hand NOUN bracis-28380 199 8 , , PUNCT bracis-28380 199 9 the the DET bracis-28380 199 10 test test NOUN bracis-28380 199 11 data datum NOUN bracis-28380 199 12 did do AUX bracis-28380 199 13 not not PART bracis-28380 199 14 perform perform VERB bracis-28380 199 15 well well ADV bracis-28380 199 16 in in ADP bracis-28380 199 17 the the DET bracis-28380 199 18 sartaj sartaj NOUN bracis-28380 199 19 dataset dataset VERB bracis-28380 199 20 . . PUNCT bracis-28380 200 1 the the DET bracis-28380 200 2 basic basic ADJ bracis-28380 200 3 model model NOUN bracis-28380 200 4 and and CCONJ bracis-28380 200 5 xception xception NOUN bracis-28380 200 6 achieve achieve VERB bracis-28380 200 7 moderate moderate ADJ bracis-28380 200 8 accuracy accuracy NOUN bracis-28380 200 9 and and CCONJ bracis-28380 200 10 auc auc NOUN bracis-28380 200 11 values value NOUN bracis-28380 200 12 ( ( PUNCT bracis-28380 200 13 fig fig NOUN bracis-28380 200 14 . . PUNCT bracis-28380 200 15     SPACE bracis-28380 200 16 6 6 NUM bracis-28380 200 17 ) ) PUNCT bracis-28380 200 18 , , PUNCT bracis-28380 200 19 but but CCONJ bracis-28380 200 20 xception xception NOUN bracis-28380 200 21 performs perform VERB bracis-28380 200 22 better well ADV bracis-28380 200 23 in in ADP bracis-28380 200 24 precision precision NOUN bracis-28380 200 25 , , PUNCT bracis-28380 200 26 recall recall NOUN bracis-28380 200 27 , , PUNCT bracis-28380 200 28 and and CCONJ bracis-28380 200 29 specificity specificity NOUN bracis-28380 200 30 . . PUNCT bracis-28380 201 1 xception xception PROPN bracis-28380 201 2 achieves achieve VERB bracis-28380 201 3 a a DET bracis-28380 201 4 precision precision NOUN bracis-28380 201 5 of of ADP bracis-28380 201 6 80.15 80.15 NUM bracis-28380 201 7 % % NOUN bracis-28380 201 8 , , PUNCT bracis-28380 201 9 a a DET bracis-28380 201 10 recall recall NOUN bracis-28380 201 11 of of ADP bracis-28380 201 12 94.78 94.78 NUM bracis-28380 201 13 % % NOUN bracis-28380 201 14 , , PUNCT bracis-28380 201 15 and and CCONJ bracis-28380 201 16 a a DET bracis-28380 201 17 specificity specificity NOUN bracis-28380 201 18 of of ADP bracis-28380 201 19 74.28 74.28 NUM bracis-28380 201 20 % % NOUN bracis-28380 201 21 , , PUNCT bracis-28380 201 22 compared compare VERB bracis-28380 201 23 to to ADP bracis-28380 201 24 the the DET bracis-28380 201 25 basic basic ADJ bracis-28380 201 26 model model NOUN bracis-28380 201 27 ’s ’s PART bracis-28380 201 28 precision precision NOUN bracis-28380 201 29 of of ADP bracis-28380 201 30 77.60 77.60 NUM bracis-28380 201 31 % % NOUN bracis-28380 201 32 , , PUNCT bracis-28380 201 33 recall recall NOUN bracis-28380 201 34 of of ADP bracis-28380 201 35 84.35 84.35 NUM bracis-28380 201 36 % % NOUN bracis-28380 201 37 , , PUNCT bracis-28380 201 38 and and CCONJ bracis-28380 201 39 specificity specificity NOUN bracis-28380 201 40 of of ADP bracis-28380 201 41 73.3 73.3 NUM bracis-28380 201 42 % % NOUN bracis-28380 201 43 . . PUNCT bracis-28380 202 1 despite despite SCONJ bracis-28380 202 2 these these DET bracis-28380 202 3 improvements improvement NOUN bracis-28380 202 4 , , PUNCT bracis-28380 202 5 the the DET bracis-28380 202 6 overall overall ADJ bracis-28380 202 7 performance performance NOUN bracis-28380 202 8 of of ADP bracis-28380 202 9 the the DET bracis-28380 202 10 sartaj sartaj NOUN bracis-28380 202 11 dataset dataset VERB bracis-28380 202 12 is be AUX bracis-28380 202 13 still still ADV bracis-28380 202 14 relatively relatively ADV bracis-28380 202 15 low low ADJ bracis-28380 202 16 . . PUNCT bracis-28380 203 1 one one NUM bracis-28380 203 2 hypothesis hypothesis NOUN bracis-28380 203 3 for for ADP bracis-28380 203 4 this this DET bracis-28380 203 5 low low ADJ bracis-28380 203 6 performance performance NOUN bracis-28380 203 7 could could AUX bracis-28380 203 8 be be AUX bracis-28380 203 9 due due ADJ bracis-28380 203 10 to to ADP bracis-28380 203 11 the the DET bracis-28380 203 12 variability variability NOUN bracis-28380 203 13 of of ADP bracis-28380 203 14 the the DET bracis-28380 203 15 sartaj sartaj NOUN bracis-28380 203 16 dataset dataset VERB bracis-28380 203 17 , , PUNCT bracis-28380 203 18 which which PRON bracis-28380 203 19 may may AUX bracis-28380 203 20 contain contain VERB bracis-28380 203 21 variations variation NOUN bracis-28380 203 22 in in ADP bracis-28380 203 23 image image NOUN bracis-28380 203 24 features feature NOUN bracis-28380 203 25 . . PUNCT bracis-28380 204 1 it it PRON bracis-28380 204 2 could could AUX bracis-28380 204 3 be be AUX bracis-28380 204 4 due due ADJ bracis-28380 204 5 to to ADP bracis-28380 204 6 differences difference NOUN bracis-28380 204 7 in in ADP bracis-28380 204 8 imaging imaging NOUN bracis-28380 204 9 capture capture NOUN bracis-28380 204 10 techniques technique NOUN bracis-28380 204 11 , , PUNCT bracis-28380 204 12 image image NOUN bracis-28380 204 13 quality quality NOUN bracis-28380 204 14 , , PUNCT bracis-28380 204 15 or or CCONJ bracis-28380 204 16 specific specific ADJ bracis-28380 204 17 characteristics characteristic NOUN bracis-28380 204 18 of of ADP bracis-28380 204 19 the the DET bracis-28380 204 20 tumors tumor NOUN bracis-28380 204 21 in in ADP bracis-28380 204 22 the the DET bracis-28380 204 23 dataset dataset NOUN bracis-28380 204 24 . . PUNCT bracis-28380 205 1 the the DET bracis-28380 205 2 models model NOUN bracis-28380 205 3 were be AUX bracis-28380 205 4 trained train VERB bracis-28380 205 5 on on ADP bracis-28380 205 6 a a DET bracis-28380 205 7 different different ADJ bracis-28380 205 8 dataset dataset NOUN bracis-28380 205 9 than than ADP bracis-28380 205 10 the the DET bracis-28380 205 11 sartaj sartaj NOUN bracis-28380 205 12 dataset dataset VERB bracis-28380 205 13 and and CCONJ bracis-28380 205 14 may may AUX bracis-28380 205 15 struggle struggle VERB bracis-28380 205 16 to to PART bracis-28380 205 17 generalize generalize VERB bracis-28380 205 18 effectively effectively ADV bracis-28380 205 19 . . PUNCT bracis-28380 206 1 the the DET bracis-28380 206 2 results result NOUN bracis-28380 206 3 demonstrate demonstrate VERB bracis-28380 206 4 the the DET bracis-28380 206 5 importance importance NOUN bracis-28380 206 6 of of ADP bracis-28380 206 7 evaluating evaluate VERB bracis-28380 206 8 a a DET bracis-28380 206 9 model model NOUN bracis-28380 206 10 on on ADP bracis-28380 206 11 multiple multiple ADJ bracis-28380 206 12 datasets dataset NOUN bracis-28380 206 13 to to PART bracis-28380 206 14 ensure ensure VERB bracis-28380 206 15 its its PRON bracis-28380 206 16 generalization generalization NOUN bracis-28380 206 17 performance performance NOUN bracis-28380 206 18 . . PUNCT bracis-28380 207 1 although although SCONJ bracis-28380 207 2 the the DET bracis-28380 207 3 models model NOUN bracis-28380 207 4 are be AUX bracis-28380 207 5 promising promise VERB bracis-28380 207 6 on on ADP bracis-28380 207 7 the the DET bracis-28380 207 8 br35h br35h ADJ bracis-28380 207 9 dataset dataset NOUN bracis-28380 207 10 , , PUNCT bracis-28380 207 11 it it PRON bracis-28380 207 12 has have VERB bracis-28380 207 13 yet yet ADV bracis-28380 207 14 to to PART bracis-28380 207 15 show show VERB bracis-28380 207 16 good good ADJ bracis-28380 207 17 generalizability generalizability NOUN bracis-28380 207 18 on on ADP bracis-28380 207 19 new new ADJ bracis-28380 207 20 images image NOUN bracis-28380 207 21 . . PUNCT bracis-28380 208 1 more more ADJ bracis-28380 208 2 investigations investigation NOUN bracis-28380 208 3 are be AUX bracis-28380 208 4 needed need VERB bracis-28380 208 5 to to PART bracis-28380 208 6 improve improve VERB bracis-28380 208 7 its its PRON bracis-28380 208 8 performance performance NOUN bracis-28380 208 9 on on ADP bracis-28380 208 10 the the DET bracis-28380 208 11 sartaj sartaj NOUN bracis-28380 208 12 and and CCONJ bracis-28380 208 13 in in ADP bracis-28380 208 14 other other ADJ bracis-28380 208 15 datasets dataset NOUN bracis-28380 208 16 . . PUNCT bracis-28380 209 1 table table NOUN bracis-28380 209 2 1 1 NUM bracis-28380 209 3 . . PUNCT bracis-28380 209 4 test test NOUN bracis-28380 209 5 results result NOUN bracis-28380 209 6 on on ADP bracis-28380 209 7 br35h br35h NOUN bracis-28380 209 8 and and CCONJ bracis-28380 209 9 sartaj sartaj NOUN bracis-28380 209 10 datasetsfull datasetsfull NOUN bracis-28380 209 11 size size NOUN bracis-28380 209 12 table table NOUN bracis-28380 209 13 the the DET bracis-28380 209 14 following follow VERB bracis-28380 209 15 topic topic NOUN bracis-28380 209 16 aims aim VERB bracis-28380 209 17 to to PART bracis-28380 209 18 apply apply VERB bracis-28380 209 19 the the DET bracis-28380 209 20 xai xai PROPN bracis-28380 209 21 techniques technique NOUN bracis-28380 209 22 , , PUNCT bracis-28380 209 23 following follow VERB bracis-28380 209 24 the the DET bracis-28380 209 25 interpretable interpretable ADJ bracis-28380 209 26 evaluation evaluation NOUN bracis-28380 209 27 of of ADP bracis-28380 209 28 the the DET bracis-28380 209 29 cnns cnn NOUN bracis-28380 209 30 ’ ' PUNCT bracis-28380 209 31 performance performance NOUN bracis-28380 209 32 . . PUNCT bracis-28380 210 1 our our PRON bracis-28380 210 2 investigation investigation NOUN bracis-28380 210 3 employed employ VERB bracis-28380 210 4 a a DET bracis-28380 210 5 triad triad NUM bracis-28380 210 6 of of ADP bracis-28380 210 7 xai xai PROPN bracis-28380 210 8 methodologies methodology NOUN bracis-28380 210 9 to to PART bracis-28380 210 10 extract extract VERB bracis-28380 210 11 insights insight NOUN bracis-28380 210 12 regarding regard VERB bracis-28380 210 13 the the DET bracis-28380 210 14 cnn cnn PROPN bracis-28380 210 15 models model NOUN bracis-28380 210 16 utilized utilize VERB bracis-28380 210 17 for for ADP bracis-28380 210 18 brain brain NOUN bracis-28380 210 19 tumor tumor NOUN bracis-28380 210 20 classification classification NOUN bracis-28380 210 21 . . PUNCT bracis-28380 211 1 the the DET bracis-28380 211 2 first first ADJ bracis-28380 211 3 technique technique NOUN bracis-28380 211 4 was be AUX bracis-28380 211 5 visualizing visualize VERB bracis-28380 211 6 images image NOUN bracis-28380 211 7 in in ADP bracis-28380 211 8 convolutional convolutional ADJ bracis-28380 211 9 layers layer NOUN bracis-28380 211 10 of of ADP bracis-28380 211 11 the the DET bracis-28380 211 12 basic basic ADJ bracis-28380 211 13 model model NOUN bracis-28380 211 14 , , PUNCT bracis-28380 211 15 which which PRON bracis-28380 211 16 showed show VERB bracis-28380 211 17 how how SCONJ bracis-28380 211 18 the the DET bracis-28380 211 19 input input NOUN bracis-28380 211 20 image image NOUN bracis-28380 211 21 ( ( PUNCT bracis-28380 211 22 fig fig NOUN bracis-28380 211 23 . . PUNCT bracis-28380 211 24     SPACE bracis-28380 211 25 7 7 X bracis-28380 211 26 ) ) PUNCT bracis-28380 211 27 was be AUX bracis-28380 211 28 processed process VERB bracis-28380 211 29 in in ADP bracis-28380 211 30 each each PRON bracis-28380 211 31 of of ADP bracis-28380 211 32 the the DET bracis-28380 211 33 three three NUM bracis-28380 211 34 convolutional convolutional ADJ bracis-28380 211 35 layers layer NOUN bracis-28380 211 36 ( ( PUNCT bracis-28380 211 37 fig fig NOUN bracis-28380 211 38 . . PUNCT bracis-28380 211 39     SPACE bracis-28380 211 40 8) 8) NUM bracis-28380 211 41 . . PUNCT bracis-28380 212 1 this this DET bracis-28380 212 2 technique technique NOUN bracis-28380 212 3 aimed aim VERB bracis-28380 212 4 to to PART bracis-28380 212 5 provide provide VERB bracis-28380 212 6 an an DET bracis-28380 212 7 understanding understanding NOUN bracis-28380 212 8 of of ADP bracis-28380 212 9 how how SCONJ bracis-28380 212 10 the the DET bracis-28380 212 11 model model NOUN bracis-28380 212 12 made make VERB bracis-28380 212 13 its its PRON bracis-28380 212 14 predictions prediction NOUN bracis-28380 212 15 and and CCONJ bracis-28380 212 16 identify identify VERB bracis-28380 212 17 which which PRON bracis-28380 212 18 parts part NOUN bracis-28380 212 19 of of ADP bracis-28380 212 20 the the DET bracis-28380 212 21 input input NOUN bracis-28380 212 22 image image NOUN bracis-28380 212 23 were be AUX bracis-28380 212 24 used use VERB bracis-28380 212 25 by by ADP bracis-28380 212 26 the the DET bracis-28380 212 27 model model NOUN bracis-28380 212 28 for for ADP bracis-28380 212 29 classification classification NOUN bracis-28380 212 30 . . PUNCT bracis-28380 213 1 these these DET bracis-28380 213 2 images image NOUN bracis-28380 213 3 allow allow VERB bracis-28380 213 4 us we PRON bracis-28380 213 5 to to PART bracis-28380 213 6 observe observe VERB bracis-28380 213 7 the the DET bracis-28380 213 8 features feature NOUN bracis-28380 213 9 learned learn VERB bracis-28380 213 10 by by ADP bracis-28380 213 11 the the DET bracis-28380 213 12 model model NOUN bracis-28380 213 13 at at ADP bracis-28380 213 14 each each DET bracis-28380 213 15 layer layer NOUN bracis-28380 213 16 and and CCONJ bracis-28380 213 17 the the DET bracis-28380 213 18 increasing increase VERB bracis-28380 213 19 complexity complexity NOUN bracis-28380 213 20 of of ADP bracis-28380 213 21 the the DET bracis-28380 213 22 representations representation NOUN bracis-28380 213 23 as as SCONJ bracis-28380 213 24 we we PRON bracis-28380 213 25 move move VERB bracis-28380 213 26 deeper deeply ADV bracis-28380 213 27 into into ADP bracis-28380 213 28 the the DET bracis-28380 213 29 network network NOUN bracis-28380 213 30 . . PUNCT bracis-28380 214 1 the the DET bracis-28380 214 2 grad grad NOUN bracis-28380 214 3 - - PUNCT bracis-28380 214 4 cam cam NOUN bracis-28380 214 5 technique technique NOUN bracis-28380 214 6 is be AUX bracis-28380 214 7 applied apply VERB bracis-28380 214 8 to to ADP bracis-28380 214 9 the the DET bracis-28380 214 10 basic basic ADJ bracis-28380 214 11 model model NOUN bracis-28380 214 12 ’s ’s PART bracis-28380 214 13 first first ADJ bracis-28380 214 14 , , PUNCT bracis-28380 214 15 second second ADJ bracis-28380 214 16 , , PUNCT bracis-28380 214 17 and and CCONJ bracis-28380 214 18 third third ADJ bracis-28380 214 19 convolutional convolutional ADJ bracis-28380 214 20 layers layer NOUN bracis-28380 214 21 ( ( PUNCT bracis-28380 214 22 fig fig NOUN bracis-28380 214 23 . . PUNCT bracis-28380 214 24     SPACE bracis-28380 214 25 9 9 NUM bracis-28380 214 26 ) ) PUNCT bracis-28380 214 27 . . PUNCT bracis-28380 215 1 for for ADP bracis-28380 215 2 each each DET bracis-28380 215 3 layer layer NOUN bracis-28380 215 4 , , PUNCT bracis-28380 215 5 the the DET bracis-28380 215 6 input input NOUN bracis-28380 215 7 image image NOUN bracis-28380 215 8 is be AUX bracis-28380 215 9 passed pass VERB bracis-28380 215 10 through through ADP bracis-28380 215 11 the the DET bracis-28380 215 12 model model NOUN bracis-28380 215 13 , , PUNCT bracis-28380 215 14 and and CCONJ bracis-28380 215 15 the the DET bracis-28380 215 16 gradient gradient ADJ bracis-28380 215 17 information information NOUN bracis-28380 215 18 flowing flow VERB bracis-28380 215 19 into into ADP bracis-28380 215 20 the the DET bracis-28380 215 21 layer layer NOUN bracis-28380 215 22 generates generate VERB bracis-28380 215 23 a a DET bracis-28380 215 24 heat heat NOUN bracis-28380 215 25 map map NOUN bracis-28380 215 26 highlighting highlight VERB bracis-28380 215 27 the the DET bracis-28380 215 28 most most ADV bracis-28380 215 29 important important ADJ bracis-28380 215 30 regions region NOUN bracis-28380 215 31 of of ADP bracis-28380 215 32 the the DET bracis-28380 215 33 image image NOUN bracis-28380 215 34 that that PRON bracis-28380 215 35 contributed contribute VERB bracis-28380 215 36 to to ADP bracis-28380 215 37 the the DET bracis-28380 215 38 classification classification NOUN bracis-28380 215 39 . . PUNCT bracis-28380 216 1 the the DET bracis-28380 216 2 heatmap heatmap NOUN bracis-28380 216 3 is be AUX bracis-28380 216 4 overlaid overlay VERB bracis-28380 216 5 on on ADP bracis-28380 216 6 the the DET bracis-28380 216 7 input input NOUN bracis-28380 216 8 image image NOUN bracis-28380 216 9 to to PART bracis-28380 216 10 show show VERB bracis-28380 216 11 the the DET bracis-28380 216 12 image image NOUN bracis-28380 216 13 regions region NOUN bracis-28380 216 14 that that PRON bracis-28380 216 15 contributed contribute VERB bracis-28380 216 16 to to ADP bracis-28380 216 17 the the DET bracis-28380 216 18 ranking ranking NOUN bracis-28380 216 19 . . PUNCT bracis-28380 217 1 the the DET bracis-28380 217 2 results result NOUN bracis-28380 217 3 show show VERB bracis-28380 217 4 that that SCONJ bracis-28380 217 5 grad grad NOUN bracis-28380 217 6 - - PUNCT bracis-28380 217 7 cam cam NOUN bracis-28380 217 8 can can AUX bracis-28380 217 9 provide provide VERB bracis-28380 217 10 valuable valuable ADJ bracis-28380 217 11 information information NOUN bracis-28380 217 12 about about ADP bracis-28380 217 13 the the DET bracis-28380 217 14 model model NOUN bracis-28380 217 15 ’s ’s PART bracis-28380 217 16 decision decision NOUN bracis-28380 217 17 - - PUNCT bracis-28380 217 18 making make VERB bracis-28380 217 19 process process NOUN bracis-28380 217 20 . . PUNCT bracis-28380 218 1 for for ADP bracis-28380 218 2 the the DET bracis-28380 218 3 first first ADJ bracis-28380 218 4 convolutional convolutional ADJ bracis-28380 218 5 layer layer NOUN bracis-28380 218 6 , , PUNCT bracis-28380 218 7 the the DET bracis-28380 218 8 heatmap heatmap NOUN bracis-28380 218 9 highlights highlight VERB bracis-28380 218 10 the the DET bracis-28380 218 11 edges edge NOUN bracis-28380 218 12 and and CCONJ bracis-28380 218 13 contours contours NOUN bracis-28380 218 14 of of ADP bracis-28380 218 15 the the DET bracis-28380 218 16 object object NOUN bracis-28380 218 17 in in ADP bracis-28380 218 18 the the DET bracis-28380 218 19 image image NOUN bracis-28380 218 20 . . PUNCT bracis-28380 219 1 the the DET bracis-28380 219 2 second second ADJ bracis-28380 219 3 convolutional convolutional ADJ bracis-28380 219 4 layer layer NOUN bracis-28380 219 5 highlights highlight VERB bracis-28380 219 6 more more ADJ bracis-28380 219 7 complex complex ADJ bracis-28380 219 8 features feature NOUN bracis-28380 219 9 , , PUNCT bracis-28380 219 10 such such ADJ bracis-28380 219 11 as as ADP bracis-28380 219 12 textures texture NOUN bracis-28380 219 13 and and CCONJ bracis-28380 219 14 patterns pattern NOUN bracis-28380 219 15 . . PUNCT bracis-28380 220 1 the the DET bracis-28380 220 2 heatmap heatmap NOUN bracis-28380 220 3 highlights highlight NOUN bracis-28380 220 4 even even ADV bracis-28380 220 5 more more ADV bracis-28380 220 6 complex complex ADJ bracis-28380 220 7 features feature NOUN bracis-28380 220 8 in in ADP bracis-28380 220 9 the the DET bracis-28380 220 10 third third ADJ bracis-28380 220 11 convolutional convolutional ADJ bracis-28380 220 12 layer layer NOUN bracis-28380 220 13 , , PUNCT bracis-28380 220 14 such such ADJ bracis-28380 220 15 as as ADP bracis-28380 220 16 shapes shape NOUN bracis-28380 220 17 and and CCONJ bracis-28380 220 18 structures structure NOUN bracis-28380 220 19 . . PUNCT bracis-28380 221 1 the the DET bracis-28380 221 2 third third ADJ bracis-28380 221 3 technique technique NOUN bracis-28380 221 4 was be AUX bracis-28380 221 5 the the DET bracis-28380 221 6 class class NOUN bracis-28380 221 7 activation activation NOUN bracis-28380 221 8 map map NOUN bracis-28380 221 9 , , PUNCT bracis-28380 221 10 applied apply VERB bracis-28380 221 11 to to ADP bracis-28380 221 12 the the DET bracis-28380 221 13 xception xception PROPN bracis-28380 221 14 model model PROPN bracis-28380 221 15 . . PUNCT bracis-28380 222 1 figure figure NOUN bracis-28380 222 2     SPACE bracis-28380 222 3 10 10 NUM bracis-28380 222 4 illustrates illustrate VERB bracis-28380 222 5 a a DET bracis-28380 222 6 false false ADJ bracis-28380 222 7 positive positive ADJ bracis-28380 222 8 output output NOUN bracis-28380 222 9 generated generate VERB bracis-28380 222 10 by by ADP bracis-28380 222 11 the the DET bracis-28380 222 12 xception xception PROPN bracis-28380 222 13 model model NOUN bracis-28380 222 14 , , PUNCT bracis-28380 222 15 showing show VERB bracis-28380 222 16 its its PRON bracis-28380 222 17 prediction prediction NOUN bracis-28380 222 18 with with ADP bracis-28380 222 19 100 100 NUM bracis-28380 222 20 % % NOUN bracis-28380 222 21 confidence confidence NOUN bracis-28380 222 22 . . PUNCT bracis-28380 223 1 the the DET bracis-28380 223 2 true true ADJ bracis-28380 223 3 class class NOUN bracis-28380 223 4 of of ADP bracis-28380 223 5 the the DET bracis-28380 223 6 image image NOUN bracis-28380 223 7 is be AUX bracis-28380 223 8 identified identify VERB bracis-28380 223 9 as as ADP bracis-28380 223 10 “ " PUNCT bracis-28380 223 11 no no DET bracis-28380 223 12 tumor tumor NOUN bracis-28380 223 13 ” " PUNCT bracis-28380 223 14 , , PUNCT bracis-28380 223 15 while while SCONJ bracis-28380 223 16 the the DET bracis-28380 223 17 model model NOUN bracis-28380 223 18 predicts predict VERB bracis-28380 223 19 it it PRON bracis-28380 223 20 as as ADP bracis-28380 223 21 “ " PUNCT bracis-28380 223 22 yes yes INTJ bracis-28380 223 23 tumor tumor NOUN bracis-28380 223 24 ” " PUNCT bracis-28380 223 25 . . PUNCT bracis-28380 224 1 in in ADP bracis-28380 224 2 addition addition NOUN bracis-28380 224 3 to to ADP bracis-28380 224 4 correctly correctly ADV bracis-28380 224 5 interpreting interpret VERB bracis-28380 224 6 classified classified ADJ bracis-28380 224 7 classes class NOUN bracis-28380 224 8 , , PUNCT bracis-28380 224 9 it it PRON bracis-28380 224 10 is be AUX bracis-28380 224 11 also also ADV bracis-28380 224 12 essential essential ADJ bracis-28380 224 13 to to PART bracis-28380 224 14 understand understand VERB bracis-28380 224 15 the the DET bracis-28380 224 16 factors factor NOUN bracis-28380 224 17 contributing contribute VERB bracis-28380 224 18 to to ADP bracis-28380 224 19 false false ADJ bracis-28380 224 20 positives positive NOUN bracis-28380 224 21 and and CCONJ bracis-28380 224 22 negatives negative NOUN bracis-28380 224 23 to to PART bracis-28380 224 24 improve improve VERB bracis-28380 224 25 model model NOUN bracis-28380 224 26 performance performance NOUN bracis-28380 224 27 , , PUNCT bracis-28380 224 28 reduce reduce VERB bracis-28380 224 29 classification classification NOUN bracis-28380 224 30 errors error NOUN bracis-28380 224 31 , , PUNCT bracis-28380 224 32 and and CCONJ bracis-28380 224 33 increase increase VERB bracis-28380 224 34 confidence confidence NOUN bracis-28380 224 35 in in ADP bracis-28380 224 36 model model NOUN bracis-28380 224 37 predictions prediction NOUN bracis-28380 224 38 . . PUNCT bracis-28380 225 1 interpretability interpretability NOUN bracis-28380 225 2 techniques technique NOUN bracis-28380 225 3 offer offer VERB bracis-28380 225 4 a a DET bracis-28380 225 5 unique unique ADJ bracis-28380 225 6 advantage advantage NOUN bracis-28380 225 7 , , PUNCT bracis-28380 225 8 allowing allow VERB bracis-28380 225 9 humans human NOUN bracis-28380 225 10 to to PART bracis-28380 225 11 study study VERB bracis-28380 225 12 models model NOUN bracis-28380 225 13 and and CCONJ bracis-28380 225 14 actively actively ADV bracis-28380 225 15 look look VERB bracis-28380 225 16 for for ADP bracis-28380 225 17 flaws flaw NOUN bracis-28380 225 18 . . PUNCT bracis-28380 226 1 when when SCONJ bracis-28380 226 2 flaws flaw NOUN bracis-28380 226 3 are be AUX bracis-28380 226 4 exposed expose VERB bracis-28380 226 5 and and CCONJ bracis-28380 226 6 understood understand VERB bracis-28380 226 7 , , PUNCT bracis-28380 226 8 we we PRON bracis-28380 226 9 can can AUX bracis-28380 226 10 better well ADV bracis-28380 226 11 understand understand VERB bracis-28380 226 12 the the DET bracis-28380 226 13 potential potential ADJ bracis-28380 226 14 risks risk NOUN bracis-28380 226 15 associated associate VERB bracis-28380 226 16 with with ADP bracis-28380 226 17 these these DET bracis-28380 226 18 models model NOUN bracis-28380 226 19 and and CCONJ bracis-28380 226 20 design design NOUN bracis-28380 226 21 safeguards safeguard NOUN bracis-28380 226 22 to to PART bracis-28380 226 23 mitigate mitigate VERB bracis-28380 226 24 them they PRON bracis-28380 226 25 . . PUNCT bracis-28380 227 1 it it PRON bracis-28380 227 2 could could AUX bracis-28380 227 3 ensure ensure VERB bracis-28380 227 4 that that SCONJ bracis-28380 227 5 deployed deploy VERB bracis-28380 227 6 systems system NOUN bracis-28380 227 7 meet meet VERB bracis-28380 227 8 transparency transparency NOUN bracis-28380 227 9 requirements requirement NOUN bracis-28380 227 10 and and CCONJ bracis-28380 227 11 are be AUX bracis-28380 227 12 more more ADV bracis-28380 227 13 reliable reliable ADJ bracis-28380 227 14 . . PUNCT bracis-28380 228 1 visualizing visualize VERB bracis-28380 228 2 the the DET bracis-28380 228 3 cnn cnn PROPN bracis-28380 228 4 layers layer NOUN bracis-28380 228 5 , , PUNCT bracis-28380 228 6 the the DET bracis-28380 228 7 cam cam NOUN bracis-28380 228 8 and and CCONJ bracis-28380 228 9 grad grad NOUN bracis-28380 228 10 - - PUNCT bracis-28380 228 11 cam cam NOUN bracis-28380 228 12 techniques technique NOUN bracis-28380 228 13 provide provide VERB bracis-28380 228 14 an an DET bracis-28380 228 15 understanding understanding NOUN bracis-28380 228 16 of of ADP bracis-28380 228 17 the the DET bracis-28380 228 18 inner inner ADJ bracis-28380 228 19 workings working NOUN bracis-28380 228 20 of of ADP bracis-28380 228 21 the the DET bracis-28380 228 22 cnn cnn PROPN bracis-28380 228 23 models model NOUN bracis-28380 228 24 used use VERB bracis-28380 228 25 to to PART bracis-28380 228 26 classify classify VERB bracis-28380 228 27 medical medical ADJ bracis-28380 228 28 images image NOUN bracis-28380 228 29 , , PUNCT bracis-28380 228 30 improving improve VERB bracis-28380 228 31 interpretability interpretability NOUN bracis-28380 228 32 and and CCONJ bracis-28380 228 33 consequently consequently ADV bracis-28380 228 34 making make VERB bracis-28380 228 35 the the DET bracis-28380 228 36 network network NOUN bracis-28380 228 37 more more ADV bracis-28380 228 38 transparent transparent ADJ bracis-28380 228 39 to to ADP bracis-28380 228 40 human human ADJ bracis-28380 228 41 eyes eye NOUN bracis-28380 228 42 . . PUNCT bracis-28380 229 1 these these DET bracis-28380 229 2 visualization visualization NOUN bracis-28380 229 3 techniques technique NOUN bracis-28380 229 4 help help VERB bracis-28380 229 5 explain explain VERB bracis-28380 229 6 the the DET bracis-28380 229 7 model model NOUN bracis-28380 229 8 ’s ’s PART bracis-28380 229 9 decision decision NOUN bracis-28380 229 10 - - PUNCT bracis-28380 229 11 making make VERB bracis-28380 229 12 process process NOUN bracis-28380 229 13 and and CCONJ bracis-28380 229 14 can can AUX bracis-28380 229 15 provide provide VERB bracis-28380 229 16 insights insight NOUN bracis-28380 229 17 into into ADP bracis-28380 229 18 improving improve VERB bracis-28380 229 19 model model NOUN bracis-28380 229 20 performance performance NOUN bracis-28380 229 21 . . PUNCT bracis-28380 230 1 5 5 NUM bracis-28380 230 2 conclusion conclusion NOUN bracis-28380 230 3 this this DET bracis-28380 230 4 paper paper NOUN bracis-28380 230 5 evaluated evaluate VERB bracis-28380 230 6 the the DET bracis-28380 230 7 performance performance NOUN bracis-28380 230 8 of of ADP bracis-28380 230 9 two two NUM bracis-28380 230 10 cnn cnn PROPN bracis-28380 230 11 models model NOUN bracis-28380 230 12 , , PUNCT bracis-28380 230 13 the the DET bracis-28380 230 14 basic basic ADJ bracis-28380 230 15 model model NOUN bracis-28380 230 16 and and CCONJ bracis-28380 230 17 xception xception NOUN bracis-28380 230 18 , , PUNCT bracis-28380 230 19 for for ADP bracis-28380 230 20 brain brain NOUN bracis-28380 230 21 tumor tumor NOUN bracis-28380 230 22 classification classification NOUN bracis-28380 230 23 on on ADP bracis-28380 230 24 two two NUM bracis-28380 230 25 datasets dataset NOUN bracis-28380 230 26 , , PUNCT bracis-28380 230 27 br35h br35h ADJ bracis-28380 230 28 and and CCONJ bracis-28380 230 29 sartaj sartaj NOUN bracis-28380 230 30 . . PUNCT bracis-28380 231 1 the the DET bracis-28380 231 2 accuracy accuracy NOUN bracis-28380 231 3 values value NOUN bracis-28380 231 4 of of ADP bracis-28380 231 5 the the DET bracis-28380 231 6 models model NOUN bracis-28380 231 7 reached reach VERB bracis-28380 231 8 96.33 96.33 NUM bracis-28380 231 9 % % NOUN bracis-28380 231 10 and and CCONJ bracis-28380 231 11 97.33 97.33 NUM bracis-28380 231 12 % % NOUN bracis-28380 231 13 in in ADP bracis-28380 231 14 the the DET bracis-28380 231 15 test test NOUN bracis-28380 231 16 set set NOUN bracis-28380 231 17 of of ADP bracis-28380 231 18 the the DET bracis-28380 231 19 database database NOUN bracis-28380 231 20 in in ADP bracis-28380 231 21 which which PRON bracis-28380 231 22 they they PRON bracis-28380 231 23 were be AUX bracis-28380 231 24 trained train VERB bracis-28380 231 25 . . PUNCT bracis-28380 232 1 the the DET bracis-28380 232 2 results result NOUN bracis-28380 232 3 obtained obtain VERB bracis-28380 232 4 here here ADV bracis-28380 232 5 are be AUX bracis-28380 232 6 similar similar ADJ bracis-28380 232 7 to to ADP bracis-28380 232 8 those those PRON bracis-28380 232 9 found find VERB bracis-28380 232 10 in in ADP bracis-28380 232 11 recent recent ADJ bracis-28380 232 12 research research NOUN bracis-28380 232 13 [ [ X bracis-28380 232 14 11 11 NUM bracis-28380 232 15 , , PUNCT bracis-28380 232 16 12 12 NUM bracis-28380 232 17 , , PUNCT bracis-28380 232 18 14 14 NUM bracis-28380 232 19 , , PUNCT bracis-28380 232 20 15 15 NUM bracis-28380 232 21 ] ] PUNCT bracis-28380 232 22 and and CCONJ bracis-28380 232 23 relatively relatively ADV bracis-28380 232 24 higher high ADJ bracis-28380 232 25 than than ADP bracis-28380 232 26 in in ADP bracis-28380 232 27 the the DET bracis-28380 232 28 study study NOUN bracis-28380 232 29 [ [ X bracis-28380 232 30 13 13 NUM bracis-28380 232 31 ] ] PUNCT bracis-28380 232 32 . . PUNCT bracis-28380 233 1 the the DET bracis-28380 233 2 results result NOUN bracis-28380 233 3 also also ADV bracis-28380 233 4 indicate indicate VERB bracis-28380 233 5 that that SCONJ bracis-28380 233 6 xception xception PROPN bracis-28380 233 7 outperforms outperform VERB bracis-28380 233 8 the the DET bracis-28380 233 9 basic basic ADJ bracis-28380 233 10 model model NOUN bracis-28380 233 11 on on ADP bracis-28380 233 12 both both DET bracis-28380 233 13 datasets dataset NOUN bracis-28380 233 14 . . PUNCT bracis-28380 234 1 however however ADV bracis-28380 234 2 , , PUNCT bracis-28380 234 3 the the DET bracis-28380 234 4 overall overall ADJ bracis-28380 234 5 performance performance NOUN bracis-28380 234 6 of of ADP bracis-28380 234 7 the the DET bracis-28380 234 8 sartaj sartaj NOUN bracis-28380 234 9 dataset dataset VERB bracis-28380 234 10 remains remain VERB bracis-28380 234 11 low low ADJ bracis-28380 234 12 , , PUNCT bracis-28380 234 13 suggesting suggest VERB bracis-28380 234 14 the the DET bracis-28380 234 15 need need NOUN bracis-28380 234 16 for for ADP bracis-28380 234 17 further further ADJ bracis-28380 234 18 investigation investigation NOUN bracis-28380 234 19 to to PART bracis-28380 234 20 improve improve VERB bracis-28380 234 21 model model NOUN bracis-28380 234 22 generalization generalization NOUN bracis-28380 234 23 . . PUNCT bracis-28380 235 1 using use VERB bracis-28380 235 2 different different ADJ bracis-28380 235 3 databases database NOUN bracis-28380 235 4 in in ADP bracis-28380 235 5 research research NOUN bracis-28380 235 6 is be AUX bracis-28380 235 7 crucial crucial ADJ bracis-28380 235 8 to to PART bracis-28380 235 9 support support VERB bracis-28380 235 10 robust robust ADJ bracis-28380 235 11 conclusions conclusion NOUN bracis-28380 235 12 about about ADP bracis-28380 235 13 the the DET bracis-28380 235 14 generalization generalization NOUN bracis-28380 235 15 of of ADP bracis-28380 235 16 a a DET bracis-28380 235 17 model model NOUN bracis-28380 235 18 . . PUNCT bracis-28380 236 1 it it PRON bracis-28380 236 2 helps help VERB bracis-28380 236 3 assess assess VERB bracis-28380 236 4 a a DET bracis-28380 236 5 model model NOUN bracis-28380 236 6 ’s ’s PART bracis-28380 236 7 ability ability NOUN bracis-28380 236 8 to to PART bracis-28380 236 9 generalize generalize VERB bracis-28380 236 10 diverse diverse ADJ bracis-28380 236 11 data datum NOUN bracis-28380 236 12 , , PUNCT bracis-28380 236 13 identify identify VERB bracis-28380 236 14 biases bias NOUN bracis-28380 236 15 , , PUNCT bracis-28380 236 16 increase increase VERB bracis-28380 236 17 reliability reliability NOUN bracis-28380 236 18 and and CCONJ bracis-28380 236 19 reproducibility reproducibility NOUN bracis-28380 236 20 , , PUNCT bracis-28380 236 21 and and CCONJ bracis-28380 236 22 validate validate VERB bracis-28380 236 23 real real ADJ bracis-28380 236 24 - - PUNCT bracis-28380 236 25 world world NOUN bracis-28380 236 26 applicability applicability NOUN bracis-28380 236 27 . . PUNCT bracis-28380 237 1 it it PRON bracis-28380 237 2 is be AUX bracis-28380 237 3 a a DET bracis-28380 237 4 common common ADJ bracis-28380 237 5 approach approach NOUN bracis-28380 237 6 that that SCONJ bracis-28380 237 7 researchers researcher NOUN bracis-28380 237 8 in in ADP bracis-28380 237 9 deep deep ADJ bracis-28380 237 10 learning learning NOUN bracis-28380 237 11 and and CCONJ bracis-28380 237 12 medical medical ADJ bracis-28380 237 13 imaging imaging NOUN bracis-28380 237 14 must must AUX bracis-28380 237 15 follow follow VERB bracis-28380 237 16 to to PART bracis-28380 237 17 ensure ensure VERB bracis-28380 237 18 the the DET bracis-28380 237 19 validity validity NOUN bracis-28380 237 20 and and CCONJ bracis-28380 237 21 usefulness usefulness NOUN bracis-28380 237 22 of of ADP bracis-28380 237 23 their their PRON bracis-28380 237 24 discoveries discovery NOUN bracis-28380 237 25 . . PUNCT bracis-28380 238 1 furthermore furthermore ADV bracis-28380 238 2 , , PUNCT bracis-28380 238 3 the the DET bracis-28380 238 4 paper paper NOUN bracis-28380 238 5 introduces introduce VERB bracis-28380 238 6 explainable explainable ADJ bracis-28380 238 7 artificial artificial ADJ bracis-28380 238 8 intelligence intelligence NOUN bracis-28380 238 9 ( ( PUNCT bracis-28380 238 10 xai xai PROPN bracis-28380 238 11 ) ) PUNCT bracis-28380 238 12 techniques technique NOUN bracis-28380 238 13 , , PUNCT bracis-28380 238 14 the the DET bracis-28380 238 15 visualization visualization NOUN bracis-28380 238 16 of of ADP bracis-28380 238 17 cnn cnn PROPN bracis-28380 238 18 layers layer NOUN bracis-28380 238 19 , , PUNCT bracis-28380 238 20 cam cam NOUN bracis-28380 238 21 , , PUNCT bracis-28380 238 22 and and CCONJ bracis-28380 238 23 grad grad NOUN bracis-28380 238 24 - - PUNCT bracis-28380 238 25 cam cam NOUN bracis-28380 238 26 techniques technique NOUN bracis-28380 238 27 to to PART bracis-28380 238 28 enhance enhance VERB bracis-28380 238 29 the the DET bracis-28380 238 30 cnn cnn PROPN bracis-28380 238 31 models model NOUN bracis-28380 238 32 ’ ' PUNCT bracis-28380 238 33 interpretability interpretability NOUN bracis-28380 238 34 . . PUNCT bracis-28380 239 1 the the DET bracis-28380 239 2 visualization visualization NOUN bracis-28380 239 3 of of ADP bracis-28380 239 4 cnn cnn PROPN bracis-28380 239 5 layers layer NOUN bracis-28380 239 6 demonstrates demonstrate VERB bracis-28380 239 7 the the DET bracis-28380 239 8 learned learn VERB bracis-28380 239 9 features feature NOUN bracis-28380 239 10 by by ADP bracis-28380 239 11 the the DET bracis-28380 239 12 model model NOUN bracis-28380 239 13 at at ADP bracis-28380 239 14 each each DET bracis-28380 239 15 layer layer NOUN bracis-28380 239 16 and and CCONJ bracis-28380 239 17 the the DET bracis-28380 239 18 increasing increase VERB bracis-28380 239 19 complexity complexity NOUN bracis-28380 239 20 of of ADP bracis-28380 239 21 the the DET bracis-28380 239 22 representations representation NOUN bracis-28380 239 23 . . PUNCT bracis-28380 240 1 these these DET bracis-28380 240 2 visualization visualization NOUN bracis-28380 240 3 techniques technique NOUN bracis-28380 240 4 improve improve VERB bracis-28380 240 5 the the DET bracis-28380 240 6 understanding understanding NOUN bracis-28380 240 7 of of ADP bracis-28380 240 8 the the DET bracis-28380 240 9 cnn cnn PROPN bracis-28380 240 10 models model NOUN bracis-28380 240 11 ’ ' PUNCT bracis-28380 240 12 decision decision NOUN bracis-28380 240 13 - - PUNCT bracis-28380 240 14 making make VERB bracis-28380 240 15 process process NOUN bracis-28380 240 16 and and CCONJ bracis-28380 240 17 could could AUX bracis-28380 240 18 provide provide VERB bracis-28380 240 19 valuable valuable ADJ bracis-28380 240 20 insights insight NOUN bracis-28380 240 21 into into ADP bracis-28380 240 22 enhancing enhance VERB bracis-28380 240 23 model model NOUN bracis-28380 240 24 performance performance NOUN bracis-28380 240 25 . . PUNCT bracis-28380 241 1 xai xai PROPN bracis-28380 241 2 techniques technique NOUN bracis-28380 241 3 have have VERB bracis-28380 241 4 several several ADJ bracis-28380 241 5 benefits benefit NOUN bracis-28380 241 6 , , PUNCT bracis-28380 241 7 including include VERB bracis-28380 241 8 increased increase VERB bracis-28380 241 9 trust trust NOUN bracis-28380 241 10 and and CCONJ bracis-28380 241 11 reliability reliability NOUN bracis-28380 241 12 of of ADP bracis-28380 241 13 the the DET bracis-28380 241 14 model model NOUN bracis-28380 241 15 ’s ’s PART bracis-28380 241 16 output output NOUN bracis-28380 241 17 , , PUNCT bracis-28380 241 18 better well ADJ bracis-28380 241 19 communication communication NOUN bracis-28380 241 20 between between ADP bracis-28380 241 21 medical medical ADJ bracis-28380 241 22 professionals professional NOUN bracis-28380 241 23 and and CCONJ bracis-28380 241 24 patients patient NOUN bracis-28380 241 25 , , PUNCT bracis-28380 241 26 and and CCONJ bracis-28380 241 27 potential potential ADJ bracis-28380 241 28 improvements improvement NOUN bracis-28380 241 29 in in ADP bracis-28380 241 30 model model NOUN bracis-28380 241 31 performance performance NOUN bracis-28380 241 32 . . PUNCT bracis-28380 242 1 additionally additionally ADV bracis-28380 242 2 , , PUNCT bracis-28380 242 3 the the DET bracis-28380 242 4 visualizations visualization NOUN bracis-28380 242 5 generated generate VERB bracis-28380 242 6 by by ADP bracis-28380 242 7 xai xai PROPN bracis-28380 242 8 techniques technique NOUN bracis-28380 242 9 can can AUX bracis-28380 242 10 assist assist VERB bracis-28380 242 11 radiologists radiologist NOUN bracis-28380 242 12 in in ADP bracis-28380 242 13 identifying identify VERB bracis-28380 242 14 the the DET bracis-28380 242 15 most most ADV bracis-28380 242 16 important important ADJ bracis-28380 242 17 features feature NOUN bracis-28380 242 18 in in ADP bracis-28380 242 19 medical medical ADJ bracis-28380 242 20 images image NOUN bracis-28380 242 21 , , PUNCT bracis-28380 242 22 leading lead VERB bracis-28380 242 23 to to ADP bracis-28380 242 24 more more ADV bracis-28380 242 25 accurate accurate ADJ bracis-28380 242 26 diagnoses diagnosis NOUN bracis-28380 242 27 . . PUNCT bracis-28380 243 1 however however ADV bracis-28380 243 2 , , PUNCT bracis-28380 243 3 there there PRON bracis-28380 243 4 are be VERB bracis-28380 243 5 also also ADV bracis-28380 243 6 some some DET bracis-28380 243 7 limitations limitation NOUN bracis-28380 243 8 to to PART bracis-28380 243 9 consider consider VERB bracis-28380 243 10 . . PUNCT bracis-28380 244 1 xai xai PROPN bracis-28380 244 2 techniques technique NOUN bracis-28380 244 3 can can AUX bracis-28380 244 4 be be AUX bracis-28380 244 5 computationally computationally ADV bracis-28380 244 6 expensive expensive ADJ bracis-28380 244 7 , , PUNCT bracis-28380 244 8 requiring require VERB bracis-28380 244 9 significant significant ADJ bracis-28380 244 10 processing processing NOUN bracis-28380 244 11 power power NOUN bracis-28380 244 12 and and CCONJ bracis-28380 244 13 time time NOUN bracis-28380 244 14 . . PUNCT bracis-28380 245 1 furthermore furthermore ADV bracis-28380 245 2 , , PUNCT bracis-28380 245 3 the the DET bracis-28380 245 4 visualizations visualization NOUN bracis-28380 245 5 generated generate VERB bracis-28380 245 6 by by ADP bracis-28380 245 7 these these DET bracis-28380 245 8 techniques technique NOUN bracis-28380 245 9 can can AUX bracis-28380 245 10 be be AUX bracis-28380 245 11 subjective subjective ADJ bracis-28380 245 12 . . PUNCT bracis-28380 246 1 therefore therefore ADV bracis-28380 246 2 , , PUNCT bracis-28380 246 3 it it PRON bracis-28380 246 4 is be AUX bracis-28380 246 5 crucial crucial ADJ bracis-28380 246 6 to to PART bracis-28380 246 7 perform perform VERB bracis-28380 246 8 thorough thorough ADJ bracis-28380 246 9 validation validation NOUN bracis-28380 246 10 and and CCONJ bracis-28380 246 11 testing testing NOUN bracis-28380 246 12 of of ADP bracis-28380 246 13 the the DET bracis-28380 246 14 models model NOUN bracis-28380 246 15 using use VERB bracis-28380 246 16 xai xai PROPN bracis-28380 246 17 techniques technique NOUN bracis-28380 246 18 to to PART bracis-28380 246 19 ensure ensure VERB bracis-28380 246 20 their their PRON bracis-28380 246 21 generalization generalization NOUN bracis-28380 246 22 to to ADP bracis-28380 246 23 new new ADJ bracis-28380 246 24 data datum NOUN bracis-28380 246 25 . . PUNCT bracis-28380 247 1 xai xai PROPN bracis-28380 247 2 techniques technique NOUN bracis-28380 247 3 have have AUX bracis-28380 247 4 demonstrated demonstrate VERB bracis-28380 247 5 significant significant ADJ bracis-28380 247 6 potential potential NOUN bracis-28380 247 7 for for ADP bracis-28380 247 8 improving improve VERB bracis-28380 247 9 the the DET bracis-28380 247 10 interpretability interpretability NOUN bracis-28380 247 11 of of ADP bracis-28380 247 12 cnn cnn PROPN bracis-28380 247 13 models model NOUN bracis-28380 247 14 in in ADP bracis-28380 247 15 medical medical ADJ bracis-28380 247 16 image image NOUN bracis-28380 247 17 classification classification NOUN bracis-28380 247 18 tasks task NOUN bracis-28380 247 19 . . PUNCT bracis-28380 248 1 their their PRON bracis-28380 248 2 use use NOUN bracis-28380 248 3 is be AUX bracis-28380 248 4 likely likely ADJ bracis-28380 248 5 to to PART bracis-28380 248 6 become become VERB bracis-28380 248 7 increasingly increasingly ADV bracis-28380 248 8 important important ADJ bracis-28380 248 9 in in ADP bracis-28380 248 10 the the DET bracis-28380 248 11 future future NOUN bracis-28380 248 12 . . PUNCT bracis-28380 249 1 nevertheless nevertheless ADV bracis-28380 249 2 , , PUNCT bracis-28380 249 3 further further ADJ bracis-28380 249 4 research research NOUN bracis-28380 249 5 and and CCONJ bracis-28380 249 6 development development NOUN bracis-28380 249 7 are be AUX bracis-28380 249 8 necessary necessary ADJ bracis-28380 249 9 to to PART bracis-28380 249 10 overcome overcome VERB bracis-28380 249 11 the the DET bracis-28380 249 12 limitations limitation NOUN bracis-28380 249 13 and and CCONJ bracis-28380 249 14 challenges challenge NOUN bracis-28380 249 15 associated associate VERB bracis-28380 249 16 with with ADP bracis-28380 249 17 these these DET bracis-28380 249 18 techniques technique NOUN bracis-28380 249 19 to to PART bracis-28380 249 20 ensure ensure VERB bracis-28380 249 21 their their PRON bracis-28380 249 22 reliability reliability NOUN bracis-28380 249 23 and and CCONJ bracis-28380 249 24 usefulness usefulness NOUN bracis-28380 249 25 in in ADP bracis-28380 249 26 real real ADJ bracis-28380 249 27 - - PUNCT bracis-28380 249 28 world world NOUN bracis-28380 249 29 medical medical ADJ bracis-28380 249 30 applications application NOUN bracis-28380 249 31 . . PUNCT bracis-28380 250 1 fig fig NOUN bracis-28380 250 2 . . PUNCT bracis-28380 251 1 7 7 X bracis-28380 251 2 . . X bracis-28380 251 3 basic basic ADJ bracis-28380 251 4 model model NOUN bracis-28380 251 5 the the DET bracis-28380 251 6 original original ADJ bracis-28380 251 7 image image NOUN bracis-28380 251 8 input input NOUN bracis-28380 251 9 is be AUX bracis-28380 251 10 on on ADP bracis-28380 251 11 the the DET bracis-28380 251 12 left left NOUN bracis-28380 251 13 , , PUNCT bracis-28380 251 14 and and CCONJ bracis-28380 251 15 the the DET bracis-28380 251 16 right right NOUN bracis-28380 251 17 is be AUX bracis-28380 251 18 in in ADP bracis-28380 251 19 the the DET bracis-28380 251 20 first first ADJ bracis-28380 251 21 convolutional convolutional ADJ bracis-28380 251 22 layer layer NOUN bracis-28380 251 23 . . PUNCT bracis-28380 252 1 full full ADJ bracis-28380 252 2 size size NOUN bracis-28380 252 3 image image NOUN bracis-28380 252 4 fig fig NOUN bracis-28380 252 5 . . PUNCT bracis-28380 253 1 8 8 NUM bracis-28380 253 2 . . X bracis-28380 253 3 basic basic ADJ bracis-28380 253 4 model model NOUN bracis-28380 253 5 visual visual ADJ bracis-28380 253 6 layers layer NOUN bracis-28380 253 7 technique technique VERB bracis-28380 253 8 1st 1st NOUN bracis-28380 253 9 , , PUNCT bracis-28380 253 10 2nd 2nd ADJ bracis-28380 253 11 and and CCONJ bracis-28380 253 12 3rd 3rd ADJ bracis-28380 253 13 convolutional convolutional ADJ bracis-28380 253 14 layers layer NOUN bracis-28380 253 15 showcasing showcase VERB bracis-28380 253 16 the the DET bracis-28380 253 17 learned learn VERB bracis-28380 253 18 features feature NOUN bracis-28380 253 19 and and CCONJ bracis-28380 253 20 their their PRON bracis-28380 253 21 growing grow VERB bracis-28380 253 22 complexity complexity NOUN bracis-28380 253 23 within within ADP bracis-28380 253 24 the the DET bracis-28380 253 25 network network NOUN bracis-28380 253 26 . . PUNCT bracis-28380 254 1 full full ADJ bracis-28380 254 2 size size NOUN bracis-28380 254 3 image image NOUN bracis-28380 254 4 fig fig NOUN bracis-28380 254 5 . . PUNCT bracis-28380 255 1 9 9 X bracis-28380 255 2 . . X bracis-28380 255 3 grad grad NOUN bracis-28380 255 4 - - PUNCT bracis-28380 255 5 cam cam NOUN bracis-28380 255 6 analysis analysis NOUN bracis-28380 255 7 for for ADP bracis-28380 255 8 basic basic ADJ bracis-28380 255 9 model model NOUN bracis-28380 255 10 ( ( PUNCT bracis-28380 255 11 1st 1st ADJ bracis-28380 255 12 , , PUNCT bracis-28380 255 13 2nd 2nd ADJ bracis-28380 255 14 , , PUNCT bracis-28380 255 15 and and CCONJ bracis-28380 255 16 3rd 3rd ADJ bracis-28380 255 17 convolutional convolutional ADJ bracis-28380 255 18 layers layer NOUN bracis-28380 255 19 ) ) PUNCT bracis-28380 255 20 reveals reveal VERB bracis-28380 255 21 distinct distinct ADJ bracis-28380 255 22 insights insight NOUN bracis-28380 255 23 : : PUNCT bracis-28380 255 24 edges edge NOUN bracis-28380 255 25 and and CCONJ bracis-28380 255 26 contours contours NOUN bracis-28380 255 27 ( ( PUNCT bracis-28380 255 28 1st 1st NOUN bracis-28380 255 29 ) ) PUNCT bracis-28380 255 30 , , PUNCT bracis-28380 255 31 complex complex ADJ bracis-28380 255 32 textures texture NOUN bracis-28380 255 33 and and CCONJ bracis-28380 255 34 patterns pattern NOUN bracis-28380 255 35 ( ( PUNCT bracis-28380 255 36 2nd 2nd ADJ bracis-28380 255 37 ) ) PUNCT bracis-28380 255 38 and and CCONJ bracis-28380 255 39 shapes shape NOUN bracis-28380 255 40 and and CCONJ bracis-28380 255 41 structures(3rd structures(3rd NOUN bracis-28380 255 42 ) ) PUNCT bracis-28380 255 43 . . PUNCT bracis-28380 256 1 full full ADJ bracis-28380 256 2 size size NOUN bracis-28380 256 3 image image NOUN bracis-28380 256 4 fig fig NOUN bracis-28380 256 5 . . PUNCT bracis-28380 257 1 10 10 NUM bracis-28380 257 2 . . PUNCT bracis-28380 258 1 false false ADJ bracis-28380 258 2 positive positive ADJ bracis-28380 258 3 output output NOUN bracis-28380 258 4 and and CCONJ bracis-28380 258 5 after after ADP bracis-28380 258 6 cam cam NOUN bracis-28380 258 7 in in ADP bracis-28380 258 8 xception xception PROPN bracis-28380 258 9 model model PROPN bracis-28380 258 10 . . PUNCT bracis-28380 259 1 the the DET bracis-28380 259 2 image image NOUN bracis-28380 259 3 on on ADP bracis-28380 259 4 the the DET bracis-28380 259 5 left left NOUN bracis-28380 259 6 shows show VERB bracis-28380 259 7 the the DET bracis-28380 259 8 classification classification NOUN bracis-28380 259 9 of of ADP bracis-28380 259 10 the the DET bracis-28380 259 11 xception xception PROPN bracis-28380 259 12 model model NOUN bracis-28380 259 13 with with ADP bracis-28380 259 14 a a DET bracis-28380 259 15 100 100 NUM bracis-28380 259 16 % % NOUN bracis-28380 259 17 of of ADP bracis-28380 259 18 confidence confidence NOUN bracis-28380 259 19 . . PUNCT bracis-28380 260 1 true true ADJ bracis-28380 260 2 class class NOUN bracis-28380 260 3 : : PUNCT bracis-28380 260 4 “ " PUNCT bracis-28380 260 5 no no DET bracis-28380 260 6 tumor tumor NOUN bracis-28380 260 7 ” " PUNCT bracis-28380 260 8 ; ; PUNCT bracis-28380 260 9 predicted predict VERB bracis-28380 260 10 class class NOUN bracis-28380 260 11 : : PUNCT bracis-28380 260 12 “ " PUNCT bracis-28380 260 13 yes yes INTJ bracis-28380 260 14 ” " PUNCT bracis-28380 260 15 . . PUNCT bracis-28380 261 1 the the DET bracis-28380 261 2 heatmap heatmap NOUN bracis-28380 261 3 is be AUX bracis-28380 261 4 shown show VERB bracis-28380 261 5 in in ADP bracis-28380 261 6 the the DET bracis-28380 261 7 central central ADJ bracis-28380 261 8 image image NOUN bracis-28380 261 9 . . PUNCT bracis-28380 262 1 in in ADP bracis-28380 262 2 the the DET bracis-28380 262 3 image image NOUN bracis-28380 262 4 on on ADP bracis-28380 262 5 the the DET bracis-28380 262 6 right right NOUN bracis-28380 262 7 , , PUNCT bracis-28380 262 8 the the DET bracis-28380 262 9 area area NOUN bracis-28380 262 10 colored color VERB bracis-28380 262 11 is be AUX bracis-28380 262 12 those those PRON bracis-28380 262 13 that that PRON bracis-28380 262 14 increase increase VERB bracis-28380 262 15 the the DET bracis-28380 262 16 probability probability NOUN bracis-28380 262 17 that that SCONJ bracis-28380 262 18 the the DET bracis-28380 262 19 image image NOUN bracis-28380 262 20 belongs belong VERB bracis-28380 262 21 to to ADP bracis-28380 262 22 the the DET bracis-28380 262 23 yes yes INTJ bracis-28380 262 24 class class NOUN bracis-28380 262 25 ( ( PUNCT bracis-28380 262 26 tumor tumor NOUN bracis-28380 262 27 ) ) PUNCT bracis-28380 262 28 . . PUNCT bracis-28380 263 1 full full ADJ bracis-28380 263 2 size size NOUN bracis-28380 263 3 image image NOUN bracis-28380 263 4 references reference NOUN bracis-28380 263 5 câncer câncer NOUN bracis-28380 263 6 do do AUX bracis-28380 263 7 sistema sistema PROPN bracis-28380 263 8 nervoso nervoso PROPN bracis-28380 263 9 central central PROPN bracis-28380 263 10 . . PUNCT bracis-28380 264 1 instituto instituto PROPN bracis-28380 264 2 nacional nacional PROPN bracis-28380 264 3 de de PROPN bracis-28380 264 4 câncer câncer PROPN bracis-28380 264 5 - - PUNCT bracis-28380 264 6 inca inca PROPN bracis-28380 264 7 ( ( PUNCT bracis-28380 264 8 n.d n.d PROPN bracis-28380 264 9 . . PROPN bracis-28380 264 10 ) ) PUNCT bracis-28380 264 11 . . PUNCT bracis-28380 265 1 https://www.gov.br/inca/pt-br/assuntos/cancer/tipos/sistema-nervoso-central https://www.gov.br/inca/pt-br/assuntos/cancer/tipos/sistema-nervoso-central ADJ bracis-28380 265 2 brain brain NOUN bracis-28380 265 3 and and CCONJ bracis-28380 265 4 central central ADJ bracis-28380 265 5 nervous nervous ADJ bracis-28380 265 6 system system NOUN bracis-28380 265 7 cancer cancer NOUN bracis-28380 265 8 - - PUNCT bracis-28380 265 9 iarc iarc NOUN bracis-28380 265 10 . . PUNCT bracis-28380 266 1 ( ( PUNCT bracis-28380 266 2 n.d.).https://www.iarc.who.int n.d.).https://www.iarc.who.int NOUN bracis-28380 266 3 / / SYM bracis-28380 266 4 cancer cancer NOUN bracis-28380 266 5 - - PUNCT bracis-28380 266 6 type type NOUN bracis-28380 266 7 / / SYM bracis-28380 266 8 brain brain NOUN bracis-28380 266 9 - - PUNCT bracis-28380 266 10 and and CCONJ bracis-28380 266 11 - - PUNCT bracis-28380 266 12 central central ADJ bracis-28380 266 13 - - PUNCT bracis-28380 266 14 nervous nervous ADJ bracis-28380 266 15 - - PUNCT bracis-28380 266 16 system system NOUN bracis-28380 266 17 - - PUNCT bracis-28380 266 18 cancer/. cancer/. NOUN bracis-28380 266 19 accessed access VERB bracis-28380 266 20 16 16 NUM bracis-28380 266 21 may may PROPN bracis-28380 266 22 2023 2023 NUM bracis-28380 266 23 magadza magadza NOUN bracis-28380 266 24 , , PUNCT bracis-28380 266 25 t. t. PROPN bracis-28380 266 26 , , PUNCT bracis-28380 266 27 viriri viriri PROPN bracis-28380 266 28 , , PUNCT bracis-28380 266 29 s. s. PROPN bracis-28380 266 30 : : PUNCT bracis-28380 267 1 deep deep ADJ bracis-28380 267 2 learning learning NOUN bracis-28380 267 3 for for ADP bracis-28380 267 4 brain brain NOUN bracis-28380 267 5 tumor tumor NOUN bracis-28380 267 6 segmentation segmentation NOUN bracis-28380 267 7 : : PUNCT bracis-28380 267 8 a a DET bracis-28380 267 9 survey survey NOUN bracis-28380 267 10 of of ADP bracis-28380 267 11 state state NOUN bracis-28380 267 12 - - PUNCT bracis-28380 267 13 of of ADP bracis-28380 267 14 - - PUNCT bracis-28380 267 15 the the DET bracis-28380 267 16 - - PUNCT bracis-28380 267 17 art art NOUN bracis-28380 267 18 . . PUNCT bracis-28380 268 1 j. j. PROPN bracis-28380 268 2 imaging imaging PROPN bracis-28380 268 3 7(2 7(2 PROPN bracis-28380 268 4 ) ) PUNCT bracis-28380 268 5 , , PUNCT bracis-28380 268 6 19 19 NUM bracis-28380 268 7 ( ( PUNCT bracis-28380 268 8 2021 2021 NUM bracis-28380 268 9 ) ) PUNCT bracis-28380 268 10 . . PUNCT bracis-28380 269 1 https://doi.org/10.3390/jimaging7020019 https://doi.org/10.3390/jimaging7020019 PROPN bracis-28380 269 2 fernando fernando PROPN bracis-28380 269 3 , , PUNCT bracis-28380 269 4 k.r.m k.r.m PROPN bracis-28380 269 5 . . PUNCT bracis-28380 269 6 , , PUNCT bracis-28380 269 7 tsokos tsokos PROPN bracis-28380 269 8 , , PUNCT bracis-28380 269 9 c.p c.p PROPN bracis-28380 269 10 . . PROPN bracis-28380 269 11 : : PUNCT bracis-28380 270 1 deep deep ADJ bracis-28380 270 2 and and CCONJ bracis-28380 270 3 statistical statistical ADJ bracis-28380 270 4 learning learning NOUN bracis-28380 270 5 in in ADP bracis-28380 270 6 biomedical biomedical ADJ bracis-28380 270 7 imaging imaging NOUN bracis-28380 270 8 : : PUNCT bracis-28380 270 9 state state NOUN bracis-28380 270 10 of of ADP bracis-28380 270 11 the the DET bracis-28380 270 12 art art NOUN bracis-28380 270 13 in in ADP bracis-28380 270 14 3d 3d PROPN bracis-28380 270 15 mri mri PROPN bracis-28380 270 16 brain brain NOUN bracis-28380 270 17 tumor tumor NOUN bracis-28380 270 18 segmentation segmentation NOUN bracis-28380 270 19 . . PUNCT bracis-28380 271 1 inf inf PROPN bracis-28380 271 2 . . PUNCT bracis-28380 271 3 fusion fusion PROPN bracis-28380 271 4 92 92 NUM bracis-28380 271 5 , , PUNCT bracis-28380 271 6 450–465 450–465 NUM bracis-28380 271 7 ( ( PUNCT bracis-28380 271 8 2023 2023 NUM bracis-28380 271 9 ) ) PUNCT bracis-28380 271 10 . . PUNCT bracis-28380 272 1 https://doi.org/10.1016/j.inffus.2022.12.013 https://doi.org/10.1016/j.inffus.2022.12.013 PROPN bracis-28380 272 2 dsa dsa PROPN bracis-28380 272 3 , , PUNCT bracis-28380 272 4 e. e. PROPN bracis-28380 272 5 : : PUNCT bracis-28380 272 6 chapter chapter NOUN bracis-28380 272 7 43 43 NUM bracis-28380 272 8 pooling pool VERB bracis-28380 272 9 layers layer NOUN bracis-28380 272 10 in in ADP bracis-28380 272 11 convolutional convolutional ADJ bracis-28380 272 12 neural neural ADJ bracis-28380 272 13 networks network NOUN bracis-28380 272 14 . . PUNCT bracis-28380 273 1 in in ADP bracis-28380 273 2 : : PUNCT bracis-28380 273 3 deep deep ADJ bracis-28380 273 4 learning learning NOUN bracis-28380 273 5 book book NOUN bracis-28380 273 6 , , PUNCT bracis-28380 273 7 10 10 NUM bracis-28380 273 8 december december PROPN bracis-28380 273 9 2022 2022 NUM bracis-28380 273 10 . . PUNCT bracis-28380 274 1 https://www.deeplearningbook.com.br/camadas-de-pooling-em-redes-neurais-convolucionais https://www.deeplearningbook.com.br/camadas-de-pooling-em-redes-neurais-convolucionais PROPN bracis-28380 274 2 alzheimer alzheimer PROPN bracis-28380 274 3 ’s ’s PART bracis-28380 274 4 disease disease NOUN bracis-28380 274 5 diagnosis diagnosis NOUN bracis-28380 274 6 using use VERB bracis-28380 274 7 deep deep ADJ bracis-28380 274 8 learning learning NOUN bracis-28380 274 9 techniques technique NOUN bracis-28380 274 10 . . PUNCT bracis-28380 275 1 int int NOUN bracis-28380 275 2 . . PUNCT bracis-28380 276 1 j. j. PROPN bracis-28380 276 2 eng eng PROPN bracis-28380 276 3 . . PUNCT bracis-28380 277 1 adv adv PROPN bracis-28380 277 2 . . PUNCT bracis-28380 277 3 technol technol PROPN bracis-28380 277 4 . . PROPN bracis-28380 277 5 9(3 9(3 NUM bracis-28380 277 6 ) ) PUNCT bracis-28380 277 7 , , PUNCT bracis-28380 277 8 874–880 874–880 NUM bracis-28380 277 9 ( ( PUNCT bracis-28380 277 10 2020 2020 NUM bracis-28380 277 11 ) ) PUNCT bracis-28380 277 12 . . PUNCT bracis-28380 278 1 https://doi.org/10.35940/ijeat.c5345.029320 https://doi.org/10.35940/ijeat.c5345.029320 PROPN bracis-28380 278 2 lipton lipton PROPN bracis-28380 278 3 , , PUNCT bracis-28380 278 4 z.c z.c PROPN bracis-28380 278 5 . . PROPN bracis-28380 278 6 : : PUNCT bracis-28380 279 1 the the DET bracis-28380 279 2 mythos mythos NOUN bracis-28380 279 3 of of ADP bracis-28380 279 4 model model NOUN bracis-28380 279 5 interpretability interpretability NOUN bracis-28380 279 6 . . PUNCT bracis-28380 280 1 commun commun PROPN bracis-28380 280 2 . . PUNCT bracis-28380 280 3 acm acm PROPN bracis-28380 280 4 61(10 61(10 NUM bracis-28380 280 5 ) ) PUNCT bracis-28380 280 6 , , PUNCT bracis-28380 280 7 3643 3643 NUM bracis-28380 280 8 ( ( PUNCT bracis-28380 280 9 2018 2018 NUM bracis-28380 280 10 ) ) PUNCT bracis-28380 280 11 . . PUNCT bracis-28380 281 1 https://doi.org/10.1145/3233231 https://doi.org/10.1145/3233231 VERB bracis-28380 281 2 br br PROPN bracis-28380 281 3 35h 35h NUM bracis-28380 281 4 : : PUNCT bracis-28380 281 5 brain brain NOUN bracis-28380 281 6 tumor tumor NOUN bracis-28380 281 7 detection detection NOUN bracis-28380 281 8 2020 2020 NUM bracis-28380 281 9 ( ( PUNCT bracis-28380 281 10 n.d n.d PROPN bracis-28380 281 11 . . PROPN bracis-28380 281 12 ) ) PUNCT bracis-28380 281 13 . . PUNCT bracis-28380 282 1 www.kaggle.com www.kaggle.com PROPN bracis-28380 282 2 , , PUNCT bracis-28380 282 3 https://www.kaggle.com/datasets/ahmedhamada0/brain-tumor-detection https://www.kaggle.com/datasets/ahmedhamada0/brain-tumor-detection NOUN bracis-28380 282 4 brain brain NOUN bracis-28380 282 5 tumor tumor NOUN bracis-28380 282 6 classification classification NOUN bracis-28380 282 7 ( ( PUNCT bracis-28380 282 8 mri mri NOUN bracis-28380 282 9 ) ) PUNCT bracis-28380 282 10 ( ( PUNCT bracis-28380 282 11 n.d n.d PROPN bracis-28380 282 12 . . PROPN bracis-28380 282 13 ) ) PUNCT bracis-28380 282 14 . . PUNCT bracis-28380 283 1 www.kaggle.com www.kaggle.com PROPN bracis-28380 283 2 , , PUNCT bracis-28380 283 3 https://www.kaggle.com/datasets/sartajbhuvaji/brain-tumor-classification-mri https://www.kaggle.com/datasets/sartajbhuvaji/brain-tumor-classification-mri PROPN bracis-28380 283 4 doshi doshi PROPN bracis-28380 283 5 - - PUNCT bracis-28380 283 6 velez velez PROPN bracis-28380 283 7 , , PUNCT bracis-28380 283 8 f. f. PROPN bracis-28380 283 9 , , PUNCT bracis-28380 283 10 kim kim PROPN bracis-28380 283 11 , , PUNCT bracis-28380 283 12 b.:towards b.:toward VERB bracis-28380 283 13 a a DET bracis-28380 283 14 rigorous rigorous ADJ bracis-28380 283 15 science science NOUN bracis-28380 283 16 of of ADP bracis-28380 283 17 interpretable interpretable ADJ bracis-28380 283 18 machine machine NOUN bracis-28380 283 19 learning learning NOUN bracis-28380 283 20 . . PUNCT bracis-28380 284 1 arxiv:1702.08608 arxiv:1702.08608 NOUN bracis-28380 285 1 [ [ X bracis-28380 285 2 cs cs X bracis-28380 285 3 , , PUNCT bracis-28380 285 4 stat stat NOUN bracis-28380 285 5 ] ] PUNCT bracis-28380 285 6 . . PUNCT bracis-28380 286 1 https://arxiv.org/abs/1702.08608 https://arxiv.org/abs/1702.08608 NOUN bracis-28380 286 2 ( ( PUNCT bracis-28380 286 3 2017 2017 NUM bracis-28380 286 4 ) ) PUNCT bracis-28380 286 5 maqsood maqsood PROPN bracis-28380 286 6 , , PUNCT bracis-28380 286 7 s. s. PROPN bracis-28380 286 8 , , PUNCT bracis-28380 286 9 damaševičius damaševičius PROPN bracis-28380 286 10 , , PUNCT bracis-28380 286 11 r. r. PROPN bracis-28380 286 12 , , PUNCT bracis-28380 286 13 maskeliūnas maskeliūnas PROPN bracis-28380 286 14 , , PUNCT bracis-28380 286 15 r. r. PROPN bracis-28380 286 16 : : PUNCT bracis-28380 286 17 multi multi ADJ bracis-28380 286 18 - - ADJ bracis-28380 286 19 modal modal ADJ bracis-28380 286 20 brain brain NOUN bracis-28380 286 21 tumor tumor NOUN bracis-28380 286 22 detection detection NOUN bracis-28380 286 23 using use VERB bracis-28380 286 24 deep deep ADJ bracis-28380 286 25 neural neural ADJ bracis-28380 286 26 network network NOUN bracis-28380 286 27 and and CCONJ bracis-28380 286 28 multiclass multiclass ADJ bracis-28380 286 29 svm svm PROPN bracis-28380 286 30 . . PROPN bracis-28380 286 31 medicina medicina PROPN bracis-28380 286 32 58(8 58(8 NUM bracis-28380 286 33 ) ) PUNCT bracis-28380 286 34 , , PUNCT bracis-28380 286 35 1090 1090 NUM bracis-28380 286 36 ( ( PUNCT bracis-28380 286 37 2022 2022 NUM bracis-28380 286 38 ) ) PUNCT bracis-28380 286 39 . . PUNCT bracis-28380 287 1 https://doi.org/10.3390/medicina58081090 https://doi.org/10.3390/medicina58081090 PROPN bracis-28380 287 2 gómez gómez NOUN bracis-28380 287 3 - - PUNCT bracis-28380 287 4 guzmán guzmán NOUN bracis-28380 287 5 , , PUNCT bracis-28380 287 6 m.a m.a PROPN bracis-28380 287 7 . . PROPN bracis-28380 287 8 , , PUNCT bracis-28380 287 9 et et PROPN bracis-28380 287 10 al al PROPN bracis-28380 287 11 . . PUNCT bracis-28380 287 12 : : PUNCT bracis-28380 288 1 classifying classify VERB bracis-28380 288 2 brain brain NOUN bracis-28380 288 3 tumors tumor NOUN bracis-28380 288 4 on on ADP bracis-28380 288 5 magnetic magnetic ADJ bracis-28380 288 6 resonance resonance NOUN bracis-28380 288 7 imaging imaging NOUN bracis-28380 288 8 by by ADP bracis-28380 288 9 using use VERB bracis-28380 288 10 convolutional convolutional ADJ bracis-28380 288 11 neural neural ADJ bracis-28380 288 12 networks network NOUN bracis-28380 288 13 . . PUNCT bracis-28380 289 1 electronics electronic NOUN bracis-28380 289 2 12(4 12(4 NUM bracis-28380 289 3 ) ) PUNCT bracis-28380 289 4 , , PUNCT bracis-28380 289 5 955 955 NUM bracis-28380 289 6 ( ( PUNCT bracis-28380 289 7 2023 2023 NUM bracis-28380 289 8 ) ) PUNCT bracis-28380 289 9 . . PUNCT bracis-28380 290 1 https://doi.org/10.3390/electronics12040955 https://doi.org/10.3390/electronics12040955 NOUN bracis-28380 290 2 gaur gaur NOUN bracis-28380 290 3 , , PUNCT bracis-28380 290 4 l. l. PROPN bracis-28380 290 5 , , PUNCT bracis-28380 290 6 bhandari bhandari PROPN bracis-28380 290 7 , , PUNCT bracis-28380 290 8 m. m. NOUN bracis-28380 290 9 , , PUNCT bracis-28380 290 10 razdan razdan PROPN bracis-28380 290 11 , , PUNCT bracis-28380 290 12 t. t. PROPN bracis-28380 290 13 , , PUNCT bracis-28380 290 14 mallik mallik PROPN bracis-28380 290 15 , , PUNCT bracis-28380 290 16 s. s. PROPN bracis-28380 290 17 , , PUNCT bracis-28380 290 18 zhao zhao PROPN bracis-28380 290 19 , , PUNCT bracis-28380 290 20 z. z. PROPN bracis-28380 290 21 : : PUNCT bracis-28380 290 22 explanation explanation NOUN bracis-28380 290 23 - - PUNCT bracis-28380 290 24 driven drive VERB bracis-28380 290 25 deep deep ADJ bracis-28380 290 26 learning learning NOUN bracis-28380 290 27 model model NOUN bracis-28380 290 28 for for ADP bracis-28380 290 29 prediction prediction NOUN bracis-28380 290 30 of of ADP bracis-28380 290 31 brain brain NOUN bracis-28380 290 32 tumour tumour NOUN bracis-28380 290 33 status status NOUN bracis-28380 290 34 using use VERB bracis-28380 290 35 mri mri NOUN bracis-28380 290 36 image image NOUN bracis-28380 290 37 data datum NOUN bracis-28380 290 38 . . PUNCT bracis-28380 291 1 front front NOUN bracis-28380 291 2 . . PUNCT bracis-28380 292 1 genet genet NOUN bracis-28380 292 2 . . PUNCT bracis-28380 293 1 13 13 NUM bracis-28380 293 2 . . PUNCT bracis-28380 294 1 https://doi.org/10.3389/fgene.2022.822666 https://doi.org/10.3389/fgene.2022.822666 ADJ bracis-28380 294 2 marmolejo marmolejo PROPN bracis-28380 294 3 - - PUNCT bracis-28380 294 4 saucedo saucedo PROPN bracis-28380 294 5 , , PUNCT bracis-28380 294 6 j.a j.a PROPN bracis-28380 294 7 . . PROPN bracis-28380 294 8 , , PUNCT bracis-28380 294 9 kose kose PROPN bracis-28380 294 10 , , PUNCT bracis-28380 294 11 u. u. PROPN bracis-28380 294 12 : : PUNCT bracis-28380 294 13 numerical numerical PROPN bracis-28380 294 14 grad grad PROPN bracis-28380 294 15 - - PUNCT bracis-28380 294 16 cam cam NOUN bracis-28380 294 17 based base VERB bracis-28380 294 18 explainable explainable ADJ bracis-28380 294 19 convolutional convolutional ADJ bracis-28380 294 20 neural neural ADJ bracis-28380 294 21 network network NOUN bracis-28380 294 22 for for ADP bracis-28380 294 23 brain brain NOUN bracis-28380 294 24 tumor tumor NOUN bracis-28380 294 25 diagnosis diagnosis NOUN bracis-28380 294 26 . . PUNCT bracis-28380 295 1 mob mob NOUN bracis-28380 295 2 . . PUNCT bracis-28380 296 1 netw netw PROPN bracis-28380 296 2 . . PUNCT bracis-28380 297 1 appl appl PROPN bracis-28380 297 2 . . PUNCT bracis-28380 298 1 ( ( PUNCT bracis-28380 298 2 2022 2022 NUM bracis-28380 298 3 ) ) PUNCT bracis-28380 298 4 . . PUNCT bracis-28380 299 1 https://doi.org/10.1007/s11036-022-02021-6 https://doi.org/10.1007/s11036-022-02021-6 NUM bracis-28380 299 2 article article NOUN bracis-28380 299 3     SPACE bracis-28380 299 4 google google PROPN bracis-28380 299 5 scholar scholar NOUN bracis-28380 299 6     SPACE bracis-28380 299 7 islam islam PROPN bracis-28380 299 8 , , PUNCT bracis-28380 299 9 md.a md.a NOUN bracis-28380 299 10 . . PUNCT bracis-28380 299 11 , , PUNCT bracis-28380 299 12 et et PROPN bracis-28380 299 13 al.:a al.:a ADV bracis-28380 299 14 low low PROPN bracis-28380 299 15 parametric parametric PROPN bracis-28380 299 16 cnn cnn PROPN bracis-28380 299 17 based base VERB bracis-28380 299 18 solution solution NOUN bracis-28380 299 19 to to PART bracis-28380 299 20 efficiently efficiently ADV bracis-28380 299 21 detect detect VERB bracis-28380 299 22 brain brain NOUN bracis-28380 299 23 tumor tumor NOUN bracis-28380 299 24 cells cell NOUN bracis-28380 299 25 from from ADP bracis-28380 299 26 ultrasound ultrasound ADJ bracis-28380 299 27 scans scan NOUN bracis-28380 299 28 ( ( PUNCT bracis-28380 299 29 2023 2023 NUM bracis-28380 299 30 ) ) PUNCT bracis-28380 299 31 . . PUNCT bracis-28380 300 1 https://doi.org/10.1109/ccwc57344.2023.10099302 https://doi.org/10.1109/ccwc57344.2023.10099302 PROPN bracis-28380 300 2 westphal westphal PROPN bracis-28380 300 3 , , PUNCT bracis-28380 300 4 e. e. PROPN bracis-28380 300 5 , , PUNCT bracis-28380 300 6 seitz seitz PROPN bracis-28380 300 7 , , PUNCT bracis-28380 300 8 h. h. PROPN bracis-28380 300 9 : : PUNCT bracis-28380 300 10 a a DET bracis-28380 300 11 machine machine NOUN bracis-28380 300 12 learning learn VERB bracis-28380 300 13 method method NOUN bracis-28380 300 14 for for ADP bracis-28380 300 15 defect defect NOUN bracis-28380 300 16 detection detection NOUN bracis-28380 300 17 and and CCONJ bracis-28380 300 18 visualization visualization NOUN bracis-28380 300 19 in in ADP bracis-28380 300 20 selective selective ADJ bracis-28380 300 21 laser laser NOUN bracis-28380 300 22 sintering sintering NOUN bracis-28380 300 23 based base VERB bracis-28380 300 24 on on ADP bracis-28380 300 25 convolutional convolutional ADJ bracis-28380 300 26 neural neural ADJ bracis-28380 300 27 networks network NOUN bracis-28380 300 28 . . PUNCT bracis-28380 301 1 additive additive ADJ bracis-28380 301 2 manuf manuf PROPN bracis-28380 301 3 . . PROPN bracis-28380 301 4 41 41 NUM bracis-28380 301 5 , , PUNCT bracis-28380 301 6 101965 101965 NUM bracis-28380 301 7 ( ( PUNCT bracis-28380 301 8 2021 2021 NUM bracis-28380 301 9 ) ) PUNCT bracis-28380 301 10 . . PUNCT bracis-28380 302 1 https://doi.org/10.1016/j.addma.2021.101965 https://doi.org/10.1016/j.addma.2021.101965 PROPN bracis-28380 302 2 chollet chollet NOUN bracis-28380 302 3 , , PUNCT bracis-28380 302 4 f. f. PROPN bracis-28380 302 5 : : PUNCT bracis-28380 303 1 xception xception NOUN bracis-28380 303 2 : : PUNCT bracis-28380 303 3 deep deep ADJ bracis-28380 303 4 learning learn VERB bracis-28380 303 5 with with ADP bracis-28380 303 6 depthwise depthwise NOUN bracis-28380 303 7 separable separable ADJ bracis-28380 303 8 convolutions convolution NOUN bracis-28380 303 9 ( ( PUNCT bracis-28380 303 10 2017 2017 NUM bracis-28380 303 11 ) ) PUNCT bracis-28380 303 12 . . PUNCT bracis-28380 304 1 arxiv:1610.02357 arxiv:1610.02357 X bracis-28380 304 2 [ [ X bracis-28380 304 3 cs cs X bracis-28380 304 4 ] ] X bracis-28380 304 5 . . PUNCT bracis-28380 305 1 https://arxiv.org/abs/1610.02357v3 https://arxiv.org/abs/1610.02357v3 NUM bracis-28380 305 2 chollet chollet PROPN bracis-28380 305 3 , , PUNCT bracis-28380 305 4 f. f. PROPN bracis-28380 305 5 : : PUNCT bracis-28380 305 6 deep deep ADJ bracis-28380 305 7 learning learn VERB bracis-28380 305 8 with with ADP bracis-28380 305 9 python python PROPN bracis-28380 305 10 . . PUNCT bracis-28380 306 1 manning man VERB bracis-28380 306 2 publications publication NOUN bracis-28380 306 3 , , PUNCT bracis-28380 306 4 new new PROPN bracis-28380 306 5 york york PROPN bracis-28380 306 6 ( ( PUNCT bracis-28380 306 7 2017 2017 NUM bracis-28380 306 8 ) ) PUNCT bracis-28380 306 9 google google PROPN bracis-28380 306 10 scholar scholar NOUN bracis-28380 306 11     SPACE bracis-28380 306 12 selvaraju selvaraju NOUN bracis-28380 306 13 , , PUNCT bracis-28380 306 14 r.r r.r PROPN bracis-28380 306 15 . . PROPN bracis-28380 306 16 , , PUNCT bracis-28380 306 17 cogswell cogswell PROPN bracis-28380 306 18 , , PUNCT bracis-28380 306 19 m. m. NOUN bracis-28380 306 20 , , PUNCT bracis-28380 306 21 das das PROPN bracis-28380 306 22 , , PUNCT bracis-28380 306 23 a. a. NOUN bracis-28380 306 24 , , PUNCT bracis-28380 306 25 vedantam vedantam PROPN bracis-28380 306 26 , , PUNCT bracis-28380 306 27 r. r. PROPN bracis-28380 306 28 , , PUNCT bracis-28380 306 29 parikh parikh PROPN bracis-28380 306 30 , , PUNCT bracis-28380 306 31 d. d. PROPN bracis-28380 306 32 , , PUNCT bracis-28380 306 33 batra batra PROPN bracis-28380 306 34 , , PUNCT bracis-28380 306 35 d. d. PROPN bracis-28380 306 36 : : PUNCT bracis-28380 306 37 grad grad NOUN bracis-28380 306 38 - - PUNCT bracis-28380 306 39 cam cam PROPN bracis-28380 306 40 : : PUNCT bracis-28380 306 41 visual visual ADJ bracis-28380 306 42 explanations explanation NOUN bracis-28380 306 43 from from ADP bracis-28380 306 44 deep deep ADJ bracis-28380 306 45 networks network NOUN bracis-28380 306 46 via via ADP bracis-28380 306 47 gradient gradient NOUN bracis-28380 306 48 - - PUNCT bracis-28380 306 49 based base VERB bracis-28380 306 50 localization localization NOUN bracis-28380 306 51 . . PUNCT bracis-28380 307 1 int int NOUN bracis-28380 307 2 . . PUNCT bracis-28380 308 1 j. j. PROPN bracis-28380 308 2 comput comput PROPN bracis-28380 308 3 . . PUNCT bracis-28380 309 1 vision vision NOUN bracis-28380 309 2 128(2 128(2 NUM bracis-28380 309 3 ) ) PUNCT bracis-28380 309 4 , , PUNCT bracis-28380 309 5 336–359 336–359 NUM bracis-28380 309 6 ( ( PUNCT bracis-28380 309 7 2020 2020 NUM bracis-28380 309 8 ) ) PUNCT bracis-28380 309 9 . . PUNCT bracis-28380 310 1 https://doi.org/10.1007/s11263-019-01228-7 https://doi.org/10.1007/s11263-019-01228-7 NUM bracis-28380 310 2 download download NOUN bracis-28380 310 3 references reference NOUN bracis-28380 310 4 author author NOUN bracis-28380 310 5 information information NOUN bracis-28380 310 6 authors author NOUN bracis-28380 310 7 and and CCONJ bracis-28380 310 8 affiliations affiliation NOUN bracis-28380 310 9 federal federal PROPN bracis-28380 310 10 university university PROPN bracis-28380 310 11 of of ADP bracis-28380 310 12 health health PROPN bracis-28380 310 13 sciences sciences PROPN bracis-28380 310 14 of of ADP bracis-28380 310 15 porto porto PROPN bracis-28380 310 16 alegre alegre PROPN bracis-28380 310 17 , , PUNCT bracis-28380 310 18 porto porto PROPN bracis-28380 310 19 alegre alegre PROPN bracis-28380 310 20 , , PUNCT bracis-28380 310 21 rs r NOUN bracis-28380 310 22 , , PUNCT bracis-28380 310 23 90050 90050 NUM bracis-28380 310 24 - - SYM bracis-28380 310 25 170 170 NUM bracis-28380 310 26 , , PUNCT bracis-28380 310 27 brazil brazil PROPN bracis-28380 310 28 dieine dieine PROPN bracis-28380 310 29 estela estela PROPN bracis-28380 310 30 bernieri bernieri PROPN bracis-28380 310 31 schiavon schiavon PROPN bracis-28380 310 32 , , PUNCT bracis-28380 310 33     SPACE bracis-28380 310 34 carla carla PROPN bracis-28380 310 35 diniz diniz PROPN bracis-28380 310 36 lopes lopes PROPN bracis-28380 310 37 becker becker PROPN bracis-28380 310 38 , , PUNCT bracis-28380 310 39     SPACE bracis-28380 310 40 viviane viviane PROPN bracis-28380 310 41 rodrigues rodrigues PROPN bracis-28380 310 42 botelho botelho PROPN bracis-28380 310 43     SPACE bracis-28380 310 44 & & CCONJ bracis-28380 310 45     SPACE bracis-28380 310 46 thatiane thatiane NOUN bracis-28380 310 47 alves alves PROPN bracis-28380 310 48 pianoski pianoski PROPN bracis-28380 310 49 authors author NOUN bracis-28380 310 50 dieine dieine PROPN bracis-28380 310 51 estela estela PROPN bracis-28380 310 52 bernieri bernieri PROPN bracis-28380 310 53 schiavonview schiavonview PROPN bracis-28380 310 54 author author NOUN bracis-28380 310 55 publications publication NOUN bracis-28380 310 56 search search NOUN bracis-28380 310 57 author author NOUN bracis-28380 310 58 on on ADP bracis-28380 310 59 : : PUNCT bracis-28380 310 60 pubmed pubmed PROPN bracis-28380 310 61     SPACE bracis-28380 310 62 google google PROPN bracis-28380 310 63 scholar scholar PROPN bracis-28380 310 64 carla carla PROPN bracis-28380 310 65 diniz diniz PROPN bracis-28380 310 66 lopes lopes PROPN bracis-28380 310 67 beckerview beckerview PROPN bracis-28380 310 68 author author NOUN bracis-28380 310 69 publications publication NOUN bracis-28380 310 70 search search NOUN bracis-28380 310 71 author author NOUN bracis-28380 310 72 on on ADP bracis-28380 310 73 : : PUNCT bracis-28380 310 74 pubmed pubmed PROPN bracis-28380 310 75     SPACE bracis-28380 310 76 google google PROPN bracis-28380 310 77 scholar scholar NOUN bracis-28380 310 78 viviane viviane PROPN bracis-28380 310 79 rodrigues rodrigue VERB bracis-28380 310 80 botelhoview botelhoview DET bracis-28380 310 81 author author NOUN bracis-28380 310 82 publications publication NOUN bracis-28380 310 83 search search NOUN bracis-28380 310 84 author author NOUN bracis-28380 310 85 on on ADP bracis-28380 310 86 : : PUNCT bracis-28380 310 87 pubmed pubmed PROPN bracis-28380 310 88     SPACE bracis-28380 310 89 google google PROPN bracis-28380 310 90 scholar scholar NOUN bracis-28380 310 91 thatiane thatiane NOUN bracis-28380 310 92 alves alves PROPN bracis-28380 310 93 pianoskiview pianoskiview PROPN bracis-28380 310 94 author author NOUN bracis-28380 310 95 publications publication NOUN bracis-28380 310 96 search search NOUN bracis-28380 310 97 author author NOUN bracis-28380 310 98 on on ADP bracis-28380 310 99 : : PUNCT bracis-28380 310 100 pubmed pubmed PROPN bracis-28380 310 101     SPACE bracis-28380 310 102 google google PROPN bracis-28380 310 103 scholar scholar NOUN bracis-28380 310 104 corresponding correspond VERB bracis-28380 310 105 author author NOUN bracis-28380 310 106 correspondence correspondence NOUN bracis-28380 310 107 to to ADP bracis-28380 310 108 dieine dieine PROPN bracis-28380 310 109 estela estela PROPN bracis-28380 310 110 bernieri bernieri PROPN bracis-28380 310 111 schiavon schiavon PROPN bracis-28380 310 112 . . PUNCT bracis-28380 311 1 editor editor NOUN bracis-28380 311 2 information information NOUN bracis-28380 311 3 editors editor NOUN bracis-28380 311 4 and and CCONJ bracis-28380 311 5 affiliations affiliation NOUN bracis-28380 311 6 federal federal PROPN bracis-28380 311 7 university university PROPN bracis-28380 311 8 of of ADP bracis-28380 311 9 são são PROPN bracis-28380 311 10 carlos carlos PROPN bracis-28380 311 11 , , PUNCT bracis-28380 311 12 são são PROPN bracis-28380 311 13 carlos carlos PROPN bracis-28380 311 14 , , PUNCT bracis-28380 311 15 brazil brazil PROPN bracis-28380 311 16 murilo murilo PROPN bracis-28380 311 17 c. c. PROPN bracis-28380 311 18 naldi naldi PROPN bracis-28380 311 19 centro centro PROPN bracis-28380 311 20 universitario universitario PROPN bracis-28380 311 21 da da PROPN bracis-28380 311 22 fei fei PROPN bracis-28380 311 23 , , PUNCT bracis-28380 311 24 são são PROPN bracis-28380 311 25 bernardo bernardo PROPN bracis-28380 311 26 do do AUX bracis-28380 311 27 campo campo PROPN bracis-28380 311 28 , , PUNCT bracis-28380 311 29 brazil brazil PROPN bracis-28380 311 30 reinaldo reinaldo PROPN bracis-28380 311 31 a. a. PROPN bracis-28380 311 32 c. c. PROPN bracis-28380 311 33 bianchi bianchi PROPN bracis-28380 311 34 rights right NOUN bracis-28380 311 35 and and CCONJ bracis-28380 311 36 permissions permission NOUN bracis-28380 311 37 reprints reprint NOUN bracis-28380 311 38 and and CCONJ bracis-28380 311 39 permissions permission VERB bracis-28380 311 40 copyright copyright NOUN bracis-28380 311 41 information information NOUN bracis-28380 311 42 © © ADP bracis-28380 311 43 2023 2023 NUM bracis-28380 311 44 the the DET bracis-28380 311 45 author(s author(s NOUN bracis-28380 311 46 ) ) PUNCT bracis-28380 311 47 , , PUNCT bracis-28380 311 48 under under ADP bracis-28380 311 49 exclusive exclusive ADJ bracis-28380 311 50 license license NOUN bracis-28380 311 51 to to ADP bracis-28380 311 52 springer springer NOUN bracis-28380 311 53 nature nature PROPN bracis-28380 311 54 switzerland switzerland PROPN bracis-28380 311 55 ag ag PROPN bracis-28380 311 56 about about ADP bracis-28380 311 57 this this DET bracis-28380 311 58 paper paper NOUN bracis-28380 311 59 cite cite VERB bracis-28380 311 60 this this DET bracis-28380 311 61 paper paper NOUN bracis-28380 311 62 schiavon schiavon ADJ bracis-28380 311 63 , , PUNCT bracis-28380 311 64 d.e.b d.e.b NOUN bracis-28380 311 65 . . PROPN bracis-28380 311 66 , , PUNCT bracis-28380 311 67 becker becker PROPN bracis-28380 311 68 , , PUNCT bracis-28380 311 69 c.d.l c.d.l PROPN bracis-28380 311 70 . . PUNCT bracis-28380 311 71 , , PUNCT bracis-28380 311 72 botelho botelho NOUN bracis-28380 311 73 , , PUNCT bracis-28380 311 74 v.r v.r PROPN bracis-28380 311 75 . . PROPN bracis-28380 311 76 , , PUNCT bracis-28380 311 77 pianoski pianoski PROPN bracis-28380 311 78 , , PUNCT bracis-28380 311 79 t.a t.a PROPN bracis-28380 311 80 . . PROPN bracis-28380 311 81 ( ( PUNCT bracis-28380 311 82 2023 2023 NUM bracis-28380 311 83 ) ) PUNCT bracis-28380 311 84 . . PUNCT bracis-28380 312 1 interpreting interpret VERB bracis-28380 312 2 convolutional convolutional ADJ bracis-28380 312 3 neural neural ADJ bracis-28380 312 4 networks network NOUN bracis-28380 312 5 for for ADP bracis-28380 312 6     SPACE bracis-28380 312 7 brain brain NOUN bracis-28380 312 8 tumor tumor NOUN bracis-28380 312 9 classification classification NOUN bracis-28380 312 10 : : PUNCT bracis-28380 312 11 an an DET bracis-28380 312 12 explainable explainable ADJ bracis-28380 312 13 artificial artificial ADJ bracis-28380 312 14 intelligence intelligence NOUN bracis-28380 312 15 approach approach NOUN bracis-28380 312 16 . . PUNCT bracis-28380 313 1 in in ADP bracis-28380 313 2 : : PUNCT bracis-28380 313 3 naldi naldi PROPN bracis-28380 313 4 , , PUNCT bracis-28380 313 5 m.c m.c PROPN bracis-28380 313 6 . . PROPN bracis-28380 313 7 , , PUNCT bracis-28380 313 8 bianchi bianchi PROPN bracis-28380 313 9 , , PUNCT bracis-28380 313 10 r.a.c r.a.c ADP bracis-28380 313 11 . . PUNCT bracis-28380 313 12 ( ( PUNCT bracis-28380 313 13 eds ed NOUN bracis-28380 313 14 ) ) PUNCT bracis-28380 313 15 intelligent intelligent ADJ bracis-28380 313 16 systems system NOUN bracis-28380 313 17 . . PUNCT bracis-28380 314 1 bracis bracis PROPN bracis-28380 314 2 2023 2023 NUM bracis-28380 314 3 . . PUNCT bracis-28380 315 1 lecture lecture NOUN bracis-28380 315 2 notes note NOUN bracis-28380 315 3 in in ADP bracis-28380 315 4 computer computer NOUN bracis-28380 315 5 science science NOUN bracis-28380 315 6 ( ( PUNCT bracis-28380 315 7 ) ) PUNCT bracis-28380 315 8 , , PUNCT bracis-28380 315 9 vol vol NOUN bracis-28380 315 10 14196 14196 NUM bracis-28380 315 11 . . PUNCT bracis-28380 316 1 springer springer NOUN bracis-28380 316 2 , , PUNCT bracis-28380 316 3 cham cham PROPN bracis-28380 316 4 . . PUNCT bracis-28380 317 1 https://doi.org/10.1007/978-3-031-45389-2_6 https://doi.org/10.1007/978-3-031-45389-2_6 X bracis-28380 318 1 download download PROPN bracis-28380 318 2 citation citation NOUN bracis-28380 318 3 .ris .ris PUNCT bracis-28380 319 1 .enw .enw PROPN bracis-28380 319 2 .bib .bib PUNCT bracis-28380 320 1 doi doi PROPN bracis-28380 320 2 : : PUNCT bracis-28380 320 3 https://doi.org/10.1007/978-3-031-45389-2_6 https://doi.org/10.1007/978-3-031-45389-2_6 X bracis-28380 320 4 published publish VERB bracis-28380 320 5 : : PUNCT bracis-28380 320 6 12 12 NUM bracis-28380 320 7 october october PROPN bracis-28380 320 8 2023 2023 NUM bracis-28380 320 9 publisher publisher NOUN bracis-28380 320 10 name name NOUN bracis-28380 320 11 : : PUNCT bracis-28380 320 12 springer springer NOUN bracis-28380 320 13 , , PUNCT bracis-28380 320 14 cham cham PROPN bracis-28380 320 15 print print PROPN bracis-28380 320 16 isbn isbn PROPN bracis-28380 320 17 : : PUNCT bracis-28380 320 18 978 978 NUM bracis-28380 320 19 - - SYM bracis-28380 320 20 3 3 NUM bracis-28380 320 21 - - PUNCT bracis-28380 320 22 031 031 NUM bracis-28380 320 23 - - PUNCT bracis-28380 320 24 45388 45388 NUM bracis-28380 320 25 - - SYM bracis-28380 320 26 5 5 NUM bracis-28380 320 27 online online ADJ bracis-28380 320 28 isbn isbn NOUN bracis-28380 320 29 : : PUNCT bracis-28380 320 30 978 978 NUM bracis-28380 320 31 - - SYM bracis-28380 320 32 3 3 NUM bracis-28380 320 33 - - PUNCT bracis-28380 320 34 031 031 NUM bracis-28380 320 35 - - PUNCT bracis-28380 320 36 45389 45389 NUM bracis-28380 320 37 - - SYM bracis-28380 320 38 2 2 NUM bracis-28380 320 39 ebook ebook NOUN bracis-28380 320 40 packages package NOUN bracis-28380 320 41 : : PUNCT bracis-28380 320 42 computer computer NOUN bracis-28380 320 43 sciencecomputer sciencecomputer NOUN bracis-28380 320 44 science science NOUN bracis-28380 320 45 ( ( PUNCT bracis-28380 320 46 r0 r0 NOUN bracis-28380 320 47 ) ) PUNCT bracis-28380 320 48 share share VERB bracis-28380 320 49 this this DET bracis-28380 320 50 paper paper NOUN bracis-28380 320 51 anyone anyone PRON bracis-28380 320 52 you you PRON bracis-28380 320 53 share share VERB bracis-28380 320 54 the the DET bracis-28380 320 55 following follow VERB bracis-28380 320 56 link link NOUN bracis-28380 320 57 with with ADP bracis-28380 320 58 will will AUX bracis-28380 320 59 be be AUX bracis-28380 320 60 able able ADJ bracis-28380 320 61 to to PART bracis-28380 320 62 read read VERB bracis-28380 320 63 this this DET bracis-28380 320 64 content content NOUN bracis-28380 320 65 : : PUNCT bracis-28380 320 66 get get VERB bracis-28380 320 67 shareable shareable ADJ bracis-28380 320 68 linksorry linksorry NOUN bracis-28380 320 69 , , PUNCT bracis-28380 320 70 a a DET bracis-28380 320 71 shareable shareable ADJ bracis-28380 320 72 link link NOUN bracis-28380 320 73 is be AUX bracis-28380 320 74 not not PART bracis-28380 320 75 currently currently ADV bracis-28380 320 76 available available ADJ bracis-28380 320 77 for for ADP bracis-28380 320 78 this this DET bracis-28380 320 79 article article NOUN bracis-28380 320 80 . . PUNCT bracis-28380 321 1 copy copy VERB bracis-28380 321 2 shareable shareable ADJ bracis-28380 321 3 link link NOUN bracis-28380 321 4 to to PART bracis-28380 321 5 clipboard clipboard NOUN bracis-28380 321 6 provided provide VERB bracis-28380 321 7 by by ADP bracis-28380 321 8 the the DET bracis-28380 321 9 springer springer NOUN bracis-28380 321 10 nature nature PROPN bracis-28380 321 11 sharedit sharedit PROPN bracis-28380 321 12 content content NOUN bracis-28380 321 13 - - PUNCT bracis-28380 321 14 sharing share VERB bracis-28380 321 15 initiative initiative NOUN bracis-28380 321 16 keywords keyword NOUN bracis-28380 321 17 brain brain VERB bracis-28380 321 18 tumor tumor NOUN bracis-28380 321 19 convolutional convolutional ADJ bracis-28380 321 20 neural neural ADJ bracis-28380 321 21 networks network NOUN bracis-28380 321 22 xai xai PROPN bracis-28380 321 23 techniques technique NOUN bracis-28380 321 24 publish publish VERB bracis-28380 321 25 with with ADP bracis-28380 321 26 us us PROPN bracis-28380 321 27 policies policy NOUN bracis-28380 321 28 and and CCONJ bracis-28380 321 29 ethics ethic NOUN bracis-28380 321 30 profiles profile NOUN bracis-28380 321 31 dieine dieine PROPN bracis-28380 321 32 estela estela PROPN bracis-28380 321 33 bernieri bernieri VERB bracis-28380 321 34 schiavon schiavon ADJ bracis-28380 321 35 view view NOUN bracis-28380 321 36 author author NOUN bracis-28380 321 37 profile profile NOUN bracis-28380 321 38 search search NOUN bracis-28380 321 39 search search NOUN bracis-28380 321 40 by by ADP bracis-28380 321 41 keyword keyword NOUN bracis-28380 321 42 or or CCONJ bracis-28380 321 43 author author NOUN bracis-28380 321 44 search search NOUN bracis-28380 321 45 navigation navigation NOUN bracis-28380 321 46 find find VERB bracis-28380 321 47 a a DET bracis-28380 321 48 journal journal NOUN bracis-28380 321 49 publish publish VERB bracis-28380 321 50 with with ADP bracis-28380 321 51 us we PRON bracis-28380 321 52 track track VERB bracis-28380 321 53 your your PRON bracis-28380 321 54 research research NOUN bracis-28380 321 55 discover discover VERB bracis-28380 321 56 content content NOUN bracis-28380 321 57 journals journal NOUN bracis-28380 321 58 a a DET bracis-28380 321 59 - - PUNCT bracis-28380 321 60 z z NOUN bracis-28380 321 61 books book NOUN bracis-28380 321 62 a a DET bracis-28380 321 63 - - PUNCT bracis-28380 321 64 z z NOUN bracis-28380 321 65 publish publish NOUN bracis-28380 321 66 with with ADP bracis-28380 321 67 us us PROPN bracis-28380 321 68 journal journal PROPN bracis-28380 321 69 finder finder PROPN bracis-28380 321 70 publish publish VERB bracis-28380 321 71 your your PRON bracis-28380 321 72 research research NOUN bracis-28380 321 73 language language NOUN bracis-28380 321 74 editing edit VERB bracis-28380 321 75 open open ADJ bracis-28380 321 76 access access NOUN bracis-28380 321 77 publishing publishing NOUN bracis-28380 321 78 products product NOUN bracis-28380 321 79 and and CCONJ bracis-28380 321 80 services service NOUN bracis-28380 321 81 our our PRON bracis-28380 321 82 products product NOUN bracis-28380 321 83 librarians librarian VERB bracis-28380 321 84 societies society NOUN bracis-28380 321 85 partners partner NOUN bracis-28380 321 86 and and CCONJ bracis-28380 321 87 advertisers advertiser NOUN bracis-28380 321 88 our our PRON bracis-28380 321 89 brands brand NOUN bracis-28380 321 90 springer springer NOUN bracis-28380 321 91 nature nature PROPN bracis-28380 321 92 portfolio portfolio PROPN bracis-28380 321 93 bmc bmc PROPN bracis-28380 321 94 palgrave palgrave PROPN bracis-28380 321 95 macmillan macmillan PROPN bracis-28380 321 96 apress apress PROPN bracis-28380 321 97 discover discover VERB bracis-28380 321 98 your your PRON bracis-28380 321 99 privacy privacy NOUN bracis-28380 321 100 choices choice NOUN bracis-28380 321 101 / / SYM bracis-28380 321 102 manage manage NOUN bracis-28380 321 103 cookies cookie NOUN bracis-28380 321 104 your your PRON bracis-28380 321 105 us us PROPN bracis-28380 322 1 state state NOUN bracis-28380 322 2 privacy privacy NOUN bracis-28380 322 3 rights right NOUN bracis-28380 322 4 accessibility accessibility NOUN bracis-28380 322 5 statement statement NOUN bracis-28380 322 6 terms term NOUN bracis-28380 322 7 and and CCONJ bracis-28380 322 8 conditions condition NOUN bracis-28380 322 9 privacy privacy NOUN bracis-28380 322 10 policy policy NOUN bracis-28380 322 11 help help NOUN bracis-28380 322 12 and and CCONJ bracis-28380 322 13 support support VERB bracis-28380 322 14 legal legal ADJ bracis-28380 322 15 notice notice NOUN bracis-28380 322 16 cancel cancel VERB bracis-28380 322 17 contracts contract NOUN bracis-28380 322 18 here here ADV bracis-28380 322 19 not not PART bracis-28380 322 20 affiliated affiliate VERB bracis-28380 322 21 © © PROPN bracis-28380 322 22 2025 2025 NUM bracis-28380 322 23 springer springer NOUN bracis-28380 322 24 nature nature NOUN