id	sid	tid	token	lemma	pos
fcis-20320	1	1	frontiers	frontier	NOUN
fcis-20320	1	2	in	in	ADP
fcis-20320	1	3	computing	computing	NOUN
fcis-20320	1	4	and	and	CCONJ
fcis-20320	1	5	intelligent	intelligent	ADJ
fcis-20320	1	6	systems	system	NOUN
fcis-20320	1	7	issn	issn	VERB
fcis-20320	1	8	:	:	PUNCT
fcis-20320	1	9	2832	2832	NUM
fcis-20320	1	10	-	-	SYM
fcis-20320	1	11	6024	6024	NUM
fcis-20320	1	12	|	|	NOUN
fcis-20320	1	13	vol	vol	NOUN
fcis-20320	1	14	.	.	PROPN
fcis-20320	2	1	8	8	NUM
fcis-20320	2	2	,	,	PUNCT
fcis-20320	2	3	no	no	INTJ
fcis-20320	2	4	.	.	NOUN
fcis-20320	2	5	1	1	NUM
fcis-20320	2	6	,	,	PUNCT
fcis-20320	2	7	2024	2024	NUM
fcis-20320	2	8	6	6	NUM
fcis-20320	2	9	a	a	DET
fcis-20320	2	10	hybrid	hybrid	ADJ
fcis-20320	2	11	deep	deep	ADJ
fcis-20320	2	12	learning	learning	NOUN
fcis-20320	2	13	approach	approach	NOUN
fcis-20320	2	14	for	for	ADP
fcis-20320	2	15	lung	lung	NOUN
fcis-20320	2	16	nodule	nodule	NOUN
fcis-20320	2	17	classification	classification	NOUN
fcis-20320	2	18	cheng	cheng	PROPN
fcis-20320	2	19	ren	ren	PROPN
fcis-20320	2	20	and	and	CCONJ
fcis-20320	2	21	shouming	shoume	VERB
fcis-20320	2	22	hou	hou	PROPN
fcis-20320	2	23	school	school	NOUN
fcis-20320	2	24	of	of	ADP
fcis-20320	2	25	computer	computer	NOUN
fcis-20320	2	26	science	science	NOUN
fcis-20320	2	27	and	and	CCONJ
fcis-20320	2	28	technology	technology	NOUN
fcis-20320	2	29	,	,	PUNCT
fcis-20320	2	30	henan	henan	PROPN
fcis-20320	2	31	polytechnic	polytechnic	PROPN
fcis-20320	2	32	university	university	PROPN
fcis-20320	2	33	,	,	PUNCT
fcis-20320	2	34	jiaozuo	jiaozuo	PROPN
fcis-20320	2	35	henan	henan	PROPN
fcis-20320	2	36	,	,	PUNCT
fcis-20320	2	37	454003	454003	NUM
fcis-20320	2	38	,	,	PUNCT
fcis-20320	2	39	china	china	PROPN
fcis-20320	2	40	abstract	abstract	NOUN
fcis-20320	2	41	:	:	PUNCT
fcis-20320	2	42	lung	lung	NOUN
fcis-20320	2	43	cancer	cancer	NOUN
fcis-20320	2	44	has	have	VERB
fcis-20320	2	45	the	the	DET
fcis-20320	2	46	highest	high	ADJ
fcis-20320	2	47	morbidity	morbidity	NOUN
fcis-20320	2	48	and	and	CCONJ
fcis-20320	2	49	mortality	mortality	NOUN
fcis-20320	2	50	rates	rate	NOUN
fcis-20320	2	51	worldwide	worldwide	ADV
fcis-20320	2	52	.	.	PUNCT
fcis-20320	3	1	pulmonary	pulmonary	ADJ
fcis-20320	3	2	nodules	nodule	NOUN
fcis-20320	3	3	are	be	AUX
fcis-20320	3	4	an	an	DET
fcis-20320	3	5	early	early	ADJ
fcis-20320	3	6	manifestation	manifestation	NOUN
fcis-20320	3	7	of	of	ADP
fcis-20320	3	8	lung	lung	NOUN
fcis-20320	3	9	cancer	cancer	NOUN
fcis-20320	3	10	.	.	PUNCT
fcis-20320	4	1	therefore	therefore	ADV
fcis-20320	4	2	,	,	PUNCT
fcis-20320	4	3	accurate	accurate	ADJ
fcis-20320	4	4	classification	classification	NOUN
fcis-20320	4	5	of	of	ADP
fcis-20320	4	6	pulmonary	pulmonary	ADJ
fcis-20320	4	7	nodules	nodule	NOUN
fcis-20320	4	8	is	be	AUX
fcis-20320	4	9	of	of	ADP
fcis-20320	4	10	great	great	ADJ
fcis-20320	4	11	significance	significance	NOUN
fcis-20320	4	12	for	for	ADP
fcis-20320	4	13	the	the	DET
fcis-20320	4	14	early	early	ADJ
fcis-20320	4	15	diagnosis	diagnosis	NOUN
fcis-20320	4	16	and	and	CCONJ
fcis-20320	4	17	treatment	treatment	NOUN
fcis-20320	4	18	of	of	ADP
fcis-20320	4	19	lung	lung	NOUN
fcis-20320	4	20	cancer	cancer	NOUN
fcis-20320	4	21	.	.	PUNCT
fcis-20320	5	1	however	however	ADV
fcis-20320	5	2	,	,	PUNCT
fcis-20320	5	3	the	the	DET
fcis-20320	5	4	classification	classification	NOUN
fcis-20320	5	5	of	of	ADP
fcis-20320	5	6	lung	lung	NOUN
fcis-20320	5	7	nodules	nodule	NOUN
fcis-20320	5	8	is	be	AUX
fcis-20320	5	9	a	a	DET
fcis-20320	5	10	complex	complex	ADJ
fcis-20320	5	11	and	and	CCONJ
fcis-20320	5	12	time	time	NOUN
fcis-20320	5	13	-	-	PUNCT
fcis-20320	5	14	consuming	consume	VERB
fcis-20320	5	15	task	task	NOUN
fcis-20320	5	16	requiring	require	VERB
fcis-20320	5	17	extensive	extensive	ADJ
fcis-20320	5	18	image	image	NOUN
fcis-20320	5	19	reading	reading	NOUN
fcis-20320	5	20	and	and	CCONJ
fcis-20320	5	21	analysis	analysis	NOUN
fcis-20320	5	22	by	by	ADP
fcis-20320	5	23	expert	expert	NOUN
fcis-20320	5	24	radiologists	radiologist	NOUN
fcis-20320	5	25	.	.	PUNCT
fcis-20320	6	1	therefore	therefore	ADV
fcis-20320	6	2	,	,	PUNCT
fcis-20320	6	3	using	use	VERB
fcis-20320	6	4	deep	deep	ADJ
fcis-20320	6	5	learning	learning	NOUN
fcis-20320	6	6	technology	technology	NOUN
fcis-20320	6	7	to	to	PART
fcis-20320	6	8	assist	assist	VERB
fcis-20320	6	9	doctors	doctor	NOUN
fcis-20320	6	10	in	in	ADP
fcis-20320	6	11	detecting	detect	VERB
fcis-20320	6	12	and	and	CCONJ
fcis-20320	6	13	classifying	classify	VERB
fcis-20320	6	14	pulmonary	pulmonary	ADJ
fcis-20320	6	15	nodules	nodule	NOUN
fcis-20320	6	16	has	have	AUX
fcis-20320	6	17	become	become	VERB
fcis-20320	6	18	a	a	DET
fcis-20320	6	19	current	current	ADJ
fcis-20320	6	20	research	research	NOUN
fcis-20320	6	21	trend	trend	NOUN
fcis-20320	6	22	.	.	PUNCT
fcis-20320	7	1	a	a	DET
fcis-20320	7	2	lightweight	lightweight	ADJ
fcis-20320	7	3	classification	classification	NOUN
fcis-20320	7	4	model	model	NOUN
fcis-20320	7	5	named	name	VERB
fcis-20320	7	6	res	re	NOUN
fcis-20320	7	7	-	-	PUNCT
fcis-20320	7	8	vgg	vgg	NOUN
fcis-20320	7	9	is	be	AUX
fcis-20320	7	10	proposed	propose	VERB
fcis-20320	7	11	for	for	ADP
fcis-20320	7	12	classifying	classify	VERB
fcis-20320	7	13	lung	lung	NOUN
fcis-20320	7	14	nodules	nodule	NOUN
fcis-20320	7	15	as	as	ADP
fcis-20320	7	16	benign	benign	ADJ
fcis-20320	7	17	or	or	CCONJ
fcis-20320	7	18	malignant	malignant	ADJ
fcis-20320	7	19	.	.	PUNCT
fcis-20320	8	1	the	the	DET
fcis-20320	8	2	res	re	NOUN
fcis-20320	8	3	-	-	PUNCT
fcis-20320	8	4	vgg	vgg	NOUN
fcis-20320	8	5	model	model	NOUN
fcis-20320	8	6	improves	improve	VERB
fcis-20320	8	7	on	on	ADP
fcis-20320	8	8	vgg16	vgg16	NOUN
fcis-20320	8	9	by	by	ADP
fcis-20320	8	10	reducing	reduce	VERB
fcis-20320	8	11	the	the	DET
fcis-20320	8	12	use	use	NOUN
fcis-20320	8	13	of	of	ADP
fcis-20320	8	14	convolutional	convolutional	ADJ
fcis-20320	8	15	and	and	CCONJ
fcis-20320	8	16	fully	fully	ADV
fcis-20320	8	17	connected	connected	ADJ
fcis-20320	8	18	layers	layer	NOUN
fcis-20320	8	19	.	.	PUNCT
fcis-20320	9	1	to	to	PART
fcis-20320	9	2	reduce	reduce	VERB
fcis-20320	9	3	overfitting	overfitte	VERB
fcis-20320	9	4	,	,	PUNCT
fcis-20320	9	5	residual	residual	ADJ
fcis-20320	9	6	connections	connection	NOUN
fcis-20320	9	7	are	be	AUX
fcis-20320	9	8	introduced	introduce	VERB
fcis-20320	9	9	.	.	PUNCT
fcis-20320	10	1	the	the	DET
fcis-20320	10	2	training	training	NOUN
fcis-20320	10	3	of	of	ADP
fcis-20320	10	4	the	the	DET
fcis-20320	10	5	model	model	NOUN
fcis-20320	10	6	was	be	AUX
fcis-20320	10	7	performed	perform	VERB
fcis-20320	10	8	on	on	ADP
fcis-20320	10	9	the	the	DET
fcis-20320	10	10	luna16	luna16	ADJ
fcis-20320	10	11	database	database	NOUN
fcis-20320	10	12	,	,	PUNCT
fcis-20320	10	13	and	and	CCONJ
fcis-20320	10	14	a	a	DET
fcis-20320	10	15	ten	ten	ADJ
fcis-20320	10	16	-	-	ADJ
fcis-20320	10	17	fold	fold	ADJ
fcis-20320	10	18	cross	cross	ADJ
fcis-20320	10	19	-	-	ADJ
fcis-20320	10	20	validation	validation	ADJ
fcis-20320	10	21	method	method	NOUN
fcis-20320	10	22	was	be	AUX
fcis-20320	10	23	used	use	VERB
fcis-20320	10	24	to	to	PART
fcis-20320	10	25	evaluate	evaluate	VERB
fcis-20320	10	26	the	the	DET
fcis-20320	10	27	performance	performance	NOUN
fcis-20320	10	28	of	of	ADP
fcis-20320	10	29	the	the	DET
fcis-20320	10	30	model	model	NOUN
fcis-20320	10	31	.	.	PUNCT
fcis-20320	11	1	in	in	ADP
fcis-20320	11	2	addition	addition	NOUN
fcis-20320	11	3	,	,	PUNCT
fcis-20320	11	4	the	the	DET
fcis-20320	11	5	res	re	NOUN
fcis-20320	11	6	-	-	PUNCT
fcis-20320	11	7	vgg	vgg	NOUN
fcis-20320	11	8	model	model	NOUN
fcis-20320	11	9	was	be	AUX
fcis-20320	11	10	compared	compare	VERB
fcis-20320	11	11	with	with	ADP
fcis-20320	11	12	three	three	NUM
fcis-20320	11	13	other	other	ADJ
fcis-20320	11	14	common	common	ADJ
fcis-20320	11	15	classification	classification	NOUN
fcis-20320	11	16	networks	network	NOUN
fcis-20320	11	17	,	,	PUNCT
fcis-20320	11	18	and	and	CCONJ
fcis-20320	11	19	the	the	DET
fcis-20320	11	20	results	result	NOUN
fcis-20320	11	21	showed	show	VERB
fcis-20320	11	22	that	that	SCONJ
fcis-20320	11	23	the	the	DET
fcis-20320	11	24	res	re	NOUN
fcis-20320	11	25	-	-	PUNCT
fcis-20320	11	26	vgg	vgg	NOUN
fcis-20320	11	27	model	model	NOUN
fcis-20320	11	28	outperformed	outperform	VERB
fcis-20320	11	29	the	the	DET
fcis-20320	11	30	other	other	ADJ
fcis-20320	11	31	models	model	NOUN
fcis-20320	11	32	in	in	ADP
fcis-20320	11	33	terms	term	NOUN
fcis-20320	11	34	of	of	ADP
fcis-20320	11	35	accuracy	accuracy	NOUN
fcis-20320	11	36	,	,	PUNCT
fcis-20320	11	37	sensitivity	sensitivity	NOUN
fcis-20320	11	38	,	,	PUNCT
fcis-20320	11	39	and	and	CCONJ
fcis-20320	11	40	specificity	specificity	NOUN
fcis-20320	11	41	.	.	PUNCT
fcis-20320	12	1	keywords	keyword	NOUN
fcis-20320	12	2	:	:	PUNCT
fcis-20320	12	3	lung	lung	NOUN
fcis-20320	12	4	nodule	nodule	NOUN
fcis-20320	12	5	benign	benign	ADJ
fcis-20320	12	6	and	and	CCONJ
fcis-20320	12	7	malignant	malignant	ADJ
fcis-20320	12	8	classification	classification	NOUN
fcis-20320	12	9	;	;	PUNCT
fcis-20320	12	10	deep	deep	ADJ
fcis-20320	12	11	learning	learning	NOUN
fcis-20320	12	12	;	;	PUNCT
fcis-20320	12	13	resnet	resnet	NOUN
fcis-20320	12	14	;	;	PUNCT
fcis-20320	12	15	convolutional	convolutional	ADJ
fcis-20320	12	16	neural	neural	ADJ
fcis-20320	12	17	network	network	NOUN
fcis-20320	12	18	;	;	PUNCT
fcis-20320	12	19	vggnet	vggnet	NOUN
fcis-20320	12	20	.	.	PUNCT
fcis-20320	13	1	1	1	X
fcis-20320	13	2	.	.	X
fcis-20320	13	3	introduction	introduction	NOUN
fcis-20320	13	4	in	in	ADP
fcis-20320	13	5	the	the	DET
fcis-20320	13	6	medical	medical	ADJ
fcis-20320	13	7	field	field	NOUN
fcis-20320	13	8	,	,	PUNCT
fcis-20320	13	9	the	the	DET
fcis-20320	13	10	detection	detection	NOUN
fcis-20320	13	11	and	and	CCONJ
fcis-20320	13	12	classification	classification	NOUN
fcis-20320	13	13	of	of	ADP
fcis-20320	13	14	pulmonary	pulmonary	ADJ
fcis-20320	13	15	nodules	nodule	NOUN
fcis-20320	13	16	is	be	AUX
fcis-20320	13	17	an	an	DET
fcis-20320	13	18	important	important	ADJ
fcis-20320	13	19	and	and	CCONJ
fcis-20320	13	20	challenging	challenging	ADJ
fcis-20320	13	21	task	task	NOUN
fcis-20320	13	22	.	.	PUNCT
fcis-20320	14	1	pulmonary	pulmonary	ADJ
fcis-20320	14	2	nodules	nodule	NOUN
fcis-20320	14	3	are	be	AUX
fcis-20320	14	4	small	small	ADJ
fcis-20320	14	5	pieces	piece	NOUN
fcis-20320	14	6	of	of	ADP
fcis-20320	14	7	lung	lung	NOUN
fcis-20320	14	8	tissue	tissue	NOUN
fcis-20320	14	9	that	that	PRON
fcis-20320	14	10	appear	appear	VERB
fcis-20320	14	11	on	on	ADP
fcis-20320	14	12	lung	lung	PROPN
fcis-20320	14	13	computed	compute	VERB
fcis-20320	14	14	tomography	tomography	NOUN
fcis-20320	14	15	(	(	PUNCT
fcis-20320	14	16	ct	ct	NOUN
fcis-20320	14	17	)	)	PUNCT
fcis-20320	14	18	scans	scan	NOUN
fcis-20320	14	19	,	,	PUNCT
fcis-20320	14	20	and	and	CCONJ
fcis-20320	14	21	they	they	PRON
fcis-20320	14	22	may	may	AUX
fcis-20320	14	23	be	be	AUX
fcis-20320	14	24	benign	benign	ADJ
fcis-20320	14	25	or	or	CCONJ
fcis-20320	14	26	malignant	malignant	ADJ
fcis-20320	14	27	[	[	X
fcis-20320	14	28	1	1	NUM
fcis-20320	14	29	]	]	PUNCT
fcis-20320	14	30	.	.	PUNCT
fcis-20320	15	1	most	most	ADJ
fcis-20320	15	2	lung	lung	NOUN
fcis-20320	15	3	cancers	cancer	NOUN
fcis-20320	15	4	originate	originate	VERB
fcis-20320	15	5	from	from	ADP
fcis-20320	15	6	small	small	ADJ
fcis-20320	15	7	malignant	malignant	ADJ
fcis-20320	15	8	nodules	nodule	NOUN
fcis-20320	15	9	.	.	PUNCT
fcis-20320	16	1	however	however	ADV
fcis-20320	16	2	,	,	PUNCT
fcis-20320	16	3	automatic	automatic	ADJ
fcis-20320	16	4	identification	identification	NOUN
fcis-20320	16	5	of	of	ADP
fcis-20320	16	6	benign	benign	ADJ
fcis-20320	16	7	and	and	CCONJ
fcis-20320	16	8	malignant	malignant	ADJ
fcis-20320	16	9	pulmonary	pulmonary	ADJ
fcis-20320	16	10	nodules	nodule	NOUN
fcis-20320	16	11	in	in	ADP
fcis-20320	16	12	ct	ct	NUM
fcis-20320	16	13	images	image	NOUN
fcis-20320	16	14	becomes	become	VERB
fcis-20320	16	15	a	a	DET
fcis-20320	16	16	challenge	challenge	NOUN
fcis-20320	16	17	due	due	ADP
fcis-20320	16	18	to	to	ADP
fcis-20320	16	19	the	the	DET
fcis-20320	16	20	wide	wide	ADJ
fcis-20320	16	21	range	range	NOUN
fcis-20320	16	22	of	of	ADP
fcis-20320	16	23	shape	shape	NOUN
fcis-20320	16	24	and	and	CCONJ
fcis-20320	16	25	texture	texture	ADJ
fcis-20320	16	26	variations	variation	NOUN
fcis-20320	16	27	of	of	ADP
fcis-20320	16	28	pulmonary	pulmonary	ADJ
fcis-20320	16	29	nodules	nodule	NOUN
fcis-20320	16	30	,	,	PUNCT
fcis-20320	16	31	as	as	ADV
fcis-20320	16	32	well	well	ADV
fcis-20320	16	33	as	as	ADP
fcis-20320	16	34	the	the	DET
fcis-20320	16	35	visual	visual	ADJ
fcis-20320	16	36	similarity	similarity	NOUN
fcis-20320	16	37	between	between	ADP
fcis-20320	16	38	benign	benign	ADJ
fcis-20320	16	39	and	and	CCONJ
fcis-20320	16	40	malignant	malignant	ADJ
fcis-20320	16	41	pulmonary	pulmonary	ADJ
fcis-20320	16	42	nodules	nodule	NOUN
fcis-20320	16	43	[	[	X
fcis-20320	16	44	2	2	NUM
fcis-20320	16	45	]	]	PUNCT
fcis-20320	16	46	.	.	PUNCT
fcis-20320	17	1	deep	deep	ADJ
fcis-20320	17	2	learning	learning	NOUN
fcis-20320	17	3	,	,	PUNCT
fcis-20320	17	4	as	as	ADP
fcis-20320	17	5	an	an	DET
fcis-20320	17	6	important	important	ADJ
fcis-20320	17	7	branch	branch	NOUN
fcis-20320	17	8	of	of	ADP
fcis-20320	17	9	machine	machine	NOUN
fcis-20320	17	10	learning	learning	NOUN
fcis-20320	17	11	,	,	PUNCT
fcis-20320	17	12	has	have	AUX
fcis-20320	17	13	achieved	achieve	VERB
fcis-20320	17	14	remarkable	remarkable	ADJ
fcis-20320	17	15	success	success	NOUN
fcis-20320	17	16	in	in	ADP
fcis-20320	17	17	various	various	ADJ
fcis-20320	17	18	fields	field	NOUN
fcis-20320	17	19	,	,	PUNCT
fcis-20320	17	20	including	include	VERB
fcis-20320	17	21	natural	natural	ADJ
fcis-20320	17	22	language	language	NOUN
fcis-20320	17	23	processing	processing	NOUN
fcis-20320	17	24	,	,	PUNCT
fcis-20320	17	25	speech	speech	NOUN
fcis-20320	17	26	recognition	recognition	NOUN
fcis-20320	17	27	,	,	PUNCT
fcis-20320	17	28	computer	computer	NOUN
fcis-20320	17	29	vision	vision	NOUN
fcis-20320	17	30	,	,	PUNCT
fcis-20320	17	31	etc	etc	X
fcis-20320	17	32	.	.	X
fcis-20320	17	33	deep	deep	ADJ
fcis-20320	17	34	learning	learning	NOUN
fcis-20320	17	35	models	model	NOUN
fcis-20320	17	36	,	,	PUNCT
fcis-20320	17	37	such	such	ADJ
fcis-20320	17	38	as	as	ADP
fcis-20320	17	39	convolutional	convolutional	ADJ
fcis-20320	17	40	neural	neural	ADJ
fcis-20320	17	41	networks	network	NOUN
fcis-20320	17	42	(	(	PUNCT
fcis-20320	17	43	cnn	cnn	PROPN
fcis-20320	17	44	)	)	PUNCT
fcis-20320	17	45	,	,	PUNCT
fcis-20320	17	46	have	have	AUX
fcis-20320	17	47	been	be	AUX
fcis-20320	17	48	widely	widely	ADV
fcis-20320	17	49	used	use	VERB
fcis-20320	17	50	in	in	ADP
fcis-20320	17	51	medical	medical	ADJ
fcis-20320	17	52	image	image	NOUN
fcis-20320	17	53	analysis	analysis	NOUN
fcis-20320	17	54	,	,	PUNCT
fcis-20320	17	55	including	include	VERB
fcis-20320	17	56	the	the	DET
fcis-20320	17	57	detection	detection	NOUN
fcis-20320	17	58	and	and	CCONJ
fcis-20320	17	59	classification	classification	NOUN
fcis-20320	17	60	of	of	ADP
fcis-20320	17	61	pulmonary	pulmonary	ADJ
fcis-20320	17	62	nodules	nodule	NOUN
fcis-20320	17	63	.	.	PUNCT
fcis-20320	18	1	however	however	ADV
fcis-20320	18	2	,	,	PUNCT
fcis-20320	18	3	although	although	SCONJ
fcis-20320	18	4	deep	deep	ADJ
fcis-20320	18	5	learning	learning	NOUN
fcis-20320	18	6	has	have	AUX
fcis-20320	18	7	made	make	VERB
fcis-20320	18	8	significant	significant	ADJ
fcis-20320	18	9	progress	progress	NOUN
fcis-20320	18	10	in	in	ADP
fcis-20320	18	11	these	these	DET
fcis-20320	18	12	fields	field	NOUN
fcis-20320	18	13	,	,	PUNCT
fcis-20320	18	14	some	some	DET
fcis-20320	18	15	challenges	challenge	NOUN
fcis-20320	18	16	still	still	ADV
fcis-20320	18	17	exist	exist	VERB
fcis-20320	18	18	.	.	PUNCT
fcis-20320	19	1	for	for	ADP
fcis-20320	19	2	example	example	NOUN
fcis-20320	19	3	,	,	PUNCT
fcis-20320	19	4	some	some	DET
fcis-20320	19	5	deep	deep	ADJ
fcis-20320	19	6	learning	learning	NOUN
fcis-20320	19	7	models	model	NOUN
fcis-20320	19	8	,	,	PUNCT
fcis-20320	19	9	such	such	ADJ
fcis-20320	19	10	as	as	ADP
fcis-20320	19	11	alexnet	alexnet	ADJ
fcis-20320	19	12	,	,	PUNCT
fcis-20320	19	13	vggnet	vggnet	PROPN
fcis-20320	19	14	,	,	PUNCT
fcis-20320	19	15	microsoft	microsoft	PROPN
fcis-20320	19	16	resnet	resnet	PROPN
fcis-20320	19	17	,	,	PUNCT
fcis-20320	19	18	densenets	densenet	NOUN
fcis-20320	19	19	,	,	PUNCT
fcis-20320	19	20	etc	etc	X
fcis-20320	19	21	.	.	X
fcis-20320	19	22	,	,	PUNCT
fcis-20320	19	23	may	may	AUX
fcis-20320	19	24	face	face	VERB
fcis-20320	19	25	the	the	DET
fcis-20320	19	26	problem	problem	NOUN
fcis-20320	19	27	of	of	ADP
fcis-20320	19	28	decreased	decrease	VERB
fcis-20320	19	29	accuracy	accuracy	NOUN
fcis-20320	19	30	when	when	SCONJ
fcis-20320	19	31	increasing	increase	VERB
fcis-20320	19	32	the	the	DET
fcis-20320	19	33	network	network	NOUN
fcis-20320	19	34	depth	depth	NOUN
fcis-20320	19	35	.	.	PUNCT
fcis-20320	20	1	to	to	PART
fcis-20320	20	2	solve	solve	VERB
fcis-20320	20	3	this	this	DET
fcis-20320	20	4	problem	problem	NOUN
fcis-20320	20	5	,	,	PUNCT
fcis-20320	20	6	this	this	DET
fcis-20320	20	7	paper	paper	NOUN
fcis-20320	20	8	proposes	propose	VERB
fcis-20320	20	9	a	a	DET
fcis-20320	20	10	new	new	ADJ
fcis-20320	20	11	deep	deep	ADJ
fcis-20320	20	12	learning	learning	NOUN
fcis-20320	20	13	model	model	NOUN
fcis-20320	20	14	named	name	VERB
fcis-20320	20	15	res	re	NOUN
fcis-20320	20	16	-	-	PUNCT
fcis-20320	20	17	vgg	vgg	NOUN
fcis-20320	20	18	.	.	PUNCT
fcis-20320	21	1	this	this	DET
fcis-20320	21	2	model	model	NOUN
fcis-20320	21	3	combines	combine	VERB
fcis-20320	21	4	two	two	NUM
fcis-20320	21	5	different	different	ADJ
fcis-20320	21	6	deep	deep	ADJ
fcis-20320	21	7	learning	learning	NOUN
fcis-20320	21	8	models	model	NOUN
fcis-20320	21	9	,	,	PUNCT
fcis-20320	21	10	namely	namely	ADV
fcis-20320	21	11	vgg16	vgg16	NOUN
fcis-20320	21	12	and	and	CCONJ
fcis-20320	21	13	resnet	resnet	NOUN
fcis-20320	21	14	.	.	PUNCT
fcis-20320	22	1	the	the	DET
fcis-20320	22	2	rest	rest	NOUN
fcis-20320	22	3	of	of	ADP
fcis-20320	22	4	the	the	DET
fcis-20320	22	5	paper	paper	NOUN
fcis-20320	22	6	is	be	AUX
fcis-20320	22	7	organized	organize	VERB
fcis-20320	22	8	as	as	SCONJ
fcis-20320	22	9	follows	follow	VERB
fcis-20320	22	10	.	.	PUNCT
fcis-20320	23	1	section	section	NOUN
fcis-20320	23	2	2	2	NUM
fcis-20320	23	3	analyzes	analyze	VERB
fcis-20320	23	4	the	the	DET
fcis-20320	23	5	related	relate	VERB
fcis-20320	23	6	works	work	NOUN
fcis-20320	23	7	.	.	PUNCT
fcis-20320	24	1	section	section	NOUN
fcis-20320	24	2	3	3	NUM
fcis-20320	24	3	presents	present	VERB
fcis-20320	24	4	the	the	DET
fcis-20320	24	5	proposed	propose	VERB
fcis-20320	24	6	methodology	methodology	NOUN
fcis-20320	24	7	for	for	ADP
fcis-20320	24	8	the	the	DET
fcis-20320	24	9	classification	classification	NOUN
fcis-20320	24	10	of	of	ADP
fcis-20320	24	11	lung	lung	NOUN
fcis-20320	24	12	nodules	nodule	NOUN
fcis-20320	24	13	.	.	PUNCT
fcis-20320	25	1	the	the	DET
fcis-20320	25	2	experimental	experimental	ADJ
fcis-20320	25	3	results	result	NOUN
fcis-20320	25	4	obtained	obtain	VERB
fcis-20320	25	5	are	be	AUX
fcis-20320	25	6	discussed	discuss	VERB
fcis-20320	25	7	in	in	ADP
fcis-20320	25	8	section	section	NOUN
fcis-20320	25	9	4	4	NUM
fcis-20320	25	10	.	.	PUNCT
fcis-20320	26	1	the	the	DET
fcis-20320	26	2	conclusion	conclusion	NOUN
fcis-20320	26	3	of	of	ADP
fcis-20320	26	4	the	the	DET
fcis-20320	26	5	paper	paper	NOUN
fcis-20320	26	6	can	can	AUX
fcis-20320	26	7	be	be	AUX
fcis-20320	26	8	found	find	VERB
fcis-20320	26	9	in	in	ADP
fcis-20320	26	10	section	section	NOUN
fcis-20320	26	11	5	5	NUM
fcis-20320	26	12	.	.	NOUN
fcis-20320	26	13	2	2	NUM
fcis-20320	26	14	.	.	NUM
fcis-20320	26	15	related	relate	VERB
fcis-20320	26	16	works	work	NOUN
fcis-20320	26	17	after	after	SCONJ
fcis-20320	26	18	lung	lung	NOUN
fcis-20320	26	19	nodules	nodule	NOUN
fcis-20320	26	20	are	be	AUX
fcis-20320	26	21	successfully	successfully	ADV
fcis-20320	26	22	detected	detect	VERB
fcis-20320	26	23	,	,	PUNCT
fcis-20320	26	24	the	the	DET
fcis-20320	26	25	next	next	ADJ
fcis-20320	26	26	key	key	ADJ
fcis-20320	26	27	task	task	NOUN
fcis-20320	26	28	is	be	AUX
fcis-20320	26	29	to	to	PART
fcis-20320	26	30	determine	determine	VERB
fcis-20320	26	31	whether	whether	SCONJ
fcis-20320	26	32	these	these	DET
fcis-20320	26	33	detected	detect	VERB
fcis-20320	26	34	nodules	nodule	NOUN
fcis-20320	26	35	are	be	AUX
fcis-20320	26	36	benign	benign	ADJ
fcis-20320	26	37	or	or	CCONJ
fcis-20320	26	38	malignant	malignant	ADJ
fcis-20320	26	39	.	.	PUNCT
fcis-20320	27	1	the	the	DET
fcis-20320	27	2	diagnostic	diagnostic	ADJ
fcis-20320	27	3	part	part	NOUN
fcis-20320	27	4	of	of	ADP
fcis-20320	27	5	the	the	DET
fcis-20320	27	6	pulmonary	pulmonary	ADJ
fcis-20320	27	7	nodule	nodule	NOUN
fcis-20320	27	8	computer	computer	NOUN
fcis-20320	27	9	aided	aid	VERB
fcis-20320	27	10	diagnosis	diagnosis	NOUN
fcis-20320	27	11	(	(	PUNCT
fcis-20320	27	12	cad	cad	NOUN
fcis-20320	27	13	)	)	PUNCT
fcis-20320	27	14	system	system	NOUN
fcis-20320	27	15	can	can	AUX
fcis-20320	27	16	automatically	automatically	ADV
fcis-20320	27	17	distinguish	distinguish	VERB
fcis-20320	27	18	between	between	ADP
fcis-20320	27	19	benign	benign	ADJ
fcis-20320	27	20	and	and	CCONJ
fcis-20320	27	21	malignant	malignant	ADJ
fcis-20320	27	22	nodules	nodule	NOUN
fcis-20320	27	23	based	base	VERB
fcis-20320	27	24	on	on	ADP
fcis-20320	27	25	characteristics	characteristic	NOUN
fcis-20320	27	26	such	such	ADJ
fcis-20320	27	27	as	as	ADP
fcis-20320	27	28	size	size	NOUN
fcis-20320	27	29	,	,	PUNCT
fcis-20320	27	30	shape	shape	NOUN
fcis-20320	27	31	,	,	PUNCT
fcis-20320	27	32	and	and	CCONJ
fcis-20320	27	33	appearance	appearance	NOUN
fcis-20320	27	34	of	of	ADP
fcis-20320	27	35	the	the	DET
fcis-20320	27	36	nodules	nodule	NOUN
fcis-20320	27	37	.	.	PUNCT
fcis-20320	28	1	to	to	PART
fcis-20320	28	2	evaluate	evaluate	VERB
fcis-20320	28	3	the	the	DET
fcis-20320	28	4	performance	performance	NOUN
fcis-20320	28	5	of	of	ADP
fcis-20320	28	6	cad	cad	PROPN
fcis-20320	28	7	systems	system	NOUN
fcis-20320	28	8	on	on	ADP
fcis-20320	28	9	binary	binary	ADJ
fcis-20320	28	10	classification	classification	NOUN
fcis-20320	28	11	tasks	task	NOUN
fcis-20320	28	12	,	,	PUNCT
fcis-20320	28	13	the	the	DET
fcis-20320	28	14	receiver	receiver	NOUN
fcis-20320	28	15	operating	operate	VERB
fcis-20320	28	16	characteristic	characteristic	NOUN
fcis-20320	28	17	(	(	PUNCT
fcis-20320	28	18	roc	roc	PROPN
fcis-20320	28	19	)	)	PUNCT
fcis-20320	28	20	curve	curve	NOUN
fcis-20320	28	21	is	be	AUX
fcis-20320	28	22	usually	usually	ADV
fcis-20320	28	23	used	use	VERB
fcis-20320	28	24	for	for	ADP
fcis-20320	28	25	evaluation	evaluation	NOUN
fcis-20320	28	26	.	.	PUNCT
fcis-20320	29	1	the	the	DET
fcis-20320	29	2	area	area	NOUN
fcis-20320	29	3	under	under	ADP
fcis-20320	29	4	the	the	DET
fcis-20320	29	5	roc	roc	PROPN
fcis-20320	29	6	curve	curve	NOUN
fcis-20320	29	7	(	(	PUNCT
fcis-20320	29	8	auc	auc	NOUN
fcis-20320	29	9	)	)	PUNCT
fcis-20320	29	10	is	be	AUX
fcis-20320	29	11	a	a	DET
fcis-20320	29	12	commonly	commonly	ADV
fcis-20320	29	13	used	use	VERB
fcis-20320	29	14	evaluation	evaluation	NOUN
fcis-20320	29	15	index	index	NOUN
fcis-20320	29	16	,	,	PUNCT
fcis-20320	29	17	which	which	PRON
fcis-20320	29	18	can	can	AUX
fcis-20320	29	19	quantitatively	quantitatively	ADV
fcis-20320	29	20	reflect	reflect	VERB
fcis-20320	29	21	the	the	DET
fcis-20320	29	22	performance	performance	NOUN
fcis-20320	29	23	of	of	ADP
fcis-20320	29	24	a	a	DET
fcis-20320	29	25	classifier	classifier	NOUN
fcis-20320	29	26	under	under	ADP
fcis-20320	29	27	different	different	ADJ
fcis-20320	29	28	thresholds	threshold	NOUN
fcis-20320	29	29	.	.	PUNCT
fcis-20320	30	1	the	the	DET
fcis-20320	30	2	malignancy	malignancy	NOUN
fcis-20320	30	3	of	of	ADP
fcis-20320	30	4	pulmonary	pulmonary	ADJ
fcis-20320	30	5	nodules	nodule	NOUN
fcis-20320	30	6	is	be	AUX
fcis-20320	30	7	closely	closely	ADV
fcis-20320	30	8	related	relate	VERB
fcis-20320	30	9	to	to	ADP
fcis-20320	30	10	their	their	PRON
fcis-20320	30	11	geometric	geometric	ADJ
fcis-20320	30	12	size	size	NOUN
fcis-20320	30	13	,	,	PUNCT
fcis-20320	30	14	shape	shape	NOUN
fcis-20320	30	15	,	,	PUNCT
fcis-20320	30	16	and	and	CCONJ
fcis-20320	30	17	appearance	appearance	NOUN
fcis-20320	30	18	description	description	NOUN
fcis-20320	30	19	.	.	PUNCT
fcis-20320	31	1	therefore	therefore	ADV
fcis-20320	31	2	,	,	PUNCT
fcis-20320	31	3	the	the	DET
fcis-20320	31	4	pulmonary	pulmonary	ADJ
fcis-20320	31	5	nodule	nodule	NOUN
fcis-20320	31	6	cad	cad	NOUN
fcis-20320	31	7	system	system	NOUN
fcis-20320	31	8	diagnosis	diagnosis	NOUN
fcis-20320	31	9	part	part	NOUN
fcis-20320	31	10	automatically	automatically	ADV
fcis-20320	31	11	extracts	extract	VERB
fcis-20320	31	12	effective	effective	ADJ
fcis-20320	31	13	features	feature	NOUN
fcis-20320	31	14	such	such	ADJ
fcis-20320	31	15	as	as	ADP
fcis-20320	31	16	texture	texture	NOUN
fcis-20320	31	17	,	,	PUNCT
fcis-20320	31	18	shape	shape	NOUN
fcis-20320	31	19	,	,	PUNCT
fcis-20320	31	20	and	and	CCONJ
fcis-20320	31	21	growth	growth	NOUN
fcis-20320	31	22	rate	rate	NOUN
fcis-20320	31	23	of	of	ADP
fcis-20320	31	24	pulmonary	pulmonary	ADJ
fcis-20320	31	25	nodules	nodule	NOUN
fcis-20320	31	26	from	from	ADP
fcis-20320	31	27	ct	ct	NUM
fcis-20320	31	28	images	image	NOUN
fcis-20320	31	29	to	to	PART
fcis-20320	31	30	distinguish	distinguish	VERB
fcis-20320	31	31	pulmonary	pulmonary	ADJ
fcis-20320	31	32	nodules	nodule	NOUN
fcis-20320	31	33	.	.	PUNCT
fcis-20320	32	1	nature	nature	NOUN
fcis-20320	32	2	.	.	PUNCT
fcis-20320	33	1	cnn	cnn	PROPN
fcis-20320	33	2	and	and	CCONJ
fcis-20320	33	3	big	big	ADJ
fcis-20320	33	4	data	datum	NOUN
fcis-20320	33	5	technology	technology	NOUN
fcis-20320	33	6	have	have	AUX
fcis-20320	33	7	also	also	ADV
fcis-20320	33	8	achieved	achieve	VERB
fcis-20320	33	9	remarkable	remarkable	ADJ
fcis-20320	33	10	results	result	NOUN
fcis-20320	33	11	in	in	ADP
fcis-20320	33	12	the	the	DET
fcis-20320	33	13	diagnosis	diagnosis	NOUN
fcis-20320	33	14	and	and	CCONJ
fcis-20320	33	15	classification	classification	NOUN
fcis-20320	33	16	of	of	ADP
fcis-20320	33	17	pulmonary	pulmonary	ADJ
fcis-20320	33	18	nodules	nodule	NOUN
fcis-20320	33	19	.	.	PUNCT
fcis-20320	34	1	in	in	ADP
fcis-20320	34	2	fact	fact	NOUN
fcis-20320	34	3	,	,	PUNCT
fcis-20320	34	4	classification	classification	NOUN
fcis-20320	34	5	is	be	AUX
fcis-20320	34	6	the	the	DET
fcis-20320	34	7	most	most	ADV
fcis-20320	34	8	widely	widely	ADV
fcis-20320	34	9	used	use	VERB
fcis-20320	34	10	field	field	NOUN
fcis-20320	34	11	of	of	ADP
fcis-20320	34	12	cnn	cnn	PROPN
fcis-20320	34	13	.	.	PUNCT
fcis-20320	35	1	starting	start	VERB
fcis-20320	35	2	from	from	ADP
fcis-20320	35	3	alexnet	alexnet	ADJ
fcis-20320	35	4	,	,	PUNCT
fcis-20320	35	5	resnet	resnet	NOUN
fcis-20320	35	6	,	,	PUNCT
fcis-20320	35	7	densenet	densenet	NOUN
fcis-20320	35	8	,	,	PUNCT
fcis-20320	35	9	vgg	vgg	NOUN
fcis-20320	35	10	network	network	NOUN
fcis-20320	35	11	,	,	PUNCT
fcis-20320	35	12	etc	etc	X
fcis-20320	35	13	.	.	X
fcis-20320	35	14	,	,	PUNCT
fcis-20320	35	15	most	most	ADV
fcis-20320	35	16	deep	deep	ADJ
fcis-20320	35	17	learning	learning	NOUN
fcis-20320	35	18	networks	network	NOUN
fcis-20320	35	19	originating	originate	VERB
fcis-20320	35	20	from	from	ADP
fcis-20320	35	21	the	the	DET
fcis-20320	35	22	field	field	NOUN
fcis-20320	35	23	of	of	ADP
fcis-20320	35	24	computer	computer	NOUN
fcis-20320	35	25	vision	vision	NOUN
fcis-20320	35	26	have	have	AUX
fcis-20320	35	27	been	be	AUX
fcis-20320	35	28	applied	apply	VERB
fcis-20320	35	29	in	in	ADP
fcis-20320	35	30	medical	medical	ADJ
fcis-20320	35	31	image	image	NOUN
fcis-20320	35	32	classification	classification	NOUN
fcis-20320	35	33	.	.	PUNCT
fcis-20320	36	1	first	first	ADV
fcis-20320	36	2	,	,	PUNCT
fcis-20320	36	3	based	base	VERB
fcis-20320	36	4	on	on	ADP
fcis-20320	36	5	the	the	DET
fcis-20320	36	6	traditional	traditional	ADJ
fcis-20320	36	7	cnn	cnn	PROPN
fcis-20320	36	8	structure	structure	NOUN
fcis-20320	36	9	,	,	PUNCT
fcis-20320	36	10	classification	classification	NOUN
fcis-20320	36	11	can	can	AUX
fcis-20320	36	12	be	be	AUX
fcis-20320	36	13	achieved	achieve	VERB
fcis-20320	36	14	through	through	ADP
fcis-20320	36	15	a	a	DET
fcis-20320	36	16	series	series	NOUN
fcis-20320	36	17	of	of	ADP
fcis-20320	36	18	convolution	convolution	NOUN
fcis-20320	36	19	operations	operation	NOUN
fcis-20320	36	20	[	[	X
fcis-20320	36	21	3	3	NUM
fcis-20320	36	22	]	]	PUNCT
fcis-20320	36	23	.	.	PUNCT
fcis-20320	37	1	the	the	DET
fcis-20320	37	2	study	study	NOUN
fcis-20320	37	3	by	by	ADP
fcis-20320	37	4	yang	yang	PROPN
fcis-20320	37	5	et	et	PROPN
fcis-20320	37	6	al	al	PROPN
fcis-20320	37	7	.	.	PUNCT
fcis-20320	38	1	[	[	X
fcis-20320	38	2	3	3	X
fcis-20320	38	3	]	]	PUNCT
fcis-20320	38	4	showed	show	VERB
fcis-20320	38	5	that	that	SCONJ
fcis-20320	38	6	simple	simple	ADJ
fcis-20320	38	7	geometric	geometric	ADJ
fcis-20320	38	8	features	feature	NOUN
fcis-20320	38	9	can	can	AUX
fcis-20320	38	10	not	not	PART
fcis-20320	38	11	capture	capture	VERB
fcis-20320	38	12	important	important	ADJ
fcis-20320	38	13	features	feature	NOUN
fcis-20320	38	14	of	of	ADP
fcis-20320	38	15	lung	lung	NOUN
fcis-20320	38	16	nodules	nodule	NOUN
fcis-20320	38	17	that	that	PRON
fcis-20320	38	18	support	support	VERB
fcis-20320	38	19	classification	classification	NOUN
fcis-20320	38	20	using	use	VERB
fcis-20320	38	21	original	original	ADJ
fcis-20320	38	22	images	image	NOUN
fcis-20320	38	23	and	and	CCONJ
fcis-20320	38	24	nodule	nodule	NOUN
fcis-20320	38	25	masks	mask	NOUN
fcis-20320	38	26	containing	contain	VERB
fcis-20320	38	27	rich	rich	ADJ
fcis-20320	38	28	nodule	nodule	NOUN
fcis-20320	38	29	information	information	NOUN
fcis-20320	38	30	.	.	PUNCT
fcis-20320	39	1	hua	hua	PROPN
fcis-20320	39	2	et	et	PROPN
fcis-20320	39	3	al	al	PROPN
fcis-20320	39	4	.	.	PUNCT
fcis-20320	40	1	[	[	X
fcis-20320	40	2	4	4	X
fcis-20320	40	3	]	]	PUNCT
fcis-20320	40	4	introduced	introduce	VERB
fcis-20320	40	5	deep	deep	ADJ
fcis-20320	40	6	belief	belief	NOUN
fcis-20320	40	7	networks	network	NOUN
fcis-20320	40	8	and	and	CCONJ
fcis-20320	40	9	cnn	cnn	PROPN
fcis-20320	40	10	in	in	ADP
fcis-20320	40	11	the	the	DET
fcis-20320	40	12	context	context	NOUN
fcis-20320	40	13	of	of	ADP
fcis-20320	40	14	nodule	nodule	NOUN
fcis-20320	40	15	classification	classification	NOUN
fcis-20320	40	16	.	.	PUNCT
fcis-20320	41	1	experimental	experimental	ADJ
fcis-20320	41	2	results	result	NOUN
fcis-20320	41	3	show	show	VERB
fcis-20320	41	4	that	that	SCONJ
fcis-20320	41	5	deep	deep	ADJ
fcis-20320	41	6	learning	learning	NOUN
fcis-20320	41	7	methods	method	NOUN
fcis-20320	41	8	can	can	AUX
fcis-20320	41	9	achieve	achieve	VERB
fcis-20320	41	10	better	well	ADJ
fcis-20320	41	11	recognition	recognition	NOUN
fcis-20320	41	12	results	result	NOUN
fcis-20320	41	13	and	and	CCONJ
fcis-20320	41	14	have	have	VERB
fcis-20320	41	15	broad	broad	ADJ
fcis-20320	41	16	application	application	NOUN
fcis-20320	41	17	prospects	prospect	NOUN
fcis-20320	41	18	in	in	ADP
fcis-20320	41	19	the	the	DET
fcis-20320	41	20	field	field	NOUN
fcis-20320	41	21	of	of	ADP
fcis-20320	41	22	cad	cad	PROPN
fcis-20320	41	23	diagnostic	diagnostic	ADJ
fcis-20320	41	24	applications	application	NOUN
fcis-20320	41	25	.	.	PUNCT
fcis-20320	42	1	li	li	PROPN
fcis-20320	42	2	et	et	PROPN
fcis-20320	42	3	al	al	PROPN
fcis-20320	42	4	.	.	PUNCT
fcis-20320	43	1	[	[	X
fcis-20320	43	2	5	5	NUM
fcis-20320	43	3	]	]	PUNCT
fcis-20320	43	4	designed	design	VERB
fcis-20320	43	5	a	a	DET
fcis-20320	43	6	network	network	NOUN
fcis-20320	43	7	consisting	consist	VERB
fcis-20320	43	8	of	of	ADP
fcis-20320	43	9	three	three	NUM
fcis-20320	43	10	convolutional	convolutional	ADJ
fcis-20320	43	11	layers	layer	NOUN
fcis-20320	43	12	,	,	PUNCT
fcis-20320	43	13	a	a	DET
fcis-20320	43	14	max	max	PROPN
fcis-20320	43	15	pooling	pool	VERB
fcis-20320	43	16	layer	layer	NOUN
fcis-20320	43	17	,	,	PUNCT
fcis-20320	43	18	and	and	CCONJ
fcis-20320	43	19	three	three	NUM
fcis-20320	43	20	fully	fully	ADV
fcis-20320	43	21	connected	connected	ADJ
fcis-20320	43	22	layers	layer	NOUN
fcis-20320	43	23	.	.	PUNCT
fcis-20320	44	1	they	they	PRON
fcis-20320	44	2	identified	identify	VERB
fcis-20320	44	3	input	input	NOUN
fcis-20320	44	4	nodule	nodule	NOUN
fcis-20320	44	5	images	image	NOUN
fcis-20320	44	6	as	as	ADP
fcis-20320	44	7	solid	solid	ADJ
fcis-20320	44	8	,	,	PUNCT
fcis-20320	44	9	semi	semi	ADJ
fcis-20320	44	10	-	-	ADJ
fcis-20320	44	11	solid	solid	ADJ
fcis-20320	44	12	,	,	PUNCT
fcis-20320	44	13	and	and	CCONJ
fcis-20320	44	14	ground	ground	NOUN
fcis-20320	44	15	-	-	PUNCT
fcis-20320	44	16	glass	glass	NOUN
fcis-20320	44	17	opacity	opacity	NOUN
fcis-20320	44	18	(	(	PUNCT
fcis-20320	44	19	ggo	ggo	NOUN
fcis-20320	44	20	)	)	PUNCT
fcis-20320	44	21	nodules	nodule	NOUN
fcis-20320	44	22	or	or	CCONJ
fcis-20320	44	23	non	non	NOUN
fcis-20320	44	24	-	-	NOUN
fcis-20320	44	25	nodules	nodule	NOUN
fcis-20320	44	26	.	.	PUNCT
fcis-20320	45	1	on	on	ADP
fcis-20320	45	2	the	the	DET
fcis-20320	45	3	lidc	lidc	ADJ
fcis-20320	45	4	dataset	dataset	NOUN
fcis-20320	45	5	,	,	PUNCT
fcis-20320	45	6	the	the	DET
fcis-20320	45	7	sensitivity	sensitivity	NOUN
fcis-20320	45	8	7	7	NUM
fcis-20320	45	9	of	of	ADP
fcis-20320	45	10	this	this	DET
fcis-20320	45	11	method	method	NOUN
fcis-20320	45	12	is	be	AUX
fcis-20320	45	13	87.1	87.1	NUM
fcis-20320	45	14	%	%	NOUN
fcis-20320	45	15	.	.	PUNCT
fcis-20320	46	1	instead	instead	ADV
fcis-20320	46	2	of	of	ADP
fcis-20320	46	3	using	use	VERB
fcis-20320	46	4	3d	3d	PROPN
fcis-20320	46	5	volumes	volume	NOUN
fcis-20320	46	6	,	,	PUNCT
fcis-20320	46	7	sahu	sahu	PROPN
fcis-20320	46	8	et	et	PROPN
fcis-20320	46	9	al	al	PROPN
fcis-20320	46	10	.	.	PUNCT
fcis-20320	47	1	[	[	X
fcis-20320	47	2	6	6	NUM
fcis-20320	47	3	]	]	PUNCT
fcis-20320	47	4	used	use	VERB
fcis-20320	47	5	a	a	DET
fcis-20320	47	6	lightweight	lightweight	ADJ
fcis-20320	47	7	network	network	NOUN
fcis-20320	47	8	to	to	PART
fcis-20320	47	9	process	process	VERB
fcis-20320	47	10	cross	cros	NOUN
fcis-20320	47	11	-	-	NOUN
fcis-20320	47	12	sections	section	NOUN
fcis-20320	47	13	of	of	ADP
fcis-20320	47	14	nodules	nodule	NOUN
fcis-20320	47	15	,	,	PUNCT
fcis-20320	47	16	used	use	VERB
fcis-20320	47	17	compact	compact	ADJ
fcis-20320	47	18	representations	representation	NOUN
fcis-20320	47	19	and	and	CCONJ
fcis-20320	47	20	convolutions	convolution	NOUN
fcis-20320	47	21	for	for	ADP
fcis-20320	47	22	nodule	nodule	NOUN
fcis-20320	47	23	classification	classification	NOUN
fcis-20320	47	24	,	,	PUNCT
fcis-20320	47	25	and	and	CCONJ
fcis-20320	47	26	achieved	achieve	VERB
fcis-20320	47	27	an	an	DET
fcis-20320	47	28	average	average	NOUN
fcis-20320	47	29	of	of	ADP
fcis-20320	47	30	93.18	93.18	NUM
fcis-20320	47	31	%	%	NOUN
fcis-20320	47	32	on	on	ADP
fcis-20320	47	33	the	the	DET
fcis-20320	47	34	lidc	lidc	ADJ
fcis-20320	47	35	dataset	dataset	NOUN
fcis-20320	47	36	accuracy	accuracy	NOUN
fcis-20320	47	37	.	.	PUNCT
fcis-20320	48	1	residual	residual	ADJ
fcis-20320	48	2	networks	network	NOUN
fcis-20320	48	3	(	(	PUNCT
fcis-20320	48	4	resnet	resnet	NOUN
fcis-20320	48	5	)	)	PUNCT
fcis-20320	49	1	[	[	X
fcis-20320	49	2	7	7	X
fcis-20320	49	3	]	]	PUNCT
fcis-20320	49	4	is	be	AUX
fcis-20320	49	5	another	another	DET
fcis-20320	49	6	commonly	commonly	ADV
fcis-20320	49	7	used	use	VERB
fcis-20320	49	8	network	network	NOUN
fcis-20320	49	9	,	,	PUNCT
fcis-20320	49	10	which	which	PRON
fcis-20320	49	11	is	be	AUX
fcis-20320	49	12	easier	easy	ADJ
fcis-20320	49	13	to	to	PART
fcis-20320	49	14	converge	converge	VERB
fcis-20320	49	15	during	during	ADP
fcis-20320	49	16	the	the	DET
fcis-20320	49	17	training	training	NOUN
fcis-20320	49	18	phase	phase	NOUN
fcis-20320	49	19	due	due	ADJ
fcis-20320	49	20	to	to	PART
fcis-20320	49	21	skip	skip	VERB
fcis-20320	49	22	connections	connection	NOUN
fcis-20320	49	23	.	.	PUNCT
fcis-20320	50	1	gong	gong	PROPN
fcis-20320	50	2	et	et	PROPN
fcis-20320	50	3	al	al	PROPN
fcis-20320	50	4	.	.	PUNCT
fcis-20320	51	1	[	[	X
fcis-20320	51	2	8	8	NUM
fcis-20320	51	3	]	]	PUNCT
fcis-20320	51	4	used	use	VERB
fcis-20320	51	5	a	a	DET
fcis-20320	51	6	cnn	cnn	PROPN
fcis-20320	51	7	model	model	NOUN
fcis-20320	51	8	based	base	VERB
fcis-20320	51	9	on	on	ADP
fcis-20320	51	10	residual	residual	ADJ
fcis-20320	51	11	learning	learning	NOUN
fcis-20320	51	12	similar	similar	ADJ
fcis-20320	51	13	to	to	ADP
fcis-20320	51	14	the	the	DET
fcis-20320	51	15	resnet	resnet	NOUN
fcis-20320	51	16	structure	structure	NOUN
fcis-20320	51	17	to	to	PART
fcis-20320	51	18	predict	predict	VERB
fcis-20320	51	19	the	the	DET
fcis-20320	51	20	possibility	possibility	NOUN
fcis-20320	51	21	of	of	ADP
fcis-20320	51	22	ground	ground	NOUN
fcis-20320	51	23	glass	glass	NOUN
fcis-20320	51	24	nodules	nodule	NOUN
fcis-20320	51	25	being	be	AUX
fcis-20320	51	26	invasive	invasive	ADJ
fcis-20320	51	27	adenocarcinoma	adenocarcinoma	NOUN
fcis-20320	51	28	(	(	PUNCT
fcis-20320	51	29	ia	ia	PROPN
fcis-20320	51	30	)	)	PUNCT
fcis-20320	51	31	.	.	PUNCT
fcis-20320	52	1	the	the	DET
fcis-20320	52	2	auc	auc	NOUN
fcis-20320	52	3	of	of	ADP
fcis-20320	52	4	this	this	DET
fcis-20320	52	5	method	method	NOUN
fcis-20320	52	6	for	for	ADP
fcis-20320	52	7	classifying	classify	VERB
fcis-20320	52	8	ia	ia	PROPN
fcis-20320	52	9	and	and	CCONJ
fcis-20320	52	10	non	non	ADJ
fcis-20320	52	11	-	-	ADJ
fcis-20320	52	12	ia	ia	ADJ
fcis-20320	52	13	ground	ground	NOUN
fcis-20320	52	14	-	-	PUNCT
fcis-20320	52	15	glass	glass	NOUN
fcis-20320	52	16	nodules	nodule	NOUN
fcis-20320	52	17	was	be	AUX
fcis-20320	52	18	0.92	0.92	NUM
fcis-20320	52	19	±	±	NUM
fcis-20320	52	20	0.03	0.03	NUM
fcis-20320	52	21	.	.	PUNCT
fcis-20320	53	1	xia	xia	PROPN
fcis-20320	53	2	et	et	PROPN
fcis-20320	53	3	al	al	PROPN
fcis-20320	53	4	.	.	PUNCT
fcis-20320	54	1	[	[	X
fcis-20320	54	2	9	9	NUM
fcis-20320	54	3	]	]	PUNCT
fcis-20320	54	4	proposed	propose	VERB
fcis-20320	54	5	a	a	DET
fcis-20320	54	6	recursive	recursive	ADJ
fcis-20320	54	7	residual	residual	ADJ
fcis-20320	54	8	cnn	cnn	NOUN
fcis-20320	54	9	based	base	VERB
fcis-20320	54	10	on	on	ADP
fcis-20320	54	11	u	u	NOUN
fcis-20320	54	12	-	-	NOUN
fcis-20320	54	13	net	net	ADJ
fcis-20320	54	14	to	to	ADP
fcis-20320	54	15	segment	segment	NOUN
fcis-20320	54	16	nodules	nodule	NOUN
fcis-20320	54	17	.	.	PUNCT
fcis-20320	55	1	they	they	PRON
fcis-20320	55	2	combined	combine	VERB
fcis-20320	55	3	deep	deep	ADJ
fcis-20320	55	4	learning	learning	NOUN
fcis-20320	55	5	and	and	CCONJ
fcis-20320	55	6	radiomics	radiomic	NOUN
fcis-20320	55	7	features	feature	VERB
fcis-20320	55	8	to	to	PART
fcis-20320	55	9	classify	classify	VERB
fcis-20320	55	10	non	non	ADJ
fcis-20320	55	11	-	-	ADJ
fcis-20320	55	12	ia	ia	ADJ
fcis-20320	55	13	nodules	nodule	NOUN
fcis-20320	55	14	and	and	CCONJ
fcis-20320	55	15	ia	ia	NOUN
fcis-20320	55	16	nodules	nodule	NOUN
fcis-20320	55	17	,	,	PUNCT
fcis-20320	55	18	and	and	CCONJ
fcis-20320	55	19	the	the	DET
fcis-20320	55	20	obtained	obtain	VERB
fcis-20320	55	21	auc	auc	NOUN
fcis-20320	55	22	value	value	NOUN
fcis-20320	55	23	was	be	AUX
fcis-20320	55	24	0.90	0.90	NUM
fcis-20320	55	25	±	±	NUM
fcis-20320	55	26	0.03	0.03	NUM
fcis-20320	55	27	.	.	PUNCT
fcis-20320	56	1	wu	wu	PROPN
fcis-20320	56	2	et	et	PROPN
fcis-20320	56	3	al	al	PROPN
fcis-20320	56	4	.	.	PUNCT
fcis-20320	57	1	[	[	X
fcis-20320	57	2	10	10	NUM
fcis-20320	57	3	]	]	PUNCT
fcis-20320	57	4	used	use	VERB
fcis-20320	57	5	a	a	DET
fcis-20320	57	6	50	50	NUM
fcis-20320	57	7	-	-	PUNCT
fcis-20320	57	8	layer	layer	NOUN
fcis-20320	57	9	resnet	resnet	NOUN
fcis-20320	57	10	network	network	NOUN
fcis-20320	57	11	structure	structure	NOUN
fcis-20320	57	12	as	as	ADP
fcis-20320	57	13	the	the	DET
fcis-20320	57	14	initial	initial	ADJ
fcis-20320	57	15	model	model	NOUN
fcis-20320	57	16	and	and	CCONJ
fcis-20320	57	17	combined	combine	VERB
fcis-20320	57	18	residual	residual	ADJ
fcis-20320	57	19	learning	learning	NOUN
fcis-20320	57	20	and	and	CCONJ
fcis-20320	57	21	transfer	transfer	VERB
fcis-20320	57	22	learning	learn	VERB
fcis-20320	57	23	to	to	PART
fcis-20320	57	24	achieve	achieve	VERB
fcis-20320	57	25	an	an	DET
fcis-20320	57	26	average	average	ADJ
fcis-20320	57	27	accuracy	accuracy	NOUN
fcis-20320	57	28	of	of	ADP
fcis-20320	57	29	98.23	98.23	NUM
fcis-20320	57	30	%	%	NOUN
fcis-20320	57	31	.	.	PUNCT
fcis-20320	58	1	the	the	DET
fcis-20320	58	2	advantage	advantage	NOUN
fcis-20320	58	3	of	of	ADP
fcis-20320	58	4	densenet	densenet	NOUN
fcis-20320	58	5	is	be	AUX
fcis-20320	58	6	that	that	SCONJ
fcis-20320	58	7	it	it	PRON
fcis-20320	58	8	improves	improve	VERB
fcis-20320	58	9	the	the	DET
fcis-20320	58	10	information	information	NOUN
fcis-20320	58	11	flow	flow	NOUN
fcis-20320	58	12	and	and	CCONJ
fcis-20320	58	13	gradient	gradient	NOUN
fcis-20320	58	14	of	of	ADP
fcis-20320	58	15	the	the	DET
fcis-20320	58	16	entire	entire	ADJ
fcis-20320	58	17	network	network	NOUN
fcis-20320	58	18	.	.	PUNCT
fcis-20320	59	1	each	each	DET
fcis-20320	59	2	layer	layer	NOUN
fcis-20320	59	3	can	can	AUX
fcis-20320	59	4	directly	directly	ADV
fcis-20320	59	5	obtain	obtain	VERB
fcis-20320	59	6	the	the	DET
fcis-20320	59	7	gradient	gradient	NOUN
fcis-20320	59	8	from	from	ADP
fcis-20320	59	9	the	the	DET
fcis-20320	59	10	loss	loss	NOUN
fcis-20320	59	11	function	function	NOUN
fcis-20320	59	12	and	and	CCONJ
fcis-20320	59	13	the	the	DET
fcis-20320	59	14	original	original	ADJ
fcis-20320	59	15	input	input	NOUN
fcis-20320	59	16	signal	signal	NOUN
fcis-20320	59	17	,	,	PUNCT
fcis-20320	59	18	thus	thus	ADV
fcis-20320	59	19	forming	form	VERB
fcis-20320	59	20	implicit	implicit	ADJ
fcis-20320	59	21	deep	deep	ADJ
fcis-20320	59	22	supervision	supervision	NOUN
fcis-20320	59	23	and	and	CCONJ
fcis-20320	59	24	making	make	VERB
fcis-20320	59	25	the	the	DET
fcis-20320	59	26	network	network	NOUN
fcis-20320	59	27	easy	easy	ADJ
fcis-20320	59	28	to	to	PART
fcis-20320	59	29	train	train	VERB
fcis-20320	59	30	.	.	PUNCT
fcis-20320	60	1	however	however	ADV
fcis-20320	60	2	,	,	PUNCT
fcis-20320	60	3	this	this	DET
fcis-20320	60	4	network	network	NOUN
fcis-20320	60	5	takes	take	VERB
fcis-20320	60	6	up	up	ADP
fcis-20320	60	7	more	more	ADJ
fcis-20320	60	8	gpu	gpu	NOUN
fcis-20320	60	9	memory	memory	NOUN
fcis-20320	60	10	and	and	CCONJ
fcis-20320	60	11	requires	require	VERB
fcis-20320	60	12	more	more	ADJ
fcis-20320	60	13	training	training	NOUN
fcis-20320	60	14	samples	sample	NOUN
fcis-20320	60	15	.	.	PUNCT
fcis-20320	61	1	for	for	ADP
fcis-20320	61	2	pulmonary	pulmonary	ADJ
fcis-20320	61	3	nodule	nodule	NOUN
fcis-20320	61	4	classification	classification	NOUN
fcis-20320	61	5	,	,	PUNCT
fcis-20320	61	6	li	li	PROPN
fcis-20320	61	7	et	et	PROPN
fcis-20320	61	8	al	al	PROPN
fcis-20320	61	9	.	.	PUNCT
fcis-20320	62	1	[	[	X
fcis-20320	62	2	11	11	NUM
fcis-20320	62	3	]	]	PUNCT
fcis-20320	62	4	used	use	VERB
fcis-20320	62	5	a	a	DET
fcis-20320	62	6	multi	multi	ADJ
fcis-20320	62	7	-	-	NOUN
fcis-20320	62	8	task	task	ADJ
fcis-20320	62	9	learning	learn	VERB
fcis-20320	62	10	deep	deep	ADJ
fcis-20320	62	11	neural	neural	ADJ
fcis-20320	62	12	network	network	NOUN
fcis-20320	62	13	based	base	VERB
fcis-20320	62	14	on	on	ADP
fcis-20320	62	15	3d	3d	NUM
fcis-20320	62	16	densenet	densenet	NOUN
fcis-20320	62	17	and	and	CCONJ
fcis-20320	62	18	achieved	achieve	VERB
fcis-20320	62	19	88.8	88.8	NUM
fcis-20320	62	20	%	%	NOUN
fcis-20320	62	21	results	result	NOUN
fcis-20320	62	22	in	in	ADP
fcis-20320	62	23	identifying	identify	VERB
fcis-20320	62	24	benign	benign	ADJ
fcis-20320	62	25	or	or	CCONJ
fcis-20320	62	26	malignant	malignant	ADJ
fcis-20320	62	27	pulmonary	pulmonary	ADJ
fcis-20320	62	28	nodules	nodule	NOUN
fcis-20320	62	29	.	.	PUNCT
fcis-20320	63	1	baldwin	baldwin	PROPN
fcis-20320	63	2	et	et	PROPN
fcis-20320	63	3	al	al	PROPN
fcis-20320	63	4	.	.	PUNCT
fcis-20320	64	1	[	[	X
fcis-20320	64	2	12	12	NUM
fcis-20320	64	3	]	]	PUNCT
fcis-20320	64	4	used	use	VERB
fcis-20320	64	5	a	a	DET
fcis-20320	64	6	similar	similar	ADJ
fcis-20320	64	7	densenet	densenet	NOUN
fcis-20320	64	8	structure	structure	NOUN
fcis-20320	64	9	to	to	PART
fcis-20320	64	10	design	design	VERB
fcis-20320	64	11	a	a	DET
fcis-20320	64	12	lung	lung	NOUN
fcis-20320	64	13	cancer	cancer	NOUN
fcis-20320	64	14	prediction	prediction	NOUN
fcis-20320	64	15	cnn	cnn	PROPN
fcis-20320	64	16	(	(	PUNCT
fcis-20320	64	17	lcp	lcp	PROPN
fcis-20320	64	18	-	-	PUNCT
fcis-20320	64	19	cnn	cnn	PROPN
fcis-20320	64	20	)	)	PUNCT
fcis-20320	64	21	to	to	PART
fcis-20320	64	22	predict	predict	VERB
fcis-20320	64	23	malignant	malignant	ADJ
fcis-20320	64	24	tumors	tumor	NOUN
fcis-20320	64	25	in	in	ADP
fcis-20320	64	26	pulmonary	pulmonary	ADJ
fcis-20320	64	27	nodules	nodule	NOUN
fcis-20320	64	28	,	,	PUNCT
fcis-20320	64	29	with	with	ADP
fcis-20320	64	30	an	an	DET
fcis-20320	64	31	auc	auc	NOUN
fcis-20320	64	32	of	of	ADP
fcis-20320	64	33	89.6	89.6	NUM
fcis-20320	64	34	%	%	NOUN
fcis-20320	64	35	.	.	PUNCT
fcis-20320	65	1	ren	ren	NOUN
fcis-20320	65	2	et	et	PROPN
fcis-20320	65	3	al	al	PROPN
fcis-20320	65	4	.	.	PUNCT
fcis-20320	66	1	[	[	X
fcis-20320	66	2	13	13	NUM
fcis-20320	66	3	]	]	PUNCT
fcis-20320	66	4	are	be	AUX
fcis-20320	66	5	based	base	VERB
fcis-20320	66	6	on	on	ADP
fcis-20320	66	7	the	the	DET
fcis-20320	66	8	encoder	encoder	NOUN
fcis-20320	66	9	and	and	CCONJ
fcis-20320	66	10	decoder	decoder	NOUN
fcis-20320	66	11	structure	structure	NOUN
fcis-20320	66	12	and	and	CCONJ
fcis-20320	66	13	use	use	VERB
fcis-20320	66	14	this	this	DET
fcis-20320	66	15	structure	structure	NOUN
fcis-20320	66	16	to	to	PART
fcis-20320	66	17	learn	learn	VERB
fcis-20320	66	18	the	the	DET
fcis-20320	66	19	manifold	manifold	NOUN
fcis-20320	66	20	of	of	ADP
fcis-20320	66	21	the	the	DET
fcis-20320	66	22	original	original	ADJ
fcis-20320	66	23	nodule	nodule	NOUN
fcis-20320	66	24	image	image	NOUN
fcis-20320	66	25	and	and	CCONJ
fcis-20320	66	26	classify	classify	VERB
fcis-20320	66	27	it	it	PRON
fcis-20320	66	28	based	base	VERB
fcis-20320	66	29	on	on	ADP
fcis-20320	66	30	fully	fully	ADV
fcis-20320	66	31	convolutional	convolutional	ADJ
fcis-20320	66	32	networks	network	NOUN
fcis-20320	66	33	(	(	PUNCT
fcis-20320	66	34	fcns	fcns	NOUN
fcis-20320	66	35	)	)	PUNCT
fcis-20320	66	36	from	from	ADP
fcis-20320	66	37	the	the	DET
fcis-20320	66	38	latent	latent	NOUN
fcis-20320	66	39	space	space	NOUN
fcis-20320	66	40	.	.	PUNCT
fcis-20320	67	1	the	the	DET
fcis-20320	67	2	classification	classification	NOUN
fcis-20320	67	3	accuracy	accuracy	NOUN
fcis-20320	67	4	was	be	AUX
fcis-20320	67	5	0.90	0.90	NUM
fcis-20320	67	6	,	,	PUNCT
fcis-20320	67	7	the	the	DET
fcis-20320	67	8	sensitivity	sensitivity	NOUN
fcis-20320	67	9	was	be	AUX
fcis-20320	67	10	0.81	0.81	NUM
fcis-20320	67	11	,	,	PUNCT
fcis-20320	67	12	and	and	CCONJ
fcis-20320	67	13	the	the	DET
fcis-20320	67	14	specificity	specificity	NOUN
fcis-20320	67	15	was	be	AUX
fcis-20320	67	16	0.95	0.95	NUM
fcis-20320	67	17	.	.	PUNCT
fcis-20320	68	1	lakshmanaprabu	lakshmanaprabu	PROPN
fcis-20320	68	2	et	et	PROPN
fcis-20320	68	3	al	al	PROPN
fcis-20320	68	4	.	.	PUNCT
fcis-20320	69	1	[	[	X
fcis-20320	69	2	14	14	NUM
fcis-20320	69	3	]	]	PUNCT
fcis-20320	69	4	used	use	VERB
fcis-20320	69	5	lda	lda	PROPN
fcis-20320	69	6	to	to	PART
fcis-20320	69	7	reduce	reduce	VERB
fcis-20320	69	8	the	the	DET
fcis-20320	69	9	feature	feature	NOUN
fcis-20320	69	10	dimension	dimension	NOUN
fcis-20320	69	11	and	and	CCONJ
fcis-20320	69	12	used	use	VERB
fcis-20320	69	13	an	an	DET
fcis-20320	69	14	improved	improved	ADJ
fcis-20320	69	15	gravity	gravity	NOUN
fcis-20320	69	16	search	search	NOUN
fcis-20320	69	17	algorithm	algorithm	NOUN
fcis-20320	69	18	for	for	ADP
fcis-20320	69	19	classification	classification	NOUN
fcis-20320	69	20	.	.	PUNCT
fcis-20320	70	1	the	the	DET
fcis-20320	70	2	algorithm	algorithm	NOUN
fcis-20320	70	3	has	have	VERB
fcis-20320	70	4	a	a	DET
fcis-20320	70	5	diagnostic	diagnostic	ADJ
fcis-20320	70	6	sensitivity	sensitivity	NOUN
fcis-20320	70	7	of	of	ADP
fcis-20320	70	8	95.3	95.3	NUM
fcis-20320	70	9	%	%	NOUN
fcis-20320	70	10	and	and	CCONJ
fcis-20320	70	11	a	a	DET
fcis-20320	70	12	specificity	specificity	NOUN
fcis-20320	70	13	of	of	ADP
fcis-20320	70	14	96.2	96.2	NUM
fcis-20320	70	15	%	%	NOUN
fcis-20320	70	16	for	for	ADP
fcis-20320	70	17	malignant	malignant	ADJ
fcis-20320	70	18	tumors	tumor	NOUN
fcis-20320	70	19	.	.	PUNCT
fcis-20320	71	1	in	in	ADP
fcis-20320	71	2	order	order	NOUN
fcis-20320	71	3	to	to	PART
fcis-20320	71	4	study	study	VERB
fcis-20320	71	5	different	different	ADJ
fcis-20320	71	6	network	network	NOUN
fcis-20320	71	7	structures	structure	NOUN
fcis-20320	71	8	,	,	PUNCT
fcis-20320	71	9	song	song	NOUN
fcis-20320	71	10	et	et	PROPN
fcis-20320	71	11	al	al	PROPN
fcis-20320	71	12	.	.	PUNCT
fcis-20320	72	1	[	[	X
fcis-20320	72	2	15	15	NUM
fcis-20320	72	3	]	]	PUNCT
fcis-20320	72	4	compared	compare	VERB
fcis-20320	72	5	three	three	NUM
fcis-20320	72	6	types	type	NOUN
fcis-20320	72	7	of	of	ADP
fcis-20320	72	8	deep	deep	ADJ
fcis-20320	72	9	neural	neural	ADJ
fcis-20320	72	10	networks	network	NOUN
fcis-20320	72	11	(	(	PUNCT
fcis-20320	72	12	such	such	ADJ
fcis-20320	72	13	as	as	ADP
fcis-20320	72	14	cnn	cnn	PROPN
fcis-20320	72	15	,	,	PUNCT
fcis-20320	72	16	dnn	dnn	PROPN
fcis-20320	72	17	,	,	PUNCT
fcis-20320	72	18	and	and	CCONJ
fcis-20320	72	19	stacked	stack	VERB
fcis-20320	72	20	encoders	encoder	NOUN
fcis-20320	72	21	and	and	CCONJ
fcis-20320	72	22	decoders	decoder	NOUN
fcis-20320	72	23	)	)	PUNCT
fcis-20320	72	24	for	for	ADP
fcis-20320	72	25	benign	benign	ADJ
fcis-20320	72	26	and	and	CCONJ
fcis-20320	72	27	malignant	malignant	ADJ
fcis-20320	72	28	pulmonary	pulmonary	ADJ
fcis-20320	72	29	nodules	nodule	NOUN
fcis-20320	72	30	classification	classification	NOUN
fcis-20320	72	31	.	.	PUNCT
fcis-20320	73	1	the	the	DET
fcis-20320	73	2	results	result	NOUN
fcis-20320	73	3	show	show	VERB
fcis-20320	73	4	that	that	SCONJ
fcis-20320	73	5	the	the	DET
fcis-20320	73	6	cnn	cnn	PROPN
fcis-20320	73	7	network	network	NOUN
fcis-20320	73	8	achieved	achieve	VERB
fcis-20320	73	9	the	the	DET
fcis-20320	73	10	best	good	ADJ
fcis-20320	73	11	performance	performance	NOUN
fcis-20320	73	12	,	,	PUNCT
fcis-20320	73	13	with	with	ADP
fcis-20320	73	14	an	an	DET
fcis-20320	73	15	accuracy	accuracy	NOUN
fcis-20320	73	16	of	of	ADP
fcis-20320	73	17	84.15	84.15	NUM
fcis-20320	73	18	%	%	NOUN
fcis-20320	73	19	,	,	PUNCT
fcis-20320	73	20	a	a	DET
fcis-20320	73	21	sensitivity	sensitivity	NOUN
fcis-20320	73	22	of	of	ADP
fcis-20320	73	23	83.96	83.96	NUM
fcis-20320	73	24	%	%	NOUN
fcis-20320	73	25	,	,	PUNCT
fcis-20320	73	26	and	and	CCONJ
fcis-20320	73	27	a	a	DET
fcis-20320	73	28	specificity	specificity	NOUN
fcis-20320	73	29	of	of	ADP
fcis-20320	73	30	84.32	84.32	NUM
fcis-20320	73	31	%	%	NOUN
fcis-20320	73	32	.	.	PUNCT
fcis-20320	74	1	overall	overall	ADV
fcis-20320	74	2	,	,	PUNCT
fcis-20320	74	3	the	the	DET
fcis-20320	74	4	resnet	resnet	NOUN
fcis-20320	74	5	structure	structure	NOUN
fcis-20320	74	6	outperformed	outperform	VERB
fcis-20320	74	7	cnn	cnn	PROPN
fcis-20320	74	8	and	and	CCONJ
fcis-20320	74	9	densenet	densenet	NOUN
fcis-20320	74	10	,	,	PUNCT
fcis-20320	74	11	as	as	ADV
fcis-20320	74	12	well	well	ADV
fcis-20320	74	13	as	as	ADP
fcis-20320	74	14	autoencoder	autoencoder	NOUN
fcis-20320	74	15	networks	network	NOUN
fcis-20320	74	16	for	for	ADP
fcis-20320	74	17	nodule	nodule	NOUN
fcis-20320	74	18	classification	classification	NOUN
fcis-20320	74	19	.	.	PUNCT
fcis-20320	75	1	part	part	NOUN
fcis-20320	75	2	of	of	ADP
fcis-20320	75	3	the	the	DET
fcis-20320	75	4	reason	reason	NOUN
fcis-20320	75	5	is	be	AUX
fcis-20320	75	6	that	that	SCONJ
fcis-20320	75	7	residual	residual	ADJ
fcis-20320	75	8	learning	learning	NOUN
fcis-20320	75	9	can	can	AUX
fcis-20320	75	10	effectively	effectively	ADV
fcis-20320	75	11	retain	retain	VERB
fcis-20320	75	12	the	the	DET
fcis-20320	75	13	information	information	NOUN
fcis-20320	75	14	in	in	ADP
fcis-20320	75	15	the	the	DET
fcis-20320	75	16	original	original	ADJ
fcis-20320	75	17	feature	feature	NOUN
fcis-20320	75	18	map	map	NOUN
fcis-20320	75	19	and	and	CCONJ
fcis-20320	75	20	learn	learn	VERB
fcis-20320	75	21	the	the	DET
fcis-20320	75	22	task	task	NOUN
fcis-20320	75	23	through	through	ADP
fcis-20320	75	24	residual	residual	ADJ
fcis-20320	75	25	blocks	block	NOUN
fcis-20320	75	26	,	,	PUNCT
fcis-20320	75	27	which	which	PRON
fcis-20320	75	28	is	be	AUX
fcis-20320	75	29	more	more	ADV
fcis-20320	75	30	efficient	efficient	ADJ
fcis-20320	75	31	than	than	ADP
fcis-20320	75	32	the	the	DET
fcis-20320	75	33	convolution	convolution	NOUN
fcis-20320	75	34	of	of	ADP
fcis-20320	75	35	the	the	DET
fcis-20320	75	36	original	original	ADJ
fcis-20320	75	37	feature	feature	NOUN
fcis-20320	75	38	map	map	NOUN
fcis-20320	75	39	.	.	PUNCT
fcis-20320	76	1	on	on	ADP
fcis-20320	76	2	the	the	DET
fcis-20320	76	3	other	other	ADJ
fcis-20320	76	4	hand	hand	NOUN
fcis-20320	76	5	,	,	PUNCT
fcis-20320	76	6	autoencoders	autoencoder	NOUN
fcis-20320	76	7	compress	compress	VERB
fcis-20320	76	8	the	the	DET
fcis-20320	76	9	original	original	ADJ
fcis-20320	76	10	input	input	NOUN
fcis-20320	76	11	and	and	CCONJ
fcis-20320	76	12	feature	feature	NOUN
fcis-20320	76	13	maps	map	NOUN
fcis-20320	76	14	into	into	ADP
fcis-20320	76	15	lowdimensional	lowdimensional	ADJ
fcis-20320	76	16	vectors	vector	NOUN
fcis-20320	76	17	,	,	PUNCT
fcis-20320	76	18	which	which	PRON
fcis-20320	76	19	have	have	VERB
fcis-20320	76	20	the	the	DET
fcis-20320	76	21	advantage	advantage	NOUN
fcis-20320	76	22	of	of	ADP
fcis-20320	76	23	facilitating	facilitate	VERB
fcis-20320	76	24	statistical	statistical	ADJ
fcis-20320	76	25	modeling	modeling	NOUN
fcis-20320	76	26	,	,	PUNCT
fcis-20320	76	27	but	but	CCONJ
fcis-20320	76	28	will	will	AUX
fcis-20320	76	29	lose	lose	VERB
fcis-20320	76	30	a	a	DET
fcis-20320	76	31	lot	lot	NOUN
fcis-20320	76	32	of	of	ADP
fcis-20320	76	33	image	image	NOUN
fcis-20320	76	34	information	information	NOUN
fcis-20320	76	35	,	,	PUNCT
fcis-20320	76	36	and	and	CCONJ
fcis-20320	76	37	may	may	AUX
fcis-20320	76	38	require	require	VERB
fcis-20320	76	39	a	a	DET
fcis-20320	76	40	large	large	ADJ
fcis-20320	76	41	number	number	NOUN
fcis-20320	76	42	of	of	ADP
fcis-20320	76	43	convolution	convolution	NOUN
fcis-20320	76	44	kernels	kernel	NOUN
fcis-20320	76	45	to	to	PART
fcis-20320	76	46	store	store	VERB
fcis-20320	76	47	the	the	DET
fcis-20320	76	48	information	information	NOUN
fcis-20320	76	49	learned	learn	VERB
fcis-20320	76	50	from	from	ADP
fcis-20320	76	51	training	training	NOUN
fcis-20320	76	52	samples	sample	NOUN
fcis-20320	76	53	.	.	PUNCT
fcis-20320	77	1	information	information	NOUN
fcis-20320	77	2	"	"	PUNCT
fcis-20320	77	3	.	.	PUNCT
fcis-20320	78	1	3	3	X
fcis-20320	78	2	.	.	X
fcis-20320	78	3	materials	material	NOUN
fcis-20320	78	4	and	and	CCONJ
fcis-20320	78	5	methods	method	NOUN
fcis-20320	78	6	3.1	3.1	NUM
fcis-20320	78	7	.	.	PUNCT
fcis-20320	79	1	residual	residual	ADJ
fcis-20320	79	2	networks	network	NOUN
fcis-20320	79	3	residual	residual	ADJ
fcis-20320	79	4	networks	network	NOUN
fcis-20320	79	5	(	(	PUNCT
fcis-20320	79	6	resnet	resnet	NOUN
fcis-20320	79	7	)	)	PUNCT
fcis-20320	79	8	was	be	AUX
fcis-20320	79	9	proposed	propose	VERB
fcis-20320	79	10	by	by	ADP
fcis-20320	79	11	kaiming	kaime	VERB
fcis-20320	79	12	he	he	PRON
fcis-20320	79	13	and	and	CCONJ
fcis-20320	79	14	four	four	NUM
fcis-20320	79	15	other	other	ADJ
fcis-20320	79	16	chinese	chinese	PROPN
fcis-20320	79	17	from	from	ADP
fcis-20320	79	18	microsoft	microsoft	PROPN
fcis-20320	79	19	research	research	PROPN
fcis-20320	79	20	[	[	X
fcis-20320	79	21	7	7	NUM
fcis-20320	79	22	]	]	PUNCT
fcis-20320	79	23	.	.	PUNCT
fcis-20320	80	1	they	they	PRON
fcis-20320	80	2	successfully	successfully	ADV
fcis-20320	80	3	trained	train	VERB
fcis-20320	80	4	a	a	DET
fcis-20320	80	5	152	152	NUM
fcis-20320	80	6	-	-	PUNCT
fcis-20320	80	7	layer	layer	NOUN
fcis-20320	80	8	neural	neural	ADJ
fcis-20320	80	9	network	network	NOUN
fcis-20320	80	10	and	and	CCONJ
fcis-20320	80	11	won	win	VERB
fcis-20320	80	12	the	the	DET
fcis-20320	80	13	championship	championship	NOUN
fcis-20320	80	14	in	in	ADP
fcis-20320	80	15	the	the	DET
fcis-20320	80	16	ilsvrc2015	ilsvrc2015	PROPN
fcis-20320	80	17	competition	competition	NOUN
fcis-20320	80	18	.	.	PUNCT
fcis-20320	81	1	the	the	DET
fcis-20320	81	2	error	error	NOUN
fcis-20320	81	3	rate	rate	NOUN
fcis-20320	81	4	in	in	ADP
fcis-20320	81	5	the	the	DET
fcis-20320	81	6	top	top	ADJ
fcis-20320	81	7	5	5	NUM
fcis-20320	81	8	was	be	AUX
fcis-20320	81	9	3.57	3.57	NUM
fcis-20320	81	10	%	%	NOUN
fcis-20320	81	11	.	.	PUNCT
fcis-20320	82	1	at	at	ADP
fcis-20320	82	2	the	the	DET
fcis-20320	82	3	same	same	ADJ
fcis-20320	82	4	time	time	NOUN
fcis-20320	82	5	,	,	PUNCT
fcis-20320	82	6	the	the	DET
fcis-20320	82	7	number	number	NOUN
fcis-20320	82	8	of	of	ADP
fcis-20320	82	9	parameters	parameter	NOUN
fcis-20320	82	10	was	be	AUX
fcis-20320	82	11	lower	low	ADJ
fcis-20320	82	12	than	than	ADP
fcis-20320	82	13	that	that	PRON
fcis-20320	82	14	of	of	ADP
fcis-20320	82	15	vggnet	vggnet	NOUN
fcis-20320	82	16	,	,	PUNCT
fcis-20320	82	17	and	and	CCONJ
fcis-20320	82	18	the	the	DET
fcis-20320	82	19	effect	effect	NOUN
fcis-20320	82	20	was	be	AUX
fcis-20320	82	21	very	very	ADV
fcis-20320	82	22	outstanding	outstanding	ADJ
fcis-20320	82	23	.	.	PUNCT
fcis-20320	83	1	the	the	DET
fcis-20320	83	2	principle	principle	NOUN
fcis-20320	83	3	of	of	ADP
fcis-20320	83	4	resnet	resnet	NOUN
fcis-20320	83	5	is	be	AUX
fcis-20320	83	6	that	that	SCONJ
fcis-20320	83	7	based	base	VERB
fcis-20320	83	8	on	on	ADP
fcis-20320	83	9	the	the	DET
fcis-20320	83	10	original	original	ADJ
fcis-20320	83	11	cnn	cnn	PROPN
fcis-20320	83	12	,	,	PUNCT
fcis-20320	83	13	a	a	DET
fcis-20320	83	14	residual	residual	ADJ
fcis-20320	83	15	block	block	NOUN
fcis-20320	83	16	is	be	AUX
fcis-20320	83	17	introduced	introduce	VERB
fcis-20320	83	18	.	.	PUNCT
fcis-20320	84	1	each	each	DET
fcis-20320	84	2	residual	residual	ADJ
fcis-20320	84	3	block	block	NOUN
fcis-20320	84	4	contains	contain	VERB
fcis-20320	84	5	multiple	multiple	ADJ
fcis-20320	84	6	convolution	convolution	NOUN
fcis-20320	84	7	layers	layer	NOUN
fcis-20320	84	8	and	and	CCONJ
fcis-20320	84	9	normalization	normalization	NOUN
fcis-20320	84	10	layers	layer	NOUN
fcis-20320	84	11	,	,	PUNCT
fcis-20320	84	12	and	and	CCONJ
fcis-20320	84	13	the	the	DET
fcis-20320	84	14	output	output	NOUN
fcis-20320	84	15	and	and	CCONJ
fcis-20320	84	16	input	input	NOUN
fcis-20320	84	17	are	be	AUX
fcis-20320	84	18	added	add	VERB
fcis-20320	84	19	through	through	ADP
fcis-20320	84	20	cross	cross	ADJ
fcis-20320	84	21	-	-	ADJ
fcis-20320	84	22	layer	layer	ADJ
fcis-20320	84	23	connections	connection	NOUN
fcis-20320	84	24	.	.	PUNCT
fcis-20320	85	1	combination	combination	NOUN
fcis-20320	85	2	of	of	ADP
fcis-20320	85	3	forms	form	NOUN
fcis-20320	85	4	.	.	PUNCT
fcis-20320	86	1	this	this	DET
fcis-20320	86	2	cross	cross	ADJ
fcis-20320	86	3	-	-	ADJ
fcis-20320	86	4	layer	layer	ADJ
fcis-20320	86	5	connection	connection	NOUN
fcis-20320	86	6	allows	allow	VERB
fcis-20320	86	7	the	the	DET
fcis-20320	86	8	model	model	NOUN
fcis-20320	86	9	to	to	PART
fcis-20320	86	10	be	be	AUX
fcis-20320	86	11	optimized	optimize	VERB
fcis-20320	86	12	during	during	ADP
fcis-20320	86	13	training	training	NOUN
fcis-20320	86	14	,	,	PUNCT
fcis-20320	86	15	thereby	thereby	ADV
fcis-20320	86	16	improving	improve	VERB
fcis-20320	86	17	model	model	NOUN
fcis-20320	86	18	accuracy	accuracy	NOUN
fcis-20320	86	19	.	.	PUNCT
fcis-20320	87	1	fig	fig	NOUN
fcis-20320	87	2	1	1	NUM
fcis-20320	87	3	.	.	PUNCT
fcis-20320	88	1	the	the	DET
fcis-20320	88	2	residual	residual	ADJ
fcis-20320	88	3	block	block	NOUN
fcis-20320	88	4	residual	residual	ADJ
fcis-20320	88	5	learning	learning	NOUN
fcis-20320	88	6	is	be	AUX
fcis-20320	88	7	used	use	VERB
fcis-20320	88	8	for	for	ADP
fcis-20320	88	9	every	every	DET
fcis-20320	88	10	few	few	ADJ
fcis-20320	88	11	stacked	stack	VERB
fcis-20320	88	12	layers	layer	NOUN
fcis-20320	88	13	.	.	PUNCT
fcis-20320	89	1	the	the	DET
fcis-20320	89	2	structure	structure	NOUN
fcis-20320	89	3	of	of	ADP
fcis-20320	89	4	the	the	DET
fcis-20320	89	5	residual	residual	ADJ
fcis-20320	89	6	module	module	NOUN
fcis-20320	89	7	is	be	AUX
fcis-20320	89	8	shown	show	VERB
fcis-20320	89	9	in	in	ADP
fcis-20320	89	10	figure	figure	NOUN
fcis-20320	89	11	1	1	NUM
fcis-20320	89	12	.	.	PUNCT
fcis-20320	90	1	the	the	DET
fcis-20320	90	2	module	module	NOUN
fcis-20320	90	3	is	be	AUX
fcis-20320	90	4	defined	define	VERB
fcis-20320	90	5	as	as	ADP
fcis-20320	90	6	:	:	PUNCT
fcis-20320	90	7			NOUN
fcis-20320	90	8			VERB
fcis-20320	90	9			PROPN
fcis-20320	90	10	,	,	PUNCT
fcis-20320	90	11	iy	iy	PROPN
fcis-20320	90	12	f	f	PROPN
fcis-20320	90	13	x	x	PROPN
fcis-20320	90	14	w	w	PROPN
fcis-20320	90	15	x	x	PROPN
fcis-20320	90	16			X
fcis-20320	90	17	(	(	PUNCT
fcis-20320	90	18	1	1	X
fcis-20320	90	19	)	)	PUNCT
fcis-20320	90	20	here	here	ADV
fcis-20320	90	21	x	x	X
fcis-20320	90	22	and	and	CCONJ
fcis-20320	90	23	y	y	PROPN
fcis-20320	90	24	are	be	AUX
fcis-20320	90	25	the	the	DET
fcis-20320	90	26	input	input	NOUN
fcis-20320	90	27	and	and	CCONJ
fcis-20320	90	28	output	output	NOUN
fcis-20320	90	29	vectors	vector	NOUN
fcis-20320	90	30	of	of	ADP
fcis-20320	90	31	the	the	DET
fcis-20320	90	32	layer	layer	NOUN
fcis-20320	90	33	under	under	ADP
fcis-20320	90	34	consideration	consideration	NOUN
fcis-20320	90	35	.	.	PUNCT
fcis-20320	91	1	the	the	DET
fcis-20320	91	2	function	function	NOUN
fcis-20320	91	3			PROPN
fcis-20320	91	4			VERB
fcis-20320	91	5			PROPN
fcis-20320	91	6	,	,	PUNCT
fcis-20320	91	7	if	if	SCONJ
fcis-20320	91	8	x	x	PRON
fcis-20320	91	9	w	w	NOUN
fcis-20320	91	10	represents	represent	VERB
fcis-20320	91	11	the	the	DET
fcis-20320	91	12	residual	residual	ADJ
fcis-20320	91	13	mapping	mapping	NOUN
fcis-20320	91	14	to	to	PART
fcis-20320	91	15	be	be	AUX
fcis-20320	91	16	learned	learn	VERB
fcis-20320	91	17	.	.	PUNCT
fcis-20320	92	1	for	for	ADP
fcis-20320	92	2	the	the	DET
fcis-20320	92	3	twolayer	twolayer	NOUN
fcis-20320	92	4	example	example	NOUN
fcis-20320	92	5	in	in	ADP
fcis-20320	92	6	figure	figure	NOUN
fcis-20320	92	7	1	1	NUM
fcis-20320	92	8	,	,	PUNCT
fcis-20320	92	9			NOUN
fcis-20320	92	10	2	2	PROPN
fcis-20320	92	11	1f	1f	PROPN
fcis-20320	92	12	w	w	PROPN
fcis-20320	92	13	w	w	PROPN
fcis-20320	92	14	x	x	PROPN
fcis-20320	92	15	,	,	PUNCT
fcis-20320	92	16	where	where	SCONJ
fcis-20320	92	17			PROPN
fcis-20320	92	18	represents	represent	VERB
fcis-20320	92	19	relu	relu	NOUN
fcis-20320	93	1	[	[	X
fcis-20320	93	2	16	16	NUM
fcis-20320	93	3	]	]	PUNCT
fcis-20320	93	4	,	,	PUNCT
fcis-20320	93	5	the	the	DET
fcis-20320	93	6	bias	bias	NOUN
fcis-20320	93	7	is	be	AUX
fcis-20320	93	8	omitted	omit	VERB
fcis-20320	93	9	to	to	PART
fcis-20320	93	10	simplify	simplify	VERB
fcis-20320	93	11	the	the	DET
fcis-20320	93	12	notation	notation	NOUN
fcis-20320	93	13	.	.	PUNCT
fcis-20320	94	1	the	the	DET
fcis-20320	94	2	operation	operation	NOUN
fcis-20320	94	3	f	f	PROPN
fcis-20320	94	4	x	x	NOUN
fcis-20320	94	5	is	be	AUX
fcis-20320	94	6	performed	perform	VERB
fcis-20320	94	7	via	via	ADP
fcis-20320	94	8	shortcut	shortcut	NOUN
fcis-20320	94	9	connection	connection	NOUN
fcis-20320	94	10	and	and	CCONJ
fcis-20320	94	11	element	element	NOUN
fcis-20320	94	12	addition	addition	NOUN
fcis-20320	94	13	.	.	PUNCT
fcis-20320	95	1	take	take	VERB
fcis-20320	95	2	the	the	DET
fcis-20320	95	3	second	second	ADJ
fcis-20320	95	4	nonlinearity	nonlinearity	NOUN
fcis-20320	95	5	after	after	ADP
fcis-20320	95	6	addition	addition	NOUN
fcis-20320	95	7	,	,	PUNCT
fcis-20320	95	8	which	which	PRON
fcis-20320	95	9	is	be	AUX
fcis-20320	95	10			NOUN
fcis-20320	95	11	y	y	NUM
fcis-20320	95	12	.	.	PUNCT
fcis-20320	96	1	the	the	DET
fcis-20320	96	2	shortcut	shortcut	NOUN
fcis-20320	96	3	connection	connection	NOUN
fcis-20320	96	4	in	in	ADP
fcis-20320	96	5	equation	equation	NOUN
fcis-20320	96	6	(	(	PUNCT
fcis-20320	96	7	1	1	X
fcis-20320	96	8	)	)	PUNCT
fcis-20320	96	9	neither	neither	PRON
fcis-20320	96	10	introduces	introduce	VERB
fcis-20320	96	11	additional	additional	ADJ
fcis-20320	96	12	parameters	parameter	NOUN
fcis-20320	96	13	nor	nor	CCONJ
fcis-20320	96	14	increases	increase	VERB
fcis-20320	96	15	computational	computational	ADJ
fcis-20320	96	16	complexity	complexity	NOUN
fcis-20320	96	17	.	.	PUNCT
fcis-20320	97	1	in	in	ADP
fcis-20320	97	2	formula	formula	NOUN
fcis-20320	97	3	1	1	NUM
fcis-20320	97	4	the	the	DET
fcis-20320	97	5	dimensions	dimension	NOUN
fcis-20320	97	6	of	of	ADP
fcis-20320	97	7	x	x	PUNCT
fcis-20320	97	8	and	and	CCONJ
fcis-20320	97	9	f	f	PROPN
fcis-20320	97	10	must	must	AUX
fcis-20320	97	11	be	be	AUX
fcis-20320	97	12	equal	equal	ADJ
fcis-20320	97	13	.	.	PUNCT
fcis-20320	98	1	if	if	SCONJ
fcis-20320	98	2	this	this	PRON
fcis-20320	98	3	is	be	AUX
fcis-20320	98	4	not	not	PART
fcis-20320	98	5	the	the	DET
fcis-20320	98	6	case	case	NOUN
fcis-20320	98	7	(	(	PUNCT
fcis-20320	98	8	e.g.	e.g.	ADV
fcis-20320	98	9	when	when	SCONJ
fcis-20320	98	10	changing	change	VERB
fcis-20320	98	11	input	input	NOUN
fcis-20320	98	12	or	or	CCONJ
fcis-20320	98	13	output	output	NOUN
fcis-20320	98	14	channels	channel	NOUN
fcis-20320	98	15	)	)	PUNCT
fcis-20320	98	16	,	,	PUNCT
fcis-20320	98	17	linear	linear	PROPN
fcis-20320	98	18	projection	projection	PROPN
fcis-20320	98	19	sw	sw	PROPN
fcis-20320	98	20	can	can	AUX
fcis-20320	98	21	be	be	AUX
fcis-20320	98	22	performed	perform	VERB
fcis-20320	98	23	via	via	ADP
fcis-20320	98	24	a	a	DET
fcis-20320	98	25	shortcut	shortcut	NOUN
fcis-20320	98	26	connection	connection	NOUN
fcis-20320	98	27	to	to	PART
fcis-20320	98	28	match	match	VERB
fcis-20320	98	29	the	the	DET
fcis-20320	98	30	dimensions	dimension	NOUN
fcis-20320	98	31	:	:	PUNCT
fcis-20320	98	32			NOUN
fcis-20320	98	33			VERB
fcis-20320	98	34			PROPN
fcis-20320	98	35	,	,	PUNCT
fcis-20320	98	36	i	i	PRON
fcis-20320	98	37	sy	sy	VERB
fcis-20320	98	38	f	f	NOUN
fcis-20320	99	1	x	x	PROPN
fcis-20320	99	2	w	w	PROPN
fcis-20320	99	3	w	w	PROPN
fcis-20320	99	4	x	x	PROPN
fcis-20320	99	5			X
fcis-20320	99	6	(	(	PUNCT
fcis-20320	99	7	2	2	X
fcis-20320	99	8	)	)	PUNCT
fcis-20320	99	9	the	the	DET
fcis-20320	99	10	structure	structure	NOUN
fcis-20320	99	11	of	of	ADP
fcis-20320	99	12	resnet	resnet	NOUN
fcis-20320	99	13	mainly	mainly	ADV
fcis-20320	99	14	consists	consist	VERB
fcis-20320	99	15	of	of	ADP
fcis-20320	99	16	the	the	DET
fcis-20320	99	17	following	follow	VERB
fcis-20320	99	18	parts	part	NOUN
fcis-20320	99	19	:	:	PUNCT
fcis-20320	99	20	the	the	DET
fcis-20320	99	21	first	first	ADJ
fcis-20320	99	22	construction	construction	NOUN
fcis-20320	99	23	layer	layer	NOUN
fcis-20320	99	24	,	,	PUNCT
fcis-20320	99	25	which	which	PRON
fcis-20320	99	26	is	be	AUX
fcis-20320	99	27	constructed	construct	VERB
fcis-20320	99	28	of	of	ADP
fcis-20320	99	29	a	a	DET
fcis-20320	99	30	common	common	ADJ
fcis-20320	99	31	convolution	convolution	NOUN
fcis-20320	99	32	layer	layer	NOUN
fcis-20320	99	33	and	and	CCONJ
fcis-20320	99	34	a	a	DET
fcis-20320	99	35	maximum	maximum	ADJ
fcis-20320	99	36	pooling	pool	VERB
fcis-20320	99	37	layer	layer	NOUN
fcis-20320	99	38	.	.	PUNCT
fcis-20320	100	1	the	the	DET
fcis-20320	100	2	second	second	ADJ
fcis-20320	100	3	construction	construction	NOUN
fcis-20320	100	4	layer	layer	NOUN
fcis-20320	100	5	consists	consist	VERB
fcis-20320	100	6	of	of	ADP
fcis-20320	100	7	3	3	NUM
fcis-20320	100	8	residual	residual	ADJ
fcis-20320	100	9	modules	module	NOUN
fcis-20320	100	10	.	.	PUNCT
fcis-20320	101	1	the	the	DET
fcis-20320	101	2	third	third	ADJ
fcis-20320	101	3	,	,	PUNCT
fcis-20320	101	4	fourth	fourth	ADJ
fcis-20320	101	5	,	,	PUNCT
fcis-20320	101	6	and	and	CCONJ
fcis-20320	101	7	fifth	fifth	ADJ
fcis-20320	101	8	construction	construction	NOUN
fcis-20320	101	9	layers	layer	NOUN
fcis-20320	101	10	all	all	PRON
fcis-20320	101	11	start	start	VERB
fcis-20320	101	12	with	with	ADP
fcis-20320	101	13	the	the	DET
fcis-20320	101	14	downsampling	downsample	VERB
fcis-20320	101	15	residual	residual	ADJ
fcis-20320	101	16	module	module	NOUN
fcis-20320	101	17	,	,	PUNCT
fcis-20320	101	18	followed	follow	VERB
fcis-20320	101	19	by	by	ADP
fcis-20320	101	20	3	3	NUM
fcis-20320	101	21	,	,	PUNCT
fcis-20320	101	22	5	5	NUM
fcis-20320	101	23	,	,	PUNCT
fcis-20320	101	24	and	and	CCONJ
fcis-20320	101	25	2	2	NUM
fcis-20320	101	26	residual	residual	ADJ
fcis-20320	101	27	modules	module	NOUN
fcis-20320	101	28	.	.	PUNCT
fcis-20320	102	1	there	there	PRON
fcis-20320	102	2	are	be	VERB
fcis-20320	102	3	two	two	NUM
fcis-20320	102	4	different	different	ADJ
fcis-20320	102	5	residual	residual	ADJ
fcis-20320	102	6	structures	structure	NOUN
fcis-20320	102	7	,	,	PUNCT
fcis-20320	102	8	8	8	NUM
fcis-20320	102	9	one	one	NUM
fcis-20320	102	10	is	be	AUX
fcis-20320	102	11	used	use	VERB
fcis-20320	102	12	for	for	ADP
fcis-20320	102	13	fewer	few	ADJ
fcis-20320	102	14	network	network	NOUN
fcis-20320	102	15	layers	layer	NOUN
fcis-20320	102	16	(	(	PUNCT
fcis-20320	102	17	34	34	NUM
fcis-20320	102	18	layers	layer	NOUN
fcis-20320	102	19	)	)	PUNCT
fcis-20320	102	20	,	,	PUNCT
fcis-20320	102	21	as	as	SCONJ
fcis-20320	102	22	shown	show	VERB
fcis-20320	102	23	in	in	ADP
fcis-20320	102	24	figure	figure	NOUN
fcis-20320	102	25	2(a	2(a	NUM
fcis-20320	102	26	)	)	PUNCT
fcis-20320	102	27	,	,	PUNCT
fcis-20320	102	28	and	and	CCONJ
fcis-20320	102	29	the	the	DET
fcis-20320	102	30	other	other	ADJ
fcis-20320	102	31	is	be	AUX
fcis-20320	102	32	used	use	VERB
fcis-20320	102	33	for	for	ADP
fcis-20320	102	34	50/101/152	50/101/152	NUM
fcis-20320	102	35	layers	layer	NOUN
fcis-20320	102	36	,	,	PUNCT
fcis-20320	102	37	as	as	SCONJ
fcis-20320	102	38	shown	show	VERB
fcis-20320	102	39	in	in	ADP
fcis-20320	102	40	figure	figure	NOUN
fcis-20320	102	41	2(b	2(b	NUM
fcis-20320	102	42	)	)	PUNCT
fcis-20320	102	43	.	.	PUNCT
fcis-20320	103	1	fig	fig	NOUN
fcis-20320	103	2	2	2	NUM
fcis-20320	103	3	.	.	PUNCT
fcis-20320	104	1	the	the	DET
fcis-20320	104	2	structure	structure	NOUN
fcis-20320	104	3	of	of	ADP
fcis-20320	104	4	resnet	resnet	VERB
fcis-20320	104	5	the	the	DET
fcis-20320	104	6	main	main	ADJ
fcis-20320	104	7	contribution	contribution	NOUN
fcis-20320	104	8	of	of	ADP
fcis-20320	104	9	resnet	resnet	NOUN
fcis-20320	104	10	is	be	AUX
fcis-20320	104	11	not	not	PART
fcis-20320	104	12	to	to	PART
fcis-20320	104	13	improve	improve	VERB
fcis-20320	104	14	the	the	DET
fcis-20320	104	15	expressive	expressive	ADJ
fcis-20320	104	16	ability	ability	NOUN
fcis-20320	104	17	of	of	ADP
fcis-20320	104	18	the	the	DET
fcis-20320	104	19	model	model	NOUN
fcis-20320	104	20	but	but	CCONJ
fcis-20320	104	21	to	to	PART
fcis-20320	104	22	optimize	optimize	VERB
fcis-20320	104	23	the	the	DET
fcis-20320	104	24	training	training	NOUN
fcis-20320	104	25	process	process	NOUN
fcis-20320	104	26	of	of	ADP
fcis-20320	104	27	the	the	DET
fcis-20320	104	28	model	model	NOUN
fcis-20320	104	29	.	.	PUNCT
fcis-20320	105	1	although	although	SCONJ
fcis-20320	105	2	resnet	resnet	NOUN
fcis-20320	105	3	has	have	VERB
fcis-20320	105	4	the	the	DET
fcis-20320	105	5	same	same	ADJ
fcis-20320	105	6	solution	solution	NOUN
fcis-20320	105	7	space	space	NOUN
fcis-20320	105	8	as	as	ADP
fcis-20320	105	9	the	the	DET
fcis-20320	105	10	forward	forward	ADJ
fcis-20320	105	11	network	network	NOUN
fcis-20320	105	12	of	of	ADP
fcis-20320	105	13	the	the	DET
fcis-20320	105	14	same	same	ADJ
fcis-20320	105	15	depth	depth	NOUN
fcis-20320	105	16	,	,	PUNCT
fcis-20320	105	17	resnet	resnet	NOUN
fcis-20320	105	18	effectively	effectively	ADV
fcis-20320	105	19	improves	improve	VERB
fcis-20320	105	20	the	the	DET
fcis-20320	105	21	stability	stability	NOUN
fcis-20320	105	22	and	and	CCONJ
fcis-20320	105	23	efficiency	efficiency	NOUN
fcis-20320	105	24	of	of	ADP
fcis-20320	105	25	network	network	NOUN
fcis-20320	105	26	training	training	NOUN
fcis-20320	105	27	by	by	ADP
fcis-20320	105	28	introducing	introduce	VERB
fcis-20320	105	29	residual	residual	ADJ
fcis-20320	105	30	connections	connection	NOUN
fcis-20320	105	31	.	.	PUNCT
fcis-20320	106	1	therefore	therefore	ADV
fcis-20320	106	2	,	,	PUNCT
fcis-20320	106	3	the	the	DET
fcis-20320	106	4	core	core	NOUN
fcis-20320	106	5	advantage	advantage	NOUN
fcis-20320	106	6	of	of	ADP
fcis-20320	106	7	resnet	resnet	NOUN
fcis-20320	106	8	lies	lie	NOUN
fcis-20320	106	9	in	in	ADP
fcis-20320	106	10	its	its	PRON
fcis-20320	106	11	optimization	optimization	NOUN
fcis-20320	106	12	of	of	ADP
fcis-20320	106	13	the	the	DET
fcis-20320	106	14	deep	deep	ADJ
fcis-20320	106	15	network	network	NOUN
fcis-20320	106	16	training	training	NOUN
fcis-20320	106	17	process	process	NOUN
fcis-20320	106	18	.	.	PUNCT
fcis-20320	107	1	resnet	resnet	PROPN
fcis-20320	107	2	can	can	AUX
fcis-20320	107	3	train	train	VERB
fcis-20320	107	4	very	very	ADV
fcis-20320	107	5	deep	deep	ADJ
fcis-20320	107	6	neural	neural	ADJ
fcis-20320	107	7	networks	network	NOUN
fcis-20320	107	8	,	,	PUNCT
fcis-20320	107	9	avoid	avoid	VERB
fcis-20320	107	10	the	the	DET
fcis-20320	107	11	vanishing	vanish	VERB
fcis-20320	107	12	gradient	gradient	NOUN
fcis-20320	107	13	problem	problem	NOUN
fcis-20320	107	14	,	,	PUNCT
fcis-20320	107	15	and	and	CCONJ
fcis-20320	107	16	improve	improve	VERB
fcis-20320	107	17	the	the	DET
fcis-20320	107	18	expression	expression	NOUN
fcis-20320	107	19	ability	ability	NOUN
fcis-20320	107	20	and	and	CCONJ
fcis-20320	107	21	performance	performance	NOUN
fcis-20320	107	22	of	of	ADP
fcis-20320	107	23	the	the	DET
fcis-20320	107	24	model	model	NOUN
fcis-20320	107	25	.	.	PUNCT
fcis-20320	108	1	using	use	VERB
fcis-20320	108	2	residual	residual	ADJ
fcis-20320	108	3	connections	connection	NOUN
fcis-20320	108	4	can	can	AUX
fcis-20320	108	5	preserve	preserve	VERB
fcis-20320	108	6	the	the	DET
fcis-20320	108	7	original	original	ADJ
fcis-20320	108	8	features	feature	NOUN
fcis-20320	108	9	,	,	PUNCT
fcis-20320	108	10	making	make	VERB
fcis-20320	108	11	the	the	DET
fcis-20320	108	12	network	network	NOUN
fcis-20320	108	13	learning	learn	VERB
fcis-20320	108	14	smoother	smooth	ADJ
fcis-20320	108	15	and	and	CCONJ
fcis-20320	108	16	more	more	ADV
fcis-20320	108	17	stable	stable	ADJ
fcis-20320	108	18	,	,	PUNCT
fcis-20320	108	19	and	and	CCONJ
fcis-20320	108	20	further	far	ADV
fcis-20320	108	21	improving	improve	VERB
fcis-20320	108	22	the	the	DET
fcis-20320	108	23	accuracy	accuracy	NOUN
fcis-20320	108	24	and	and	CCONJ
fcis-20320	108	25	generalization	generalization	NOUN
fcis-20320	108	26	ability	ability	NOUN
fcis-20320	108	27	of	of	ADP
fcis-20320	108	28	the	the	DET
fcis-20320	108	29	model	model	NOUN
fcis-20320	108	30	.	.	PUNCT
fcis-20320	109	1	in	in	ADP
fcis-20320	109	2	addition	addition	NOUN
fcis-20320	109	3	,	,	PUNCT
fcis-20320	109	4	using	use	VERB
fcis-20320	109	5	resnet	resnet	NOUN
fcis-20320	109	6	can	can	AUX
fcis-20320	109	7	avoid	avoid	VERB
fcis-20320	109	8	gradient	gradient	ADJ
fcis-20320	109	9	disappearance	disappearance	NOUN
fcis-20320	109	10	and	and	CCONJ
fcis-20320	109	11	gradient	gradient	ADJ
fcis-20320	109	12	explosion	explosion	NOUN
fcis-20320	109	13	problems	problem	NOUN
fcis-20320	109	14	during	during	ADP
fcis-20320	109	15	training	training	NOUN
fcis-20320	109	16	and	and	CCONJ
fcis-20320	109	17	accelerate	accelerate	VERB
fcis-20320	109	18	network	network	NOUN
fcis-20320	109	19	convergence	convergence	NOUN
fcis-20320	109	20	.	.	PUNCT
fcis-20320	110	1	in	in	ADP
fcis-20320	110	2	recent	recent	ADJ
fcis-20320	110	3	years	year	NOUN
fcis-20320	110	4	,	,	PUNCT
fcis-20320	110	5	deep	deep	ADJ
fcis-20320	110	6	learning	learning	NOUN
fcis-20320	110	7	technology	technology	NOUN
fcis-20320	110	8	has	have	AUX
fcis-20320	110	9	shown	show	VERB
fcis-20320	110	10	excellent	excellent	ADJ
fcis-20320	110	11	performance	performance	NOUN
fcis-20320	110	12	in	in	ADP
fcis-20320	110	13	different	different	ADJ
fcis-20320	110	14	fields	field	NOUN
fcis-20320	110	15	of	of	ADP
fcis-20320	110	16	medical	medical	ADJ
fcis-20320	110	17	image	image	NOUN
fcis-20320	110	18	analysis	analysis	NOUN
fcis-20320	110	19	.	.	PUNCT
fcis-20320	111	1	several	several	ADJ
fcis-20320	111	2	deep	deep	ADJ
fcis-20320	111	3	learning	learning	NOUN
fcis-20320	111	4	architectures	architecture	NOUN
fcis-20320	111	5	have	have	AUX
fcis-20320	111	6	been	be	AUX
fcis-20320	111	7	proposed	propose	VERB
fcis-20320	111	8	and	and	CCONJ
fcis-20320	111	9	used	use	VERB
fcis-20320	111	10	for	for	ADP
fcis-20320	111	11	computational	computational	ADJ
fcis-20320	111	12	pathology	pathology	NOUN
fcis-20320	111	13	classification	classification	NOUN
fcis-20320	111	14	,	,	PUNCT
fcis-20320	111	15	segmentation	segmentation	NOUN
fcis-20320	111	16	,	,	PUNCT
fcis-20320	111	17	and	and	CCONJ
fcis-20320	111	18	detection	detection	NOUN
fcis-20320	111	19	tasks	task	NOUN
fcis-20320	111	20	.	.	PUNCT
fcis-20320	112	1	due	due	ADP
fcis-20320	112	2	to	to	ADP
fcis-20320	112	3	its	its	PRON
fcis-20320	112	4	simple	simple	ADJ
fcis-20320	112	5	modular	modular	ADJ
fcis-20320	112	6	structure	structure	NOUN
fcis-20320	112	7	,	,	PUNCT
fcis-20320	112	8	most	most	ADV
fcis-20320	112	9	downstream	downstream	ADJ
fcis-20320	112	10	applications	application	NOUN
fcis-20320	112	11	still	still	ADV
fcis-20320	112	12	use	use	VERB
fcis-20320	112	13	resnet	resnet	NOUN
fcis-20320	112	14	and	and	CCONJ
fcis-20320	112	15	its	its	PRON
fcis-20320	112	16	variants	variant	NOUN
fcis-20320	112	17	.	.	PUNCT
fcis-20320	113	1	3.2	3.2	NUM
fcis-20320	113	2	.	.	PUNCT
fcis-20320	114	1	vgg	vgg	PROPN
fcis-20320	114	2	net	net	NOUN
fcis-20320	114	3	the	the	DET
fcis-20320	114	4	vgg	vgg	PROPN
fcis-20320	114	5	network	network	NOUN
fcis-20320	114	6	,	,	PUNCT
fcis-20320	114	7	proposed	propose	VERB
fcis-20320	114	8	by	by	ADP
fcis-20320	114	9	the	the	DET
fcis-20320	114	10	visual	visual	ADJ
fcis-20320	114	11	geometry	geometry	NOUN
fcis-20320	114	12	group	group	NOUN
fcis-20320	114	13	(	(	PUNCT
fcis-20320	114	14	vgg	vgg	PROPN
fcis-20320	114	15	)	)	PUNCT
fcis-20320	114	16	of	of	ADP
fcis-20320	114	17	oxford	oxford	PROPN
fcis-20320	114	18	university	university	PROPN
fcis-20320	114	19	[	[	X
fcis-20320	114	20	17	17	NUM
fcis-20320	114	21	]	]	PUNCT
fcis-20320	114	22	,	,	PUNCT
fcis-20320	114	23	is	be	AUX
fcis-20320	114	24	a	a	DET
fcis-20320	114	25	deep	deep	ADJ
fcis-20320	114	26	convolutional	convolutional	ADJ
fcis-20320	114	27	neural	neural	ADJ
fcis-20320	114	28	network	network	NOUN
fcis-20320	114	29	architecture	architecture	NOUN
fcis-20320	114	30	.	.	PUNCT
fcis-20320	115	1	in	in	ADP
fcis-20320	115	2	the	the	DET
fcis-20320	115	3	2014	2014	NUM
fcis-20320	115	4	imagenet	imagenet	NOUN
fcis-20320	115	5	image	image	NOUN
fcis-20320	115	6	classification	classification	NOUN
fcis-20320	115	7	competition	competition	NOUN
fcis-20320	115	8	,	,	PUNCT
fcis-20320	115	9	the	the	DET
fcis-20320	115	10	vgg	vgg	PROPN
fcis-20320	115	11	network	network	NOUN
fcis-20320	115	12	performed	perform	VERB
fcis-20320	115	13	well	well	ADV
fcis-20320	115	14	with	with	ADP
fcis-20320	115	15	its	its	PRON
fcis-20320	115	16	balance	balance	NOUN
fcis-20320	115	17	of	of	ADP
fcis-20320	115	18	depth	depth	NOUN
fcis-20320	115	19	and	and	CCONJ
fcis-20320	115	20	performance	performance	NOUN
fcis-20320	115	21	,	,	PUNCT
fcis-20320	115	22	winning	win	VERB
fcis-20320	115	23	the	the	DET
fcis-20320	115	24	runner	runner	NOUN
fcis-20320	115	25	-	-	PUNCT
fcis-20320	115	26	up	up	NOUN
fcis-20320	115	27	in	in	ADP
fcis-20320	115	28	the	the	DET
fcis-20320	115	29	classification	classification	NOUN
fcis-20320	115	30	task	task	NOUN
fcis-20320	115	31	and	and	CCONJ
fcis-20320	115	32	the	the	DET
fcis-20320	115	33	championship	championship	NOUN
fcis-20320	115	34	in	in	ADP
fcis-20320	115	35	the	the	DET
fcis-20320	115	36	positioning	positioning	NOUN
fcis-20320	115	37	task	task	NOUN
fcis-20320	115	38	,	,	PUNCT
fcis-20320	115	39	and	and	CCONJ
fcis-20320	115	40	has	have	AUX
fcis-20320	115	41	been	be	AUX
fcis-20320	115	42	widely	widely	ADV
fcis-20320	115	43	used	use	VERB
fcis-20320	115	44	in	in	ADP
fcis-20320	115	45	many	many	ADJ
fcis-20320	115	46	computer	computer	NOUN
fcis-20320	115	47	vision	vision	NOUN
fcis-20320	115	48	tasks	task	NOUN
fcis-20320	115	49	.	.	PUNCT
fcis-20320	116	1	the	the	DET
fcis-20320	116	2	main	main	ADJ
fcis-20320	116	3	contribution	contribution	NOUN
fcis-20320	116	4	of	of	ADP
fcis-20320	116	5	the	the	DET
fcis-20320	116	6	vgg	vgg	PROPN
fcis-20320	116	7	network	network	NOUN
fcis-20320	116	8	is	be	AUX
fcis-20320	116	9	to	to	PART
fcis-20320	116	10	prove	prove	VERB
fcis-20320	116	11	that	that	SCONJ
fcis-20320	116	12	increasing	increase	VERB
fcis-20320	116	13	the	the	DET
fcis-20320	116	14	depth	depth	NOUN
fcis-20320	116	15	of	of	ADP
fcis-20320	116	16	the	the	DET
fcis-20320	116	17	network	network	NOUN
fcis-20320	116	18	can	can	AUX
fcis-20320	116	19	affect	affect	VERB
fcis-20320	116	20	the	the	DET
fcis-20320	116	21	final	final	ADJ
fcis-20320	116	22	performance	performance	NOUN
fcis-20320	116	23	of	of	ADP
fcis-20320	116	24	the	the	DET
fcis-20320	116	25	network	network	NOUN
fcis-20320	116	26	to	to	ADP
fcis-20320	116	27	a	a	DET
fcis-20320	116	28	certain	certain	ADJ
fcis-20320	116	29	extent	extent	NOUN
fcis-20320	116	30	.	.	PUNCT
fcis-20320	117	1	to	to	PART
fcis-20320	117	2	achieve	achieve	VERB
fcis-20320	117	3	this	this	DET
fcis-20320	117	4	goal	goal	NOUN
fcis-20320	117	5	,	,	PUNCT
fcis-20320	117	6	the	the	DET
fcis-20320	117	7	vgg	vgg	PROPN
fcis-20320	117	8	network	network	NOUN
fcis-20320	117	9	uses	use	VERB
fcis-20320	117	10	three	three	NUM
fcis-20320	117	11	3×3	3×3	NUM
fcis-20320	117	12	convolution	convolution	NOUN
fcis-20320	117	13	kernels	kernel	NOUN
fcis-20320	117	14	to	to	PART
fcis-20320	117	15	replace	replace	VERB
fcis-20320	117	16	the	the	DET
fcis-20320	117	17	7×7	7×7	NUM
fcis-20320	117	18	convolution	convolution	NOUN
fcis-20320	117	19	kernel	kernel	NOUN
fcis-20320	117	20	,	,	PUNCT
fcis-20320	117	21	and	and	CCONJ
fcis-20320	117	22	two	two	NUM
fcis-20320	117	23	3×3	3×3	NUM
fcis-20320	117	24	convolution	convolution	NOUN
fcis-20320	117	25	kernels	kernel	NOUN
fcis-20320	117	26	to	to	PART
fcis-20320	117	27	replace	replace	VERB
fcis-20320	117	28	the	the	DET
fcis-20320	117	29	5×5	5×5	NUM
fcis-20320	117	30	convolution	convolution	NOUN
fcis-20320	117	31	kernel	kernel	NOUN
fcis-20320	117	32	.	.	PUNCT
fcis-20320	118	1	the	the	DET
fcis-20320	118	2	main	main	ADJ
fcis-20320	118	3	purpose	purpose	NOUN
fcis-20320	118	4	of	of	ADP
fcis-20320	118	5	this	this	PRON
fcis-20320	118	6	is	be	AUX
fcis-20320	118	7	to	to	PART
fcis-20320	118	8	improve	improve	VERB
fcis-20320	118	9	the	the	DET
fcis-20320	118	10	depth	depth	NOUN
fcis-20320	118	11	of	of	ADP
fcis-20320	118	12	the	the	DET
fcis-20320	118	13	network	network	NOUN
fcis-20320	118	14	while	while	SCONJ
fcis-20320	118	15	ensuring	ensure	VERB
fcis-20320	118	16	the	the	DET
fcis-20320	118	17	same	same	ADJ
fcis-20320	118	18	receptive	receptive	ADJ
fcis-20320	118	19	field	field	NOUN
fcis-20320	118	20	,	,	PUNCT
fcis-20320	118	21	thereby	thereby	ADV
fcis-20320	118	22	improving	improve	VERB
fcis-20320	118	23	the	the	DET
fcis-20320	118	24	effect	effect	NOUN
fcis-20320	118	25	of	of	ADP
fcis-20320	118	26	the	the	DET
fcis-20320	118	27	neural	neural	ADJ
fcis-20320	118	28	network	network	NOUN
fcis-20320	118	29	to	to	ADP
fcis-20320	118	30	a	a	DET
fcis-20320	118	31	certain	certain	ADJ
fcis-20320	118	32	extent	extent	NOUN
fcis-20320	118	33	.	.	PUNCT
fcis-20320	119	1	the	the	DET
fcis-20320	119	2	entire	entire	ADJ
fcis-20320	119	3	vgg	vgg	NOUN
fcis-20320	119	4	network	network	NOUN
fcis-20320	119	5	uses	use	VERB
fcis-20320	119	6	the	the	DET
fcis-20320	119	7	same	same	ADJ
fcis-20320	119	8	convolution	convolution	NOUN
fcis-20320	119	9	kernel	kernel	NOUN
fcis-20320	119	10	size	size	NOUN
fcis-20320	119	11	(	(	PUNCT
fcis-20320	119	12	3×3	3×3	NUM
fcis-20320	119	13	)	)	PUNCT
fcis-20320	119	14	and	and	CCONJ
fcis-20320	119	15	maximum	maximum	ADJ
fcis-20320	119	16	pooling	pool	VERB
fcis-20320	119	17	size	size	NOUN
fcis-20320	119	18	(	(	PUNCT
fcis-20320	119	19	2×2	2×2	NOUN
fcis-20320	119	20	)	)	PUNCT
fcis-20320	119	21	.	.	PUNCT
fcis-20320	120	1	in	in	ADP
fcis-20320	120	2	addition	addition	NOUN
fcis-20320	120	3	,	,	PUNCT
fcis-20320	120	4	the	the	DET
fcis-20320	120	5	vgg	vgg	PROPN
fcis-20320	120	6	network	network	NOUN
fcis-20320	120	7	can	can	AUX
fcis-20320	120	8	improve	improve	VERB
fcis-20320	120	9	performance	performance	NOUN
fcis-20320	120	10	by	by	ADP
fcis-20320	120	11	continuously	continuously	ADV
fcis-20320	120	12	deepening	deepen	VERB
fcis-20320	120	13	the	the	DET
fcis-20320	120	14	network	network	NOUN
fcis-20320	120	15	structure	structure	NOUN
fcis-20320	120	16	.	.	PUNCT
fcis-20320	121	1	as	as	SCONJ
fcis-20320	121	2	shown	show	VERB
fcis-20320	121	3	in	in	ADP
fcis-20320	121	4	figure	figure	NOUN
fcis-20320	121	5	3	3	NUM
fcis-20320	121	6	,	,	PUNCT
fcis-20320	121	7	the	the	DET
fcis-20320	121	8	vgg	vgg	PROPN
fcis-20320	121	9	network	network	NOUN
fcis-20320	121	10	has	have	VERB
fcis-20320	121	11	two	two	NUM
fcis-20320	121	12	main	main	ADJ
fcis-20320	121	13	structures	structure	NOUN
fcis-20320	121	14	,	,	PUNCT
fcis-20320	121	15	namely	namely	ADV
fcis-20320	121	16	vgg16	vgg16	NOUN
fcis-20320	121	17	and	and	CCONJ
fcis-20320	121	18	vgg19	vgg19	PROPN
fcis-20320	121	19	.	.	PUNCT
fcis-20320	122	1	vgg16	vgg16	PROPN
fcis-20320	122	2	contains	contain	VERB
fcis-20320	122	3	16	16	NUM
fcis-20320	122	4	hidden	hidden	ADJ
fcis-20320	122	5	layers	layer	NOUN
fcis-20320	122	6	(	(	PUNCT
fcis-20320	122	7	13	13	NUM
fcis-20320	122	8	convolutional	convolutional	ADJ
fcis-20320	122	9	layers	layer	NOUN
fcis-20320	122	10	and	and	CCONJ
fcis-20320	122	11	3	3	NUM
fcis-20320	122	12	fully	fully	ADV
fcis-20320	122	13	connected	connected	ADJ
fcis-20320	122	14	layers	layer	NOUN
fcis-20320	122	15	)	)	PUNCT
fcis-20320	122	16	,	,	PUNCT
fcis-20320	122	17	while	while	SCONJ
fcis-20320	122	18	vgg19	vgg19	PROPN
fcis-20320	122	19	contains	contain	VERB
fcis-20320	122	20	19	19	NUM
fcis-20320	122	21	hidden	hidden	ADJ
fcis-20320	122	22	layers	layer	NOUN
fcis-20320	122	23	(	(	PUNCT
fcis-20320	122	24	16	16	NUM
fcis-20320	122	25	convolutional	convolutional	ADJ
fcis-20320	122	26	layers	layer	NOUN
fcis-20320	122	27	and	and	CCONJ
fcis-20320	122	28	3	3	NUM
fcis-20320	122	29	fully	fully	ADV
fcis-20320	122	30	connected	connected	ADJ
fcis-20320	122	31	layers	layer	NOUN
fcis-20320	122	32	)	)	PUNCT
fcis-20320	122	33	.	.	PUNCT
fcis-20320	123	1	there	there	PRON
fcis-20320	123	2	is	be	VERB
fcis-20320	123	3	no	no	DET
fcis-20320	123	4	essential	essential	ADJ
fcis-20320	123	5	difference	difference	NOUN
fcis-20320	123	6	between	between	ADP
fcis-20320	123	7	the	the	DET
fcis-20320	123	8	two	two	NUM
fcis-20320	123	9	structures	structure	NOUN
fcis-20320	123	10	,	,	PUNCT
fcis-20320	123	11	but	but	CCONJ
fcis-20320	123	12	the	the	DET
fcis-20320	123	13	network	network	NOUN
fcis-20320	123	14	depth	depth	NOUN
fcis-20320	123	15	is	be	AUX
fcis-20320	123	16	different	different	ADJ
fcis-20320	123	17	.	.	PUNCT
fcis-20320	124	1	fig	fig	NOUN
fcis-20320	124	2	3	3	NUM
fcis-20320	124	3	.	.	PUNCT
fcis-20320	125	1	the	the	DET
fcis-20320	125	2	structure	structure	NOUN
fcis-20320	125	3	of	of	ADP
fcis-20320	125	4	the	the	DET
fcis-20320	125	5	vgg	vgg	PROPN
fcis-20320	125	6	network	network	NOUN
fcis-20320	125	7	the	the	DET
fcis-20320	125	8	vgg	vgg	PROPN
fcis-20320	125	9	network	network	NOUN
fcis-20320	125	10	also	also	ADV
fcis-20320	125	11	solves	solve	VERB
fcis-20320	125	12	some	some	DET
fcis-20320	125	13	optimization	optimization	NOUN
fcis-20320	125	14	problems	problem	NOUN
fcis-20320	125	15	,	,	PUNCT
fcis-20320	125	16	such	such	ADJ
fcis-20320	125	17	as	as	ADP
fcis-20320	125	18	lowering	lower	VERB
fcis-20320	125	19	the	the	DET
fcis-20320	125	20	learning	learning	NOUN
fcis-20320	125	21	rate	rate	NOUN
fcis-20320	125	22	,	,	PUNCT
fcis-20320	125	23	adjusting	adjust	VERB
fcis-20320	125	24	the	the	DET
fcis-20320	125	25	initialization	initialization	NOUN
fcis-20320	125	26	method	method	NOUN
fcis-20320	125	27	of	of	ADP
fcis-20320	125	28	parameters	parameter	NOUN
fcis-20320	125	29	,	,	PUNCT
fcis-20320	125	30	adjusting	adjust	VERB
fcis-20320	125	31	the	the	DET
fcis-20320	125	32	standardization	standardization	NOUN
fcis-20320	125	33	method	method	NOUN
fcis-20320	125	34	of	of	ADP
fcis-20320	125	35	input	input	NOUN
fcis-20320	125	36	data	datum	NOUN
fcis-20320	125	37	,	,	PUNCT
fcis-20320	125	38	modifying	modify	VERB
fcis-20320	125	39	the	the	DET
fcis-20320	125	40	loss	loss	NOUN
fcis-20320	125	41	function	function	NOUN
fcis-20320	125	42	,	,	PUNCT
fcis-20320	125	43	adding	add	VERB
fcis-20320	125	44	regularization	regularization	NOUN
fcis-20320	125	45	,	,	PUNCT
fcis-20320	125	46	standardizing	standardize	VERB
fcis-20320	125	47	middle	middle	ADJ
fcis-20320	125	48	layer	layer	NOUN
fcis-20320	125	49	data	datum	NOUN
fcis-20320	125	50	,	,	PUNCT
fcis-20320	125	51	and	and	CCONJ
fcis-20320	125	52	using	use	VERB
fcis-20320	125	53	dropout	dropout	NOUN
fcis-20320	125	54	.	.	PUNCT
fcis-20320	126	1	these	these	DET
fcis-20320	126	2	optimization	optimization	NOUN
fcis-20320	126	3	strategies	strategy	NOUN
fcis-20320	126	4	further	far	ADV
fcis-20320	126	5	improve	improve	VERB
fcis-20320	126	6	the	the	DET
fcis-20320	126	7	performance	performance	NOUN
fcis-20320	126	8	of	of	ADP
fcis-20320	126	9	the	the	DET
fcis-20320	126	10	vgg	vgg	PROPN
fcis-20320	126	11	network	network	NOUN
fcis-20320	126	12	.	.	PUNCT
fcis-20320	127	1	in	in	ADP
fcis-20320	127	2	medical	medical	ADJ
fcis-20320	127	3	image	image	NOUN
fcis-20320	127	4	classification	classification	NOUN
fcis-20320	127	5	,	,	PUNCT
fcis-20320	127	6	the	the	DET
fcis-20320	127	7	vgg	vgg	ADJ
fcis-20320	127	8	network	network	NOUN
fcis-20320	127	9	is	be	AUX
fcis-20320	127	10	also	also	ADV
fcis-20320	127	11	widely	widely	ADV
fcis-20320	127	12	used	use	VERB
fcis-20320	127	13	.	.	PUNCT
fcis-20320	128	1	for	for	ADP
fcis-20320	128	2	example	example	NOUN
fcis-20320	128	3	,	,	PUNCT
fcis-20320	128	4	pre	pre	ADJ
fcis-20320	128	5	-	-	ADJ
fcis-20320	128	6	trained	train	VERB
fcis-20320	128	7	vgg	vgg	NOUN
fcis-20320	128	8	models	model	NOUN
fcis-20320	128	9	can	can	AUX
fcis-20320	128	10	be	be	AUX
fcis-20320	128	11	used	use	VERB
fcis-20320	128	12	to	to	PART
fcis-20320	128	13	identify	identify	VERB
fcis-20320	128	14	and	and	CCONJ
fcis-20320	128	15	classify	classify	VERB
fcis-20320	128	16	medical	medical	ADJ
fcis-20320	128	17	images	image	NOUN
fcis-20320	128	18	through	through	ADP
fcis-20320	128	19	transfer	transfer	NOUN
fcis-20320	128	20	learning	learning	NOUN
fcis-20320	128	21	.	.	PUNCT
fcis-20320	129	1	this	this	DET
fcis-20320	129	2	method	method	NOUN
fcis-20320	129	3	takes	take	VERB
fcis-20320	129	4	advantage	advantage	NOUN
fcis-20320	129	5	of	of	ADP
fcis-20320	129	6	the	the	DET
fcis-20320	129	7	rich	rich	ADJ
fcis-20320	129	8	feature	feature	NOUN
fcis-20320	129	9	representation	representation	NOUN
fcis-20320	129	10	obtained	obtain	VERB
fcis-20320	129	11	by	by	ADP
fcis-20320	129	12	the	the	DET
fcis-20320	129	13	vgg	vgg	PROPN
fcis-20320	129	14	network	network	NOUN
fcis-20320	129	15	trained	train	VERB
fcis-20320	129	16	on	on	ADP
fcis-20320	129	17	largescale	largescale	NOUN
fcis-20320	129	18	image	image	NOUN
fcis-20320	129	19	data	data	NOUN
fcis-20320	129	20	sets	set	NOUN
fcis-20320	129	21	,	,	PUNCT
fcis-20320	129	22	and	and	CCONJ
fcis-20320	129	23	can	can	AUX
fcis-20320	129	24	effectively	effectively	ADV
fcis-20320	129	25	improve	improve	VERB
fcis-20320	129	26	the	the	DET
fcis-20320	129	27	performance	performance	NOUN
fcis-20320	129	28	of	of	ADP
fcis-20320	129	29	medical	medical	ADJ
fcis-20320	129	30	image	image	NOUN
fcis-20320	129	31	classification	classification	NOUN
fcis-20320	129	32	.	.	PUNCT
fcis-20320	130	1	3.3	3.3	NUM
fcis-20320	130	2	.	.	PUNCT
fcis-20320	131	1	loss	loss	NOUN
fcis-20320	131	2	functions	function	NOUN
fcis-20320	131	3	cross	cross	ADJ
fcis-20320	131	4	-	-	ADJ
fcis-20320	131	5	entropy	entropy	ADJ
fcis-20320	131	6	loss	loss	NOUN
fcis-20320	131	7	function	function	NOUN
fcis-20320	131	8	,	,	PUNCT
fcis-20320	131	9	derived	derive	VERB
fcis-20320	131	10	from	from	ADP
fcis-20320	131	11	the	the	DET
fcis-20320	131	12	concept	concept	NOUN
fcis-20320	131	13	of	of	ADP
fcis-20320	131	14	cross	cross	NOUN
fcis-20320	131	15	-	-	NOUN
fcis-20320	131	16	entropy	entropy	NOUN
fcis-20320	131	17	in	in	ADP
fcis-20320	131	18	information	information	NOUN
fcis-20320	131	19	theory	theory	NOUN
fcis-20320	131	20	,	,	PUNCT
fcis-20320	131	21	is	be	AUX
fcis-20320	131	22	a	a	DET
fcis-20320	131	23	commonly	commonly	ADV
fcis-20320	131	24	used	use	VERB
fcis-20320	131	25	loss	loss	NOUN
fcis-20320	131	26	function	function	NOUN
fcis-20320	131	27	and	and	CCONJ
fcis-20320	131	28	is	be	AUX
fcis-20320	131	29	mainly	mainly	ADV
fcis-20320	131	30	used	use	VERB
fcis-20320	131	31	in	in	ADP
fcis-20320	131	32	binary	binary	ADJ
fcis-20320	131	33	classification	classification	NOUN
fcis-20320	131	34	and	and	CCONJ
fcis-20320	131	35	multiclassification	multiclassification	NOUN
fcis-20320	131	36	problems	problem	NOUN
fcis-20320	131	37	.	.	PUNCT
fcis-20320	132	1	its	its	PRON
fcis-20320	132	2	core	core	ADJ
fcis-20320	132	3	idea	idea	NOUN
fcis-20320	132	4	is	be	AUX
fcis-20320	132	5	to	to	PART
fcis-20320	132	6	measure	measure	VERB
fcis-20320	132	7	the	the	DET
fcis-20320	132	8	difference	difference	NOUN
fcis-20320	132	9	between	between	ADP
fcis-20320	132	10	two	two	NUM
fcis-20320	132	11	probability	probability	NOUN
fcis-20320	132	12	distributions	distribution	NOUN
fcis-20320	132	13	,	,	PUNCT
fcis-20320	132	14	which	which	PRON
fcis-20320	132	15	are	be	AUX
fcis-20320	132	16	usually	usually	ADV
fcis-20320	132	17	the	the	DET
fcis-20320	132	18	distribution	distribution	NOUN
fcis-20320	132	19	of	of	ADP
fcis-20320	132	20	true	true	ADJ
fcis-20320	132	21	labels	label	NOUN
fcis-20320	132	22	and	and	CCONJ
fcis-20320	132	23	the	the	DET
fcis-20320	132	24	distribution	distribution	NOUN
fcis-20320	132	25	of	of	ADP
fcis-20320	132	26	model	model	NOUN
fcis-20320	132	27	prediction	prediction	NOUN
fcis-20320	132	28	results	result	NOUN
fcis-20320	132	29	.	.	PUNCT
fcis-20320	133	1	in	in	ADP
fcis-20320	133	2	machine	machine	NOUN
fcis-20320	133	3	learning	learning	NOUN
fcis-20320	133	4	,	,	PUNCT
fcis-20320	133	5	especially	especially	ADV
fcis-20320	133	6	deep	deep	ADJ
fcis-20320	133	7	learning	learning	NOUN
fcis-20320	133	8	,	,	PUNCT
fcis-20320	133	9	the	the	DET
fcis-20320	133	10	cross	cross	ADJ
fcis-20320	133	11	-	-	ADJ
fcis-20320	133	12	entropy	entropy	ADJ
fcis-20320	133	13	loss	loss	NOUN
fcis-20320	133	14	function	function	NOUN
fcis-20320	133	15	is	be	AUX
fcis-20320	133	16	widely	widely	ADV
fcis-20320	133	17	used	use	VERB
fcis-20320	133	18	as	as	ADP
fcis-20320	133	19	an	an	DET
fcis-20320	133	20	9	9	NUM
fcis-20320	133	21	optimization	optimization	NOUN
fcis-20320	133	22	target	target	NOUN
fcis-20320	133	23	and	and	CCONJ
fcis-20320	133	24	used	use	VERB
fcis-20320	133	25	to	to	PART
fcis-20320	133	26	train	train	VERB
fcis-20320	133	27	various	various	ADJ
fcis-20320	133	28	classification	classification	NOUN
fcis-20320	133	29	models	model	NOUN
fcis-20320	133	30	,	,	PUNCT
fcis-20320	133	31	such	such	ADJ
fcis-20320	133	32	as	as	ADP
fcis-20320	133	33	logistic	logistic	ADJ
fcis-20320	133	34	regression	regression	NOUN
fcis-20320	133	35	,	,	PUNCT
fcis-20320	133	36	deep	deep	ADJ
fcis-20320	133	37	neural	neural	ADJ
fcis-20320	133	38	networks	network	NOUN
fcis-20320	133	39	,	,	PUNCT
fcis-20320	133	40	etc	etc	X
fcis-20320	133	41	.	.	X
fcis-20320	133	42	by	by	ADP
fcis-20320	133	43	minimizing	minimize	VERB
fcis-20320	133	44	the	the	DET
fcis-20320	133	45	cross	cross	ADJ
fcis-20320	133	46	-	-	ADJ
fcis-20320	133	47	entropy	entropy	ADJ
fcis-20320	133	48	loss	loss	NOUN
fcis-20320	133	49	function	function	NOUN
fcis-20320	133	50	,	,	PUNCT
fcis-20320	133	51	the	the	DET
fcis-20320	133	52	prediction	prediction	NOUN
fcis-20320	133	53	results	result	NOUN
fcis-20320	133	54	of	of	ADP
fcis-20320	133	55	the	the	DET
fcis-20320	133	56	model	model	NOUN
fcis-20320	133	57	can	can	AUX
fcis-20320	133	58	be	be	AUX
fcis-20320	133	59	made	make	VERB
fcis-20320	133	60	as	as	ADV
fcis-20320	133	61	close	close	ADJ
fcis-20320	133	62	as	as	ADP
fcis-20320	133	63	possible	possible	ADJ
fcis-20320	133	64	to	to	ADP
fcis-20320	133	65	the	the	DET
fcis-20320	133	66	real	real	ADJ
fcis-20320	133	67	labels	label	NOUN
fcis-20320	133	68	,	,	PUNCT
fcis-20320	133	69	thereby	thereby	ADV
fcis-20320	133	70	improving	improve	VERB
fcis-20320	133	71	the	the	DET
fcis-20320	133	72	performance	performance	NOUN
fcis-20320	133	73	of	of	ADP
fcis-20320	133	74	the	the	DET
fcis-20320	133	75	model	model	NOUN
fcis-20320	133	76	.	.	PUNCT
fcis-20320	134	1	for	for	ADP
fcis-20320	134	2	a	a	DET
fcis-20320	134	3	binary	binary	ADJ
fcis-20320	134	4	classification	classification	NOUN
fcis-20320	134	5	problem	problem	NOUN
fcis-20320	134	6	,	,	PUNCT
fcis-20320	134	7	the	the	DET
fcis-20320	134	8	expression	expression	NOUN
fcis-20320	134	9	of	of	ADP
fcis-20320	134	10	the	the	DET
fcis-20320	134	11	cross	cross	ADJ
fcis-20320	134	12	-	-	ADJ
fcis-20320	134	13	entropy	entropy	ADJ
fcis-20320	134	14	loss	loss	NOUN
fcis-20320	134	15	function	function	NOUN
fcis-20320	134	16	is	be	AUX
fcis-20320	134	17	:	:	PUNCT
fcis-20320	134	18			ADJ
fcis-20320	134	19	1	1	NOUN
fcis-20320	134	20	log	log	NOUN
fcis-20320	134	21	(	(	PUNCT
fcis-20320	134	22	)	)	PUNCT
fcis-20320	134	23	(	(	PUNCT
fcis-20320	134	24	1	1	X
fcis-20320	134	25	)	)	PUNCT
fcis-20320	134	26	log(1	log(1	NOUN
fcis-20320	134	27	)	)	PUNCT
fcis-20320	135	1	i	i	PRON
fcis-20320	135	2	i	i	PRON
fcis-20320	136	1	i	i	PRON
fcis-20320	136	2	i	i	PRON
fcis-20320	137	1	i	i	VERB
fcis-20320	137	2	l	l	VERB
fcis-20320	138	1	y	y	PROPN
fcis-20320	138	2	p	p	X
fcis-20320	138	3	y	y	PROPN
fcis-20320	138	4	p	p	PROPN
fcis-20320	138	5	n	n	PRON
fcis-20320	138	6			PROPN
fcis-20320	138	7			PROPN
fcis-20320	138	8			NUM
fcis-20320	138	9			PUNCT
fcis-20320	138	10			NOUN
fcis-20320	138	11			PROPN
fcis-20320	138	12			NOUN
fcis-20320	138	13	(	(	PUNCT
fcis-20320	138	14	3	3	NUM
fcis-20320	138	15	)	)	PUNCT
fcis-20320	138	16	among	among	ADP
fcis-20320	138	17	them	they	PRON
fcis-20320	138	18	,	,	PUNCT
fcis-20320	138	19	n	n	X
fcis-20320	138	20	is	be	AUX
fcis-20320	138	21	the	the	DET
fcis-20320	138	22	number	number	NOUN
fcis-20320	138	23	of	of	ADP
fcis-20320	138	24	samples	sample	NOUN
fcis-20320	138	25	,	,	PUNCT
fcis-20320	138	26	iy	iy	PROPN
fcis-20320	138	27	represents	represent	VERB
fcis-20320	138	28	the	the	DET
fcis-20320	138	29	true	true	ADJ
fcis-20320	138	30	label	label	NOUN
fcis-20320	138	31	(	(	PUNCT
fcis-20320	138	32	0	0	NUM
fcis-20320	138	33	or	or	CCONJ
fcis-20320	138	34	1	1	NUM
fcis-20320	138	35	)	)	PUNCT
fcis-20320	138	36	of	of	ADP
fcis-20320	138	37	sample	sample	NOUN
fcis-20320	138	38	i	i	PRON
fcis-20320	138	39	,	,	PUNCT
fcis-20320	138	40	and	and	CCONJ
fcis-20320	138	41	ip	ip	NOUN
fcis-20320	138	42	represents	represent	VERB
fcis-20320	138	43	the	the	DET
fcis-20320	138	44	probability	probability	NOUN
fcis-20320	138	45	that	that	SCONJ
fcis-20320	138	46	sample	sample	NOUN
fcis-20320	138	47	i	i	PRON
fcis-20320	138	48	is	be	AUX
fcis-20320	138	49	predicted	predict	VERB
fcis-20320	138	50	to	to	PART
fcis-20320	138	51	be	be	AUX
fcis-20320	138	52	a	a	DET
fcis-20320	138	53	positive	positive	ADJ
fcis-20320	138	54	class	class	NOUN
fcis-20320	138	55	.	.	PUNCT
fcis-20320	139	1	for	for	ADP
fcis-20320	139	2	multi	multi	ADJ
fcis-20320	139	3	-	-	ADJ
fcis-20320	139	4	classification	classification	ADJ
fcis-20320	139	5	problems	problem	NOUN
fcis-20320	139	6	,	,	PUNCT
fcis-20320	139	7	the	the	DET
fcis-20320	139	8	expression	expression	NOUN
fcis-20320	139	9	of	of	ADP
fcis-20320	139	10	the	the	DET
fcis-20320	139	11	cross	cross	ADJ
fcis-20320	139	12	-	-	ADJ
fcis-20320	139	13	entropy	entropy	ADJ
fcis-20320	139	14	loss	loss	NOUN
fcis-20320	139	15	function	function	NOUN
fcis-20320	139	16	is	be	AUX
fcis-20320	139	17	:	:	PUNCT
fcis-20320	139	18	1	1	NUM
fcis-20320	139	19	1	1	NUM
fcis-20320	139	20	log	log	NOUN
fcis-20320	139	21	(	(	PUNCT
fcis-20320	139	22	)	)	PUNCT
fcis-20320	139	23	m	m	VERB
fcis-20320	139	24	ic	ic	INTJ
fcis-20320	139	25	ic	ic	INTJ
fcis-20320	139	26	i	i	INTJ
fcis-20320	139	27	c	c	VERB
fcis-20320	139	28	l	l	NOUN
fcis-20320	139	29	y	y	PROPN
fcis-20320	139	30	p	p	PROPN
fcis-20320	139	31	n	n	CCONJ
fcis-20320	139	32			NUM
fcis-20320	139	33			PROPN
fcis-20320	139	34			NOUN
fcis-20320	139	35			PRON
fcis-20320	139	36	(	(	PUNCT
fcis-20320	139	37	4	4	NUM
fcis-20320	139	38	)	)	PUNCT
fcis-20320	139	39	among	among	ADP
fcis-20320	139	40	them	they	PRON
fcis-20320	139	41	,	,	PUNCT
fcis-20320	139	42	n	n	X
fcis-20320	139	43	is	be	AUX
fcis-20320	139	44	the	the	DET
fcis-20320	139	45	number	number	NOUN
fcis-20320	139	46	of	of	ADP
fcis-20320	139	47	samples	sample	NOUN
fcis-20320	139	48	,	,	PUNCT
fcis-20320	139	49	m	m	VERB
fcis-20320	139	50	is	be	AUX
fcis-20320	139	51	the	the	DET
fcis-20320	139	52	number	number	NOUN
fcis-20320	139	53	of	of	ADP
fcis-20320	139	54	categories	category	NOUN
fcis-20320	139	55	,	,	PUNCT
fcis-20320	139	56	icy	icy	ADJ
fcis-20320	139	57	is	be	AUX
fcis-20320	139	58	the	the	DET
fcis-20320	139	59	sign	sign	NOUN
fcis-20320	139	60	function	function	NOUN
fcis-20320	139	61	(	(	PUNCT
fcis-20320	139	62	0	0	NUM
fcis-20320	139	63	or	or	CCONJ
fcis-20320	139	64	1	1	NUM
fcis-20320	139	65	)	)	PUNCT
fcis-20320	139	66	,	,	PUNCT
fcis-20320	139	67	taking	take	VERB
fcis-20320	139	68	1	1	NUM
fcis-20320	139	69	if	if	SCONJ
fcis-20320	139	70	the	the	DET
fcis-20320	139	71	true	true	ADJ
fcis-20320	139	72	category	category	NOUN
fcis-20320	139	73	of	of	ADP
fcis-20320	139	74	sample	sample	NOUN
fcis-20320	139	75	i	i	PRON
fcis-20320	139	76	is	be	AUX
fcis-20320	139	77	equal	equal	ADJ
fcis-20320	139	78	to	to	ADP
fcis-20320	139	79	c	c	NOUN
fcis-20320	139	80	,	,	PUNCT
fcis-20320	139	81	otherwise	otherwise	ADV
fcis-20320	139	82	taking	take	VERB
fcis-20320	139	83	0	0	NUM
fcis-20320	139	84	,	,	PUNCT
fcis-20320	139	85	icp	icp	PROPN
fcis-20320	139	86	is	be	AUX
fcis-20320	139	87	the	the	DET
fcis-20320	139	88	predicted	predict	VERB
fcis-20320	139	89	probability	probability	NOUN
fcis-20320	139	90	that	that	SCONJ
fcis-20320	139	91	the	the	DET
fcis-20320	139	92	observed	observed	ADJ
fcis-20320	139	93	sample	sample	NOUN
fcis-20320	139	94	i	i	PRON
fcis-20320	139	95	belongs	belong	VERB
fcis-20320	139	96	to	to	ADP
fcis-20320	139	97	category	category	PROPN
fcis-20320	139	98	c	c	PROPN
fcis-20320	139	99	.	.	PUNCT
fcis-20320	140	1	the	the	DET
fcis-20320	140	2	cross	cross	ADJ
fcis-20320	140	3	-	-	ADJ
fcis-20320	140	4	entropy	entropy	ADJ
fcis-20320	140	5	loss	loss	NOUN
fcis-20320	140	6	function	function	NOUN
fcis-20320	140	7	has	have	VERB
fcis-20320	140	8	the	the	DET
fcis-20320	140	9	following	follow	VERB
fcis-20320	140	10	characteristics	characteristic	NOUN
fcis-20320	140	11	:	:	PUNCT
fcis-20320	140	12	when	when	SCONJ
fcis-20320	140	13	the	the	DET
fcis-20320	140	14	prediction	prediction	NOUN
fcis-20320	140	15	result	result	NOUN
fcis-20320	140	16	is	be	AUX
fcis-20320	140	17	correct	correct	ADJ
fcis-20320	140	18	,	,	PUNCT
fcis-20320	140	19	the	the	PRON
fcis-20320	140	20	greater	great	ADJ
fcis-20320	140	21	the	the	DET
fcis-20320	140	22	probability	probability	NOUN
fcis-20320	140	23	of	of	ADP
fcis-20320	140	24	the	the	DET
fcis-20320	140	25	correct	correct	ADJ
fcis-20320	140	26	category	category	NOUN
fcis-20320	140	27	(	(	PUNCT
fcis-20320	140	28	the	the	DET
fcis-20320	140	29	closer	close	ADJ
fcis-20320	140	30	icp	icp	PROPN
fcis-20320	140	31	is	be	AUX
fcis-20320	140	32	to	to	ADP
fcis-20320	140	33	1	1	NUM
fcis-20320	140	34	)	)	PUNCT
fcis-20320	140	35	,	,	PUNCT
fcis-20320	140	36	the	the	DET
fcis-20320	140	37	smaller	small	ADJ
fcis-20320	140	38	the	the	DET
fcis-20320	140	39	cross	cross	ADJ
fcis-20320	140	40	-	-	ADJ
fcis-20320	140	41	entropy	entropy	ADJ
fcis-20320	140	42	loss	loss	NOUN
fcis-20320	140	43	function	function	NOUN
fcis-20320	140	44	.	.	PUNCT
fcis-20320	141	1	when	when	SCONJ
fcis-20320	141	2	the	the	DET
fcis-20320	141	3	prediction	prediction	NOUN
fcis-20320	141	4	result	result	NOUN
fcis-20320	141	5	is	be	AUX
fcis-20320	141	6	correct	correct	ADJ
fcis-20320	141	7	,	,	PUNCT
fcis-20320	141	8	but	but	CCONJ
fcis-20320	141	9	the	the	DET
fcis-20320	141	10	probability	probability	NOUN
fcis-20320	141	11	of	of	ADP
fcis-20320	141	12	the	the	DET
fcis-20320	141	13	correct	correct	ADJ
fcis-20320	141	14	category	category	NOUN
fcis-20320	141	15	is	be	AUX
fcis-20320	141	16	not	not	PART
fcis-20320	141	17	large	large	ADJ
fcis-20320	141	18	enough	enough	ADV
fcis-20320	141	19	(	(	PUNCT
fcis-20320	141	20	icp	icp	PROPN
fcis-20320	141	21	is	be	AUX
fcis-20320	141	22	smaller	small	ADJ
fcis-20320	141	23	)	)	PUNCT
fcis-20320	141	24	,	,	PUNCT
fcis-20320	141	25	the	the	DET
fcis-20320	141	26	crossentropy	crossentropy	NOUN
fcis-20320	141	27	loss	loss	NOUN
fcis-20320	141	28	function	function	NOUN
fcis-20320	141	29	is	be	AUX
fcis-20320	141	30	large	large	ADJ
fcis-20320	141	31	.	.	PUNCT
fcis-20320	142	1	when	when	SCONJ
fcis-20320	142	2	the	the	DET
fcis-20320	142	3	prediction	prediction	NOUN
fcis-20320	142	4	result	result	NOUN
fcis-20320	142	5	is	be	AUX
fcis-20320	142	6	wrong	wrong	ADJ
fcis-20320	142	7	,	,	PUNCT
fcis-20320	142	8	the	the	DET
fcis-20320	142	9	cross	cross	ADJ
fcis-20320	142	10	-	-	ADJ
fcis-20320	142	11	entropy	entropy	ADJ
fcis-20320	142	12	loss	loss	NOUN
fcis-20320	142	13	function	function	NOUN
fcis-20320	142	14	is	be	AUX
fcis-20320	142	15	also	also	ADV
fcis-20320	142	16	large	large	ADJ
fcis-20320	142	17	.	.	PUNCT
fcis-20320	143	1	in	in	ADP
fcis-20320	143	2	a	a	DET
fcis-20320	143	3	classification	classification	NOUN
fcis-20320	143	4	problem	problem	NOUN
fcis-20320	143	5	,	,	PUNCT
fcis-20320	143	6	the	the	DET
fcis-20320	143	7	goal	goal	NOUN
fcis-20320	143	8	is	be	AUX
fcis-20320	143	9	to	to	PART
fcis-20320	143	10	make	make	VERB
fcis-20320	143	11	the	the	DET
fcis-20320	143	12	model	model	NOUN
fcis-20320	143	13	's	's	PART
fcis-20320	143	14	predictions	prediction	NOUN
fcis-20320	143	15	as	as	ADV
fcis-20320	143	16	close	close	ADJ
fcis-20320	143	17	as	as	ADP
fcis-20320	143	18	possible	possible	ADJ
fcis-20320	143	19	to	to	ADP
fcis-20320	143	20	the	the	DET
fcis-20320	143	21	true	true	ADJ
fcis-20320	143	22	labels	label	NOUN
fcis-20320	143	23	.	.	PUNCT
fcis-20320	144	1	the	the	DET
fcis-20320	144	2	crossentropy	crossentropy	NOUN
fcis-20320	144	3	loss	loss	NOUN
fcis-20320	144	4	function	function	NOUN
fcis-20320	144	5	exactly	exactly	ADV
fcis-20320	144	6	meets	meet	VERB
fcis-20320	144	7	this	this	DET
fcis-20320	144	8	requirement	requirement	NOUN
fcis-20320	144	9	.	.	PUNCT
fcis-20320	145	1	it	it	PRON
fcis-20320	145	2	can	can	AUX
fcis-20320	145	3	effectively	effectively	ADV
fcis-20320	145	4	measure	measure	VERB
fcis-20320	145	5	the	the	DET
fcis-20320	145	6	difference	difference	NOUN
fcis-20320	145	7	between	between	ADP
fcis-20320	145	8	the	the	DET
fcis-20320	145	9	model	model	NOUN
fcis-20320	145	10	's	's	PART
fcis-20320	145	11	prediction	prediction	NOUN
fcis-20320	145	12	results	result	NOUN
fcis-20320	145	13	and	and	CCONJ
fcis-20320	145	14	the	the	DET
fcis-20320	145	15	real	real	ADJ
fcis-20320	145	16	labels	label	NOUN
fcis-20320	145	17	,	,	PUNCT
fcis-20320	145	18	and	and	CCONJ
fcis-20320	145	19	guide	guide	VERB
fcis-20320	145	20	the	the	DET
fcis-20320	145	21	optimization	optimization	NOUN
fcis-20320	145	22	of	of	ADP
fcis-20320	145	23	the	the	DET
fcis-20320	145	24	model	model	NOUN
fcis-20320	145	25	.	.	PUNCT
fcis-20320	146	1	in	in	ADP
fcis-20320	146	2	addition	addition	NOUN
fcis-20320	146	3	,	,	PUNCT
fcis-20320	146	4	the	the	DET
fcis-20320	146	5	cross	cross	ADJ
fcis-20320	146	6	-	-	ADJ
fcis-20320	146	7	entropy	entropy	ADJ
fcis-20320	146	8	loss	loss	NOUN
fcis-20320	146	9	function	function	NOUN
fcis-20320	146	10	has	have	VERB
fcis-20320	146	11	good	good	ADJ
fcis-20320	146	12	mathematical	mathematical	ADJ
fcis-20320	146	13	properties	property	NOUN
fcis-20320	146	14	,	,	PUNCT
fcis-20320	146	15	making	make	VERB
fcis-20320	146	16	the	the	DET
fcis-20320	146	17	optimization	optimization	NOUN
fcis-20320	146	18	process	process	NOUN
fcis-20320	146	19	more	more	ADV
fcis-20320	146	20	stable	stable	ADJ
fcis-20320	146	21	and	and	CCONJ
fcis-20320	146	22	efficient	efficient	ADJ
fcis-20320	146	23	.	.	PUNCT
fcis-20320	147	1	therefore	therefore	ADV
fcis-20320	147	2	,	,	PUNCT
fcis-20320	147	3	the	the	DET
fcis-20320	147	4	cross	cross	ADJ
fcis-20320	147	5	-	-	ADJ
fcis-20320	147	6	entropy	entropy	ADJ
fcis-20320	147	7	loss	loss	NOUN
fcis-20320	147	8	function	function	NOUN
fcis-20320	147	9	is	be	AUX
fcis-20320	147	10	usually	usually	ADV
fcis-20320	147	11	used	use	VERB
fcis-20320	147	12	as	as	ADP
fcis-20320	147	13	the	the	DET
fcis-20320	147	14	optimization	optimization	NOUN
fcis-20320	147	15	objective	objective	NOUN
fcis-20320	147	16	in	in	ADP
fcis-20320	147	17	classification	classification	NOUN
fcis-20320	147	18	networks	network	NOUN
fcis-20320	147	19	.	.	PUNCT
fcis-20320	148	1	the	the	DET
fcis-20320	148	2	classification	classification	NOUN
fcis-20320	148	3	of	of	ADP
fcis-20320	148	4	benign	benign	ADJ
fcis-20320	148	5	and	and	CCONJ
fcis-20320	148	6	malignant	malignant	ADJ
fcis-20320	148	7	pulmonary	pulmonary	ADJ
fcis-20320	148	8	nodules	nodule	NOUN
fcis-20320	148	9	is	be	AUX
fcis-20320	148	10	a	a	DET
fcis-20320	148	11	binary	binary	ADJ
fcis-20320	148	12	classification	classification	NOUN
fcis-20320	148	13	problem	problem	NOUN
fcis-20320	148	14	.	.	PUNCT
fcis-20320	149	1	this	this	DET
fcis-20320	149	2	section	section	NOUN
fcis-20320	149	3	will	will	AUX
fcis-20320	149	4	use	use	VERB
fcis-20320	149	5	the	the	DET
fcis-20320	149	6	binary	binary	PROPN
fcis-20320	149	7	cross	cross	ADJ
fcis-20320	149	8	-	-	ADJ
fcis-20320	149	9	entropy	entropy	ADJ
fcis-20320	149	10	function	function	NOUN
fcis-20320	149	11	as	as	ADP
fcis-20320	149	12	the	the	DET
fcis-20320	149	13	loss	loss	NOUN
fcis-20320	149	14	function	function	NOUN
fcis-20320	149	15	.	.	PUNCT
fcis-20320	150	1	in	in	ADP
fcis-20320	150	2	this	this	DET
fcis-20320	150	3	problem	problem	NOUN
fcis-20320	150	4	,	,	PUNCT
fcis-20320	150	5	one	one	NUM
fcis-20320	150	6	probability	probability	NOUN
fcis-20320	150	7	distribution	distribution	NOUN
fcis-20320	150	8	represents	represent	VERB
fcis-20320	150	9	the	the	DET
fcis-20320	150	10	true	true	ADJ
fcis-20320	150	11	label	label	NOUN
fcis-20320	150	12	(	(	PUNCT
fcis-20320	150	13	i.e.	i.e.	X
fcis-20320	150	14	,	,	PUNCT
fcis-20320	150	15	whether	whether	SCONJ
fcis-20320	150	16	the	the	DET
fcis-20320	150	17	pulmonary	pulmonary	ADJ
fcis-20320	150	18	nodule	nodule	NOUN
fcis-20320	150	19	is	be	AUX
fcis-20320	150	20	benign	benign	ADJ
fcis-20320	150	21	or	or	CCONJ
fcis-20320	150	22	malignant	malignant	ADJ
fcis-20320	150	23	)	)	PUNCT
fcis-20320	150	24	,	,	PUNCT
fcis-20320	150	25	and	and	CCONJ
fcis-20320	150	26	the	the	DET
fcis-20320	150	27	other	other	ADJ
fcis-20320	150	28	probability	probability	NOUN
fcis-20320	150	29	distribution	distribution	NOUN
fcis-20320	150	30	represents	represent	VERB
fcis-20320	150	31	the	the	DET
fcis-20320	150	32	model	model	NOUN
fcis-20320	150	33	's	's	PART
fcis-20320	150	34	predictions	prediction	NOUN
fcis-20320	150	35	.	.	PUNCT
fcis-20320	151	1	3.4	3.4	NUM
fcis-20320	151	2	.	.	PUNCT
fcis-20320	152	1	data	datum	NOUN
fcis-20320	152	2	acquisition	acquisition	NOUN
fcis-20320	152	3	the	the	DET
fcis-20320	152	4	experimental	experimental	ADJ
fcis-20320	152	5	training	training	NOUN
fcis-20320	152	6	and	and	CCONJ
fcis-20320	152	7	testing	testing	NOUN
fcis-20320	152	8	data	datum	NOUN
fcis-20320	152	9	sets	set	NOUN
fcis-20320	152	10	in	in	ADP
fcis-20320	152	11	this	this	DET
fcis-20320	152	12	article	article	NOUN
fcis-20320	152	13	come	come	VERB
fcis-20320	152	14	from	from	ADP
fcis-20320	152	15	luna16	luna16	ADJ
fcis-20320	152	16	.	.	PUNCT
fcis-20320	153	1	luna16	luna16	PROPN
fcis-20320	153	2	is	be	AUX
fcis-20320	153	3	refined	refine	VERB
fcis-20320	153	4	from	from	ADP
fcis-20320	153	5	the	the	DET
fcis-20320	153	6	publicly	publicly	ADV
fcis-20320	153	7	available	available	ADJ
fcis-20320	153	8	lung	lung	NOUN
fcis-20320	153	9	image	image	NOUN
fcis-20320	153	10	database	database	NOUN
fcis-20320	153	11	consortium	consortium	NOUN
fcis-20320	153	12	image	image	NOUN
fcis-20320	153	13	collection	collection	NOUN
fcis-20320	153	14	(	(	PUNCT
fcis-20320	153	15	lidc	lidc	NOUN
fcis-20320	153	16	-	-	PUNCT
fcis-20320	153	17	idri	idri	NOUN
fcis-20320	153	18	)	)	PUNCT
fcis-20320	154	1	[	[	X
fcis-20320	154	2	18	18	NUM
fcis-20320	154	3	]	]	PUNCT
fcis-20320	154	4	.	.	PUNCT
fcis-20320	155	1	the	the	DET
fcis-20320	155	2	american	american	PROPN
fcis-20320	155	3	college	college	PROPN
fcis-20320	155	4	of	of	ADP
fcis-20320	155	5	radiology	radiology	NOUN
fcis-20320	155	6	recommends	recommend	VERB
fcis-20320	155	7	the	the	DET
fcis-20320	155	8	use	use	NOUN
fcis-20320	155	9	of	of	ADP
fcis-20320	155	10	thin	thin	ADJ
fcis-20320	155	11	-	-	PUNCT
fcis-20320	155	12	section	section	NOUN
fcis-20320	155	13	ct	ct	NUM
fcis-20320	155	14	scans	scan	NOUN
fcis-20320	155	15	for	for	ADP
fcis-20320	155	16	nodule	nodule	NOUN
fcis-20320	155	17	detection	detection	NOUN
fcis-20320	155	18	and	and	CCONJ
fcis-20320	155	19	classification	classification	NOUN
fcis-20320	155	20	.	.	PUNCT
fcis-20320	156	1	therefore	therefore	ADV
fcis-20320	156	2	,	,	PUNCT
fcis-20320	156	3	scans	scan	NOUN
fcis-20320	156	4	with	with	ADP
fcis-20320	156	5	slice	slice	NOUN
fcis-20320	156	6	thickness	thickness	NOUN
fcis-20320	156	7	greater	great	ADJ
fcis-20320	156	8	than	than	ADP
fcis-20320	156	9	3	3	NUM
fcis-20320	156	10	mm	mm	NOUN
fcis-20320	156	11	,	,	PUNCT
fcis-20320	156	12	inconsistent	inconsistent	ADJ
fcis-20320	156	13	slice	slice	NOUN
fcis-20320	156	14	spacing	spacing	NOUN
fcis-20320	156	15	,	,	PUNCT
fcis-20320	156	16	or	or	CCONJ
fcis-20320	156	17	missing	miss	VERB
fcis-20320	156	18	slices	slice	NOUN
fcis-20320	156	19	were	be	AUX
fcis-20320	156	20	discarded	discard	VERB
fcis-20320	156	21	.	.	PUNCT
fcis-20320	157	1	in	in	ADP
fcis-20320	157	2	addition	addition	NOUN
fcis-20320	157	3	,	,	PUNCT
fcis-20320	157	4	nodules	nodule	NOUN
fcis-20320	157	5	smaller	small	ADJ
fcis-20320	157	6	than	than	SCONJ
fcis-20320	157	7	3	3	NUM
fcis-20320	157	8	mm	mm	NOUN
fcis-20320	157	9	were	be	AUX
fcis-20320	157	10	removed	remove	VERB
fcis-20320	157	11	.	.	PUNCT
fcis-20320	158	1	after	after	ADP
fcis-20320	158	2	screening	screen	VERB
fcis-20320	158	3	,	,	PUNCT
fcis-20320	158	4	888	888	NUM
fcis-20320	158	5	ct	ct	PROPN
fcis-20320	158	6	scans	scan	NOUN
fcis-20320	158	7	and	and	CCONJ
fcis-20320	158	8	1004	1004	NUM
fcis-20320	158	9	nodules	nodule	NOUN
fcis-20320	158	10	were	be	AUX
fcis-20320	158	11	selected	select	VERB
fcis-20320	158	12	for	for	ADP
fcis-20320	158	13	training	training	NOUN
fcis-20320	158	14	and	and	CCONJ
fcis-20320	158	15	testing	testing	NOUN
fcis-20320	158	16	(	(	PUNCT
fcis-20320	158	17	450	450	NUM
fcis-20320	158	18	malignant	malignant	ADJ
fcis-20320	158	19	nodules	nodule	NOUN
fcis-20320	158	20	and	and	CCONJ
fcis-20320	158	21	554	554	NUM
fcis-20320	158	22	benign	benign	ADJ
fcis-20320	158	23	nodules	nodule	NOUN
fcis-20320	158	24	)	)	PUNCT
fcis-20320	158	25	.	.	PUNCT
fcis-20320	159	1	in	in	ADP
fcis-20320	159	2	lidc	lidc	ADJ
fcis-20320	159	3	-	-	PUNCT
fcis-20320	159	4	idri	idri	NOUN
fcis-20320	159	5	,	,	PUNCT
fcis-20320	159	6	each	each	DET
fcis-20320	159	7	ct	ct	NOUN
fcis-20320	159	8	scan	scan	NOUN
fcis-20320	159	9	is	be	AUX
fcis-20320	159	10	accompanied	accompany	VERB
fcis-20320	159	11	by	by	ADP
fcis-20320	159	12	an	an	DET
fcis-20320	159	13	annotated	annotate	VERB
fcis-20320	159	14	xml	xml	NOUN
fcis-20320	159	15	file	file	NOUN
fcis-20320	159	16	provided	provide	VERB
fcis-20320	159	17	by	by	ADP
fcis-20320	159	18	four	four	NUM
fcis-20320	159	19	experienced	experienced	ADJ
fcis-20320	159	20	radiologists	radiologist	NOUN
fcis-20320	159	21	[	[	X
fcis-20320	159	22	18	18	NUM
fcis-20320	159	23	]	]	PUNCT
fcis-20320	159	24	.	.	PUNCT
fcis-20320	160	1	annotations	annotation	NOUN
fcis-20320	160	2	in	in	ADP
fcis-20320	160	3	the	the	DET
fcis-20320	160	4	dataset	dataset	NOUN
fcis-20320	160	5	were	be	AUX
fcis-20320	160	6	collected	collect	VERB
fcis-20320	160	7	during	during	ADP
fcis-20320	160	8	a	a	DET
fcis-20320	160	9	twostage	twostage	NOUN
fcis-20320	160	10	image	image	NOUN
fcis-20320	160	11	annotation	annotation	NOUN
fcis-20320	160	12	process	process	NOUN
fcis-20320	160	13	by	by	ADP
fcis-20320	160	14	at	at	ADV
fcis-20320	160	15	least	least	ADJ
fcis-20320	160	16	3	3	NUM
fcis-20320	160	17	radiologists	radiologist	NOUN
fcis-20320	160	18	.	.	PUNCT
fcis-20320	161	1	it	it	PRON
fcis-20320	161	2	includes	include	VERB
fcis-20320	161	3	almost	almost	ADV
fcis-20320	161	4	all	all	PRON
fcis-20320	161	5	defining	define	VERB
fcis-20320	161	6	characteristics	characteristic	NOUN
fcis-20320	161	7	,	,	PUNCT
fcis-20320	161	8	such	such	ADJ
fcis-20320	161	9	as	as	ADP
fcis-20320	161	10	the	the	DET
fcis-20320	161	11	location	location	NOUN
fcis-20320	161	12	,	,	PUNCT
fcis-20320	161	13	size	size	NOUN
fcis-20320	161	14	,	,	PUNCT
fcis-20320	161	15	and	and	CCONJ
fcis-20320	161	16	malignancy	malignancy	NOUN
fcis-20320	161	17	of	of	ADP
fcis-20320	161	18	the	the	DET
fcis-20320	161	19	nodule	nodule	NOUN
fcis-20320	161	20	(	(	PUNCT
fcis-20320	161	21	from	from	ADP
fcis-20320	161	22	1	1	NUM
fcis-20320	161	23	to	to	ADP
fcis-20320	161	24	5	5	NUM
fcis-20320	161	25	)	)	PUNCT
fcis-20320	161	26	.	.	PUNCT
fcis-20320	162	1	level	level	NOUN
fcis-20320	162	2	1	1	NUM
fcis-20320	162	3	indicates	indicate	VERB
fcis-20320	162	4	a	a	DET
fcis-20320	162	5	very	very	ADV
fcis-20320	162	6	small	small	ADJ
fcis-20320	162	7	likelihood	likelihood	NOUN
fcis-20320	162	8	of	of	ADP
fcis-20320	162	9	cancer	cancer	NOUN
fcis-20320	162	10	,	,	PUNCT
fcis-20320	162	11	level	level	NOUN
fcis-20320	162	12	2	2	NUM
fcis-20320	162	13	indicates	indicate	VERB
fcis-20320	162	14	a	a	DET
fcis-20320	162	15	moderate	moderate	ADJ
fcis-20320	162	16	likelihood	likelihood	NOUN
fcis-20320	162	17	of	of	ADP
fcis-20320	162	18	cancer	cancer	NOUN
fcis-20320	162	19	,	,	PUNCT
fcis-20320	162	20	level	level	NOUN
fcis-20320	162	21	3	3	NUM
fcis-20320	162	22	indicates	indicate	VERB
fcis-20320	162	23	a	a	DET
fcis-20320	162	24	moderate	moderate	ADJ
fcis-20320	162	25	likelihood	likelihood	NOUN
fcis-20320	162	26	of	of	ADP
fcis-20320	162	27	cancer	cancer	NOUN
fcis-20320	162	28	,	,	PUNCT
fcis-20320	162	29	level	level	NOUN
fcis-20320	162	30	4	4	NUM
fcis-20320	162	31	indicates	indicate	VERB
fcis-20320	162	32	a	a	DET
fcis-20320	162	33	moderate	moderate	ADJ
fcis-20320	162	34	suspicion	suspicion	NOUN
fcis-20320	162	35	of	of	ADP
fcis-20320	162	36	cancer	cancer	NOUN
fcis-20320	162	37	,	,	PUNCT
fcis-20320	162	38	and	and	CCONJ
fcis-20320	162	39	level	level	NOUN
fcis-20320	162	40	5	5	NUM
fcis-20320	162	41	indicates	indicate	VERB
fcis-20320	162	42	a	a	DET
fcis-20320	162	43	very	very	ADV
fcis-20320	162	44	strong	strong	ADJ
fcis-20320	162	45	suspicion	suspicion	NOUN
fcis-20320	162	46	of	of	ADP
fcis-20320	162	47	cancer	cancer	NOUN
fcis-20320	162	48	.	.	PUNCT
fcis-20320	163	1	the	the	DET
fcis-20320	163	2	first	first	ADJ
fcis-20320	163	3	two	two	NUM
fcis-20320	163	4	grades	grade	NOUN
fcis-20320	163	5	are	be	AUX
fcis-20320	163	6	considered	consider	VERB
fcis-20320	163	7	benign	benign	ADJ
fcis-20320	163	8	and	and	CCONJ
fcis-20320	163	9	the	the	DET
fcis-20320	163	10	last	last	ADJ
fcis-20320	163	11	two	two	NUM
fcis-20320	163	12	are	be	AUX
fcis-20320	163	13	malignant	malignant	ADJ
fcis-20320	163	14	.	.	PUNCT
fcis-20320	164	1	for	for	ADP
fcis-20320	164	2	benign	benign	ADJ
fcis-20320	164	3	and	and	CCONJ
fcis-20320	164	4	malignant	malignant	ADJ
fcis-20320	164	5	classification	classification	NOUN
fcis-20320	164	6	of	of	ADP
fcis-20320	164	7	pulmonary	pulmonary	ADJ
fcis-20320	164	8	nodules	nodule	NOUN
fcis-20320	164	9	,	,	PUNCT
fcis-20320	164	10	binary	binary	ADJ
fcis-20320	164	11	classification	classification	NOUN
fcis-20320	164	12	was	be	AUX
fcis-20320	164	13	performed	perform	VERB
fcis-20320	164	14	by	by	ADP
fcis-20320	164	15	calculating	calculate	VERB
fcis-20320	164	16	the	the	DET
fcis-20320	164	17	average	average	ADJ
fcis-20320	164	18	grade	grade	NOUN
fcis-20320	164	19	of	of	ADP
fcis-20320	164	20	each	each	DET
fcis-20320	164	21	selected	select	VERB
fcis-20320	164	22	nodule	nodule	NOUN
fcis-20320	164	23	.	.	PUNCT
fcis-20320	165	1	if	if	SCONJ
fcis-20320	165	2	the	the	DET
fcis-20320	165	3	final	final	ADJ
fcis-20320	165	4	average	average	ADJ
fcis-20320	165	5	grade	grade	NOUN
fcis-20320	165	6	equals	equal	VERB
fcis-20320	165	7	3	3	NUM
fcis-20320	165	8	,	,	PUNCT
fcis-20320	165	9	then	then	ADV
fcis-20320	165	10	the	the	DET
fcis-20320	165	11	judgment	judgment	NOUN
fcis-20320	165	12	of	of	ADP
fcis-20320	165	13	benign	benign	ADJ
fcis-20320	165	14	or	or	CCONJ
fcis-20320	165	15	malignant	malignant	ADJ
fcis-20320	165	16	nodules	nodule	NOUN
fcis-20320	165	17	is	be	AUX
fcis-20320	165	18	considered	consider	VERB
fcis-20320	165	19	uncertain	uncertain	ADJ
fcis-20320	165	20	and	and	CCONJ
fcis-20320	165	21	is	be	AUX
fcis-20320	165	22	therefore	therefore	ADV
fcis-20320	165	23	eliminated	eliminate	VERB
fcis-20320	165	24	.	.	PUNCT
fcis-20320	166	1	for	for	ADP
fcis-20320	166	2	nodules	nodule	NOUN
fcis-20320	166	3	with	with	ADP
fcis-20320	166	4	a	a	DET
fcis-20320	166	5	final	final	ADJ
fcis-20320	166	6	mean	mean	NOUN
fcis-20320	166	7	grade	grade	NOUN
fcis-20320	166	8	greater	great	ADJ
fcis-20320	166	9	than	than	ADP
fcis-20320	166	10	3	3	NUM
fcis-20320	166	11	,	,	PUNCT
fcis-20320	166	12	they	they	PRON
fcis-20320	166	13	were	be	AUX
fcis-20320	166	14	labeled	label	VERB
fcis-20320	166	15	as	as	ADP
fcis-20320	166	16	malignant	malignant	ADJ
fcis-20320	166	17	nodules	nodule	NOUN
fcis-20320	166	18	[	[	X
fcis-20320	166	19	19	19	NUM
fcis-20320	166	20	]	]	PUNCT
fcis-20320	166	21	.	.	PUNCT
fcis-20320	167	1	otherwise	otherwise	ADV
fcis-20320	167	2	,	,	PUNCT
fcis-20320	167	3	they	they	PRON
fcis-20320	167	4	are	be	AUX
fcis-20320	167	5	considered	consider	VERB
fcis-20320	167	6	benign	benign	ADJ
fcis-20320	167	7	.	.	PUNCT
fcis-20320	168	1	all	all	DET
fcis-20320	168	2	annotations	annotation	NOUN
fcis-20320	168	3	were	be	AUX
fcis-20320	168	4	obtained	obtain	VERB
fcis-20320	168	5	from	from	ADP
fcis-20320	168	6	the	the	DET
fcis-20320	168	7	lidc	lidc	ADJ
fcis-20320	168	8	-	-	PUNCT
fcis-20320	168	9	idri	idri	ADJ
fcis-20320	168	10	database	database	NOUN
fcis-20320	168	11	.	.	PUNCT
fcis-20320	169	1	4	4	X
fcis-20320	169	2	.	.	NOUN
fcis-20320	169	3	experiments	experiment	NOUN
fcis-20320	169	4	and	and	CCONJ
fcis-20320	169	5	results	result	NOUN
fcis-20320	169	6	in	in	ADP
fcis-20320	169	7	this	this	DET
fcis-20320	169	8	section	section	NOUN
fcis-20320	169	9	,	,	PUNCT
fcis-20320	169	10	the	the	DET
fcis-20320	169	11	experimental	experimental	ADJ
fcis-20320	169	12	environment	environment	NOUN
fcis-20320	169	13	and	and	CCONJ
fcis-20320	169	14	algorithm	algorithm	NOUN
fcis-20320	169	15	evaluation	evaluation	NOUN
fcis-20320	169	16	indicators	indicator	NOUN
fcis-20320	169	17	are	be	AUX
fcis-20320	169	18	first	first	ADV
fcis-20320	169	19	introduced	introduce	VERB
fcis-20320	169	20	.	.	PUNCT
fcis-20320	170	1	then	then	ADV
fcis-20320	170	2	the	the	DET
fcis-20320	170	3	model	model	NOUN
fcis-20320	170	4	was	be	AUX
fcis-20320	170	5	trained	train	VERB
fcis-20320	170	6	on	on	ADP
fcis-20320	170	7	the	the	DET
fcis-20320	170	8	luna16	luna16	ADJ
fcis-20320	170	9	data	datum	NOUN
fcis-20320	170	10	set	set	NOUN
fcis-20320	170	11	,	,	PUNCT
fcis-20320	170	12	and	and	CCONJ
fcis-20320	170	13	the	the	DET
fcis-20320	170	14	proposed	propose	VERB
fcis-20320	170	15	benign	benign	ADJ
fcis-20320	170	16	and	and	CCONJ
fcis-20320	170	17	malignant	malignant	ADJ
fcis-20320	170	18	pulmonary	pulmonary	ADJ
fcis-20320	170	19	nodule	nodule	NOUN
fcis-20320	170	20	classification	classification	NOUN
fcis-20320	170	21	method	method	NOUN
fcis-20320	170	22	was	be	AUX
fcis-20320	170	23	compared	compare	VERB
fcis-20320	170	24	with	with	ADP
fcis-20320	170	25	the	the	DET
fcis-20320	170	26	current	current	ADJ
fcis-20320	170	27	state	state	NOUN
fcis-20320	170	28	-	-	PUNCT
fcis-20320	170	29	of	of	ADP
fcis-20320	170	30	-	-	PUNCT
fcis-20320	170	31	the	the	DET
fcis-20320	170	32	-	-	PUNCT
fcis-20320	170	33	art	art	NOUN
fcis-20320	170	34	methods	method	NOUN
fcis-20320	170	35	to	to	PART
fcis-20320	170	36	verify	verify	VERB
fcis-20320	170	37	the	the	DET
fcis-20320	170	38	effectiveness	effectiveness	NOUN
fcis-20320	170	39	of	of	ADP
fcis-20320	170	40	the	the	DET
fcis-20320	170	41	proposed	propose	VERB
fcis-20320	170	42	method	method	NOUN
fcis-20320	170	43	.	.	PUNCT
fcis-20320	171	1	4.1	4.1	NUM
fcis-20320	171	2	.	.	PUNCT
fcis-20320	171	3	experiment	experiment	NOUN
fcis-20320	171	4	setup	setup	NOUN
fcis-20320	171	5	this	this	DET
fcis-20320	171	6	experiment	experiment	NOUN
fcis-20320	171	7	configured	configure	VERB
fcis-20320	171	8	python	python	NOUN
fcis-20320	171	9	for	for	ADP
fcis-20320	171	10	training	training	NOUN
fcis-20320	171	11	and	and	CCONJ
fcis-20320	171	12	testing	testing	NOUN
fcis-20320	171	13	based	base	VERB
fcis-20320	171	14	on	on	ADP
fcis-20320	171	15	the	the	DET
fcis-20320	171	16	anaconda	anaconda	NOUN
fcis-20320	171	17	environment	environment	NOUN
fcis-20320	171	18	on	on	ADP
fcis-20320	171	19	the	the	DET
fcis-20320	171	20	windows	window	NOUN
fcis-20320	171	21	10	10	NUM
fcis-20320	171	22	system	system	NOUN
fcis-20320	171	23	,	,	PUNCT
fcis-20320	171	24	and	and	CCONJ
fcis-20320	171	25	used	use	VERB
fcis-20320	171	26	the	the	DET
fcis-20320	171	27	pytorch	pytorch	NOUN
fcis-20320	171	28	deep	deep	ADJ
fcis-20320	171	29	learning	learning	NOUN
fcis-20320	171	30	framework	framework	NOUN
fcis-20320	171	31	and	and	CCONJ
fcis-20320	171	32	cuda	cuda	NOUN
fcis-20320	171	33	for	for	ADP
fcis-20320	171	34	gpu	gpu	PROPN
fcis-20320	171	35	accelerated	accelerated	ADJ
fcis-20320	171	36	training	training	NOUN
fcis-20320	171	37	and	and	CCONJ
fcis-20320	171	38	verification	verification	NOUN
fcis-20320	171	39	of	of	ADP
fcis-20320	171	40	model	model	NOUN
fcis-20320	171	41	feasibility	feasibility	NOUN
fcis-20320	171	42	.	.	PUNCT
fcis-20320	172	1	pytorch	pytorch	NOUN
fcis-20320	172	2	provides	provide	VERB
fcis-20320	172	3	many	many	ADJ
fcis-20320	172	4	extension	extension	NOUN
fcis-20320	172	5	libraries	library	NOUN
fcis-20320	172	6	and	and	CCONJ
fcis-20320	172	7	interfaces	interface	NOUN
fcis-20320	172	8	for	for	ADP
fcis-20320	172	9	easy	easy	ADJ
fcis-20320	172	10	modular	modular	ADJ
fcis-20320	172	11	design	design	NOUN
fcis-20320	172	12	,	,	PUNCT
fcis-20320	172	13	network	network	NOUN
fcis-20320	172	14	construction	construction	NOUN
fcis-20320	172	15	,	,	PUNCT
fcis-20320	172	16	and	and	CCONJ
fcis-20320	172	17	customization	customization	NOUN
fcis-20320	172	18	.	.	PUNCT
fcis-20320	173	1	in	in	ADP
fcis-20320	173	2	addition	addition	NOUN
fcis-20320	173	3	,	,	PUNCT
fcis-20320	173	4	pytorch	pytorch	NOUN
fcis-20320	173	5	supports	support	VERB
fcis-20320	173	6	efficient	efficient	ADJ
fcis-20320	173	7	parallel	parallel	ADJ
fcis-20320	173	8	computing	computing	NOUN
fcis-20320	173	9	on	on	ADP
fcis-20320	173	10	cpu	cpu	NOUN
fcis-20320	173	11	and	and	CCONJ
fcis-20320	173	12	gpu	gpu	NOUN
fcis-20320	173	13	,	,	PUNCT
fcis-20320	173	14	and	and	CCONJ
fcis-20320	173	15	also	also	ADV
fcis-20320	173	16	supports	support	VERB
fcis-20320	173	17	multimachine	multimachine	ADJ
fcis-20320	173	18	distributed	distribute	VERB
fcis-20320	173	19	computing	computing	NOUN
fcis-20320	173	20	,	,	PUNCT
fcis-20320	173	21	which	which	PRON
fcis-20320	173	22	can	can	AUX
fcis-20320	173	23	greatly	greatly	ADV
fcis-20320	173	24	improve	improve	VERB
fcis-20320	173	25	training	training	NOUN
fcis-20320	173	26	speed	speed	NOUN
fcis-20320	173	27	and	and	CCONJ
fcis-20320	173	28	model	model	NOUN
fcis-20320	173	29	performance	performance	NOUN
fcis-20320	173	30	.	.	PUNCT
fcis-20320	174	1	4.2	4.2	NUM
fcis-20320	174	2	.	.	PUNCT
fcis-20320	175	1	algorithm	algorithm	NOUN
fcis-20320	175	2	evaluation	evaluation	NOUN
fcis-20320	175	3	metrics	metric	NOUN
fcis-20320	175	4	in	in	ADP
fcis-20320	175	5	order	order	NOUN
fcis-20320	175	6	to	to	PART
fcis-20320	175	7	evaluate	evaluate	VERB
fcis-20320	175	8	the	the	DET
fcis-20320	175	9	performance	performance	NOUN
fcis-20320	175	10	of	of	ADP
fcis-20320	175	11	the	the	DET
fcis-20320	175	12	algorithm	algorithm	NOUN
fcis-20320	175	13	,	,	PUNCT
fcis-20320	175	14	accuracy	accuracy	NOUN
fcis-20320	175	15	,	,	PUNCT
fcis-20320	175	16	sensitivity	sensitivity	NOUN
fcis-20320	175	17	,	,	PUNCT
fcis-20320	175	18	specificity	specificity	NOUN
fcis-20320	175	19	,	,	PUNCT
fcis-20320	175	20	and	and	CCONJ
fcis-20320	175	21	auc	auc	NOUN
fcis-20320	175	22	indicators	indicator	NOUN
fcis-20320	175	23	are	be	AUX
fcis-20320	175	24	used	use	VERB
fcis-20320	175	25	in	in	ADP
fcis-20320	175	26	the	the	DET
fcis-20320	175	27	experiments	experiment	NOUN
fcis-20320	175	28	.	.	PUNCT
fcis-20320	176	1	accuracy	accuracy	NOUN
fcis-20320	176	2	represents	represent	VERB
fcis-20320	176	3	the	the	DET
fcis-20320	176	4	proportion	proportion	NOUN
fcis-20320	176	5	of	of	ADP
fcis-20320	176	6	correctly	correctly	ADV
fcis-20320	176	7	predicted	predict	VERB
fcis-20320	176	8	results	result	NOUN
fcis-20320	176	9	in	in	ADP
fcis-20320	176	10	the	the	DET
fcis-20320	176	11	total	total	ADJ
fcis-20320	176	12	sample	sample	NOUN
fcis-20320	176	13	.	.	PUNCT
fcis-20320	177	1	sensitivity	sensitivity	NOUN
fcis-20320	177	2	and	and	CCONJ
fcis-20320	177	3	specificity	specificity	NOUN
fcis-20320	177	4	measure	measure	VERB
fcis-20320	177	5	the	the	DET
fcis-20320	177	6	proportion	proportion	NOUN
fcis-20320	177	7	of	of	ADP
fcis-20320	177	8	correctly	correctly	ADV
fcis-20320	177	9	identified	identify	VERB
fcis-20320	177	10	malignant	malignant	ADJ
fcis-20320	177	11	nodules	nodule	NOUN
fcis-20320	177	12	and	and	CCONJ
fcis-20320	177	13	benign	benign	ADJ
fcis-20320	177	14	nodules	nodule	NOUN
fcis-20320	177	15	respectively	respectively	ADV
fcis-20320	177	16	.	.	PUNCT
fcis-20320	178	1	auc	auc	PROPN
fcis-20320	178	2	represents	represent	VERB
fcis-20320	178	3	the	the	DET
fcis-20320	178	4	area	area	NOUN
fcis-20320	178	5	under	under	ADP
fcis-20320	178	6	the	the	DET
fcis-20320	178	7	roc	roc	PROPN
fcis-20320	178	8	curve	curve	NOUN
fcis-20320	178	9	.	.	PUNCT
fcis-20320	179	1	the	the	PRON
fcis-20320	179	2	larger	large	ADJ
fcis-20320	179	3	the	the	DET
fcis-20320	179	4	auc	auc	NOUN
fcis-20320	179	5	value	value	NOUN
fcis-20320	179	6	,	,	PUNCT
fcis-20320	179	7	the	the	PRON
fcis-20320	179	8	better	well	ADJ
fcis-20320	179	9	the	the	DET
fcis-20320	179	10	classifier	classifier	ADJ
fcis-20320	179	11	effect	effect	NOUN
fcis-20320	179	12	.	.	PUNCT
fcis-20320	180	1	roc	roc	PROPN
fcis-20320	180	2	is	be	AUX
fcis-20320	180	3	a	a	DET
fcis-20320	180	4	curve	curve	NOUN
fcis-20320	180	5	obtained	obtain	VERB
fcis-20320	180	6	by	by	ADP
fcis-20320	180	7	using	use	VERB
fcis-20320	180	8	fpr	fpr	PRON
fcis-20320	180	9	(	(	PUNCT
fcis-20320	180	10	false	false	ADJ
fcis-20320	180	11	positive	positive	ADJ
fcis-20320	180	12	rate	rate	NOUN
fcis-20320	180	13	)	)	PUNCT
fcis-20320	180	14	as	as	ADP
fcis-20320	180	15	the	the	DET
fcis-20320	180	16	abscissa	abscissa	NOUN
fcis-20320	180	17	and	and	CCONJ
fcis-20320	180	18	tpr	tpr	NOUN
fcis-20320	180	19	(	(	PUNCT
fcis-20320	180	20	true	true	ADJ
fcis-20320	180	21	positive	positive	ADJ
fcis-20320	180	22	rate	rate	NOUN
fcis-20320	180	23	)	)	PUNCT
fcis-20320	180	24	as	as	ADP
fcis-20320	180	25	the	the	DET
fcis-20320	180	26	ordinate	ordinate	NOUN
fcis-20320	180	27	.	.	PUNCT
fcis-20320	181	1	tpr	tpr	PROPN
fcis-20320	181	2	is	be	AUX
fcis-20320	181	3	sensitivity	sensitivity	NOUN
fcis-20320	181	4	.	.	PUNCT
fcis-20320	182	1	the	the	DET
fcis-20320	182	2	calculation	calculation	NOUN
fcis-20320	182	3	formulas	formula	VERB
fcis-20320	182	4	for	for	ADP
fcis-20320	182	5	accuracy	accuracy	NOUN
fcis-20320	182	6	,	,	PUNCT
fcis-20320	182	7	sensitivity	sensitivity	NOUN
fcis-20320	182	8	,	,	PUNCT
fcis-20320	182	9	and	and	CCONJ
fcis-20320	182	10	specificity	specificity	NOUN
fcis-20320	182	11	are	be	AUX
fcis-20320	182	12	as	as	SCONJ
fcis-20320	182	13	follows	follow	VERB
fcis-20320	182	14	:	:	PUNCT
fcis-20320	183	1			PROPN
fcis-20320	183	2	tp	tp	NOUN
fcis-20320	183	3	tn	tn	NOUN
fcis-20320	183	4	accuracy	accuracy	NOUN
fcis-20320	183	5	tp	tp	X
fcis-20320	183	6	fn	fn	PROPN
fcis-20320	184	1	fp	fp	INTJ
fcis-20320	184	2	fn	fn	PROPN
fcis-20320	184	3			PROPN
fcis-20320	184	4			PROPN
fcis-20320	184	5			PROPN
fcis-20320	184	6			PUNCT
fcis-20320	184	7			X
fcis-20320	184	8	(	(	PUNCT
fcis-20320	184	9	5	5	X
fcis-20320	184	10	)	)	PUNCT
fcis-20320	184	11	tp	tp	ADP
fcis-20320	184	12	sensitivity	sensitivity	NOUN
fcis-20320	184	13	tpr	tpr	PROPN
fcis-20320	185	1	tp	tp	ADP
fcis-20320	185	2	fn	fn	PROPN
fcis-20320	186	1			PROPN
fcis-20320	186	2			PROPN
fcis-20320	186	3			PUNCT
fcis-20320	186	4	(	(	PUNCT
fcis-20320	186	5	6	6	NUM
fcis-20320	186	6	)	)	SYM
fcis-20320	186	7	10	10	NUM
fcis-20320	186	8			NOUN
fcis-20320	186	9			PROPN
fcis-20320	186	10	tn	tn	NOUN
fcis-20320	186	11	specificity	specificity	NOUN
fcis-20320	186	12	tnr	tnr	PROPN
fcis-20320	186	13	tn	tn	PROPN
fcis-20320	186	14	fp	fp	PROPN
fcis-20320	186	15			NUM
fcis-20320	186	16			PROPN
fcis-20320	186	17			PUNCT
fcis-20320	186	18	(	(	PUNCT
fcis-20320	186	19	7	7	NUM
fcis-20320	186	20	)	)	PUNCT
fcis-20320	186	21	among	among	ADP
fcis-20320	186	22	them	they	PRON
fcis-20320	186	23	,	,	PUNCT
fcis-20320	186	24	tp	tp	NOUN
fcis-20320	186	25	represents	represent	VERB
fcis-20320	186	26	a	a	DET
fcis-20320	186	27	positive	positive	ADJ
fcis-20320	186	28	sample	sample	NOUN
fcis-20320	186	29	predicted	predict	VERB
fcis-20320	186	30	as	as	ADP
fcis-20320	186	31	positive	positive	ADJ
fcis-20320	186	32	by	by	ADP
fcis-20320	186	33	the	the	DET
fcis-20320	186	34	model	model	NOUN
fcis-20320	186	35	,	,	PUNCT
fcis-20320	186	36	the	the	DET
fcis-20320	186	37	prediction	prediction	NOUN
fcis-20320	186	38	is	be	AUX
fcis-20320	186	39	1	1	NUM
fcis-20320	186	40	and	and	CCONJ
fcis-20320	186	41	the	the	DET
fcis-20320	186	42	prediction	prediction	NOUN
fcis-20320	186	43	is	be	AUX
fcis-20320	186	44	correct	correct	ADJ
fcis-20320	186	45	,	,	PUNCT
fcis-20320	186	46	that	that	ADV
fcis-20320	186	47	is	is	ADV
fcis-20320	186	48	,	,	PUNCT
fcis-20320	186	49	the	the	DET
fcis-20320	186	50	actual	actual	ADJ
fcis-20320	186	51	value	value	NOUN
fcis-20320	186	52	is	be	AUX
fcis-20320	186	53	1	1	NUM
fcis-20320	186	54	.	.	PUNCT
fcis-20320	187	1	fp	fp	PROPN
fcis-20320	187	2	represents	represent	VERB
fcis-20320	187	3	a	a	DET
fcis-20320	187	4	negative	negative	ADJ
fcis-20320	187	5	sample	sample	NOUN
fcis-20320	187	6	predicted	predict	VERB
fcis-20320	187	7	as	as	ADP
fcis-20320	187	8	positive	positive	ADJ
fcis-20320	187	9	by	by	ADP
fcis-20320	187	10	the	the	DET
fcis-20320	187	11	model	model	NOUN
fcis-20320	187	12	,	,	PUNCT
fcis-20320	187	13	the	the	DET
fcis-20320	187	14	prediction	prediction	NOUN
fcis-20320	187	15	is	be	AUX
fcis-20320	187	16	1	1	NUM
fcis-20320	187	17	and	and	CCONJ
fcis-20320	187	18	the	the	DET
fcis-20320	187	19	prediction	prediction	NOUN
fcis-20320	187	20	is	be	AUX
fcis-20320	187	21	wrong	wrong	ADJ
fcis-20320	187	22	,	,	PUNCT
fcis-20320	187	23	that	that	PRON
fcis-20320	187	24	is	be	AUX
fcis-20320	187	25	the	the	DET
fcis-20320	187	26	actual	actual	ADJ
fcis-20320	187	27	value	value	NOUN
fcis-20320	187	28	is	be	AUX
fcis-20320	187	29	0	0	NUM
fcis-20320	187	30	.	.	PUNCT
fcis-20320	188	1	fn	fn	PROPN
fcis-20320	188	2	represents	represent	VERB
fcis-20320	188	3	the	the	DET
fcis-20320	188	4	positive	positive	ADJ
fcis-20320	188	5	sample	sample	NOUN
fcis-20320	188	6	predicted	predict	VERB
fcis-20320	188	7	as	as	ADP
fcis-20320	188	8	negative	negative	ADJ
fcis-20320	188	9	by	by	ADP
fcis-20320	188	10	the	the	DET
fcis-20320	188	11	model	model	NOUN
fcis-20320	188	12	,	,	PUNCT
fcis-20320	188	13	the	the	DET
fcis-20320	188	14	prediction	prediction	NOUN
fcis-20320	188	15	is	be	AUX
fcis-20320	188	16	0	0	NUM
fcis-20320	188	17	and	and	CCONJ
fcis-20320	188	18	the	the	DET
fcis-20320	188	19	prediction	prediction	NOUN
fcis-20320	188	20	is	be	AUX
fcis-20320	188	21	wrong	wrong	ADJ
fcis-20320	188	22	,	,	PUNCT
fcis-20320	188	23	that	that	ADV
fcis-20320	188	24	is	is	ADV
fcis-20320	188	25	,	,	PUNCT
fcis-20320	188	26	the	the	DET
fcis-20320	188	27	actual	actual	ADJ
fcis-20320	188	28	value	value	NOUN
fcis-20320	188	29	is	be	AUX
fcis-20320	188	30	1	1	NUM
fcis-20320	188	31	.	.	X
fcis-20320	188	32	tn	tn	PROPN
fcis-20320	188	33	represents	represent	VERB
fcis-20320	188	34	the	the	DET
fcis-20320	188	35	negative	negative	ADJ
fcis-20320	188	36	sample	sample	NOUN
fcis-20320	188	37	predicted	predict	VERB
fcis-20320	188	38	as	as	ADP
fcis-20320	188	39	negative	negative	ADJ
fcis-20320	188	40	by	by	ADP
fcis-20320	188	41	the	the	DET
fcis-20320	188	42	model	model	NOUN
fcis-20320	188	43	,	,	PUNCT
fcis-20320	188	44	the	the	DET
fcis-20320	188	45	prediction	prediction	NOUN
fcis-20320	188	46	is	be	AUX
fcis-20320	188	47	0	0	NUM
fcis-20320	188	48	,	,	PUNCT
fcis-20320	188	49	and	and	CCONJ
fcis-20320	188	50	the	the	DET
fcis-20320	188	51	prediction	prediction	NOUN
fcis-20320	188	52	correct	correct	ADJ
fcis-20320	188	53	,	,	PUNCT
fcis-20320	188	54	that	that	ADV
fcis-20320	188	55	is	is	ADV
fcis-20320	188	56	,	,	PUNCT
fcis-20320	188	57	it	it	PRON
fcis-20320	188	58	is	be	AUX
fcis-20320	188	59	0	0	NUM
fcis-20320	188	60	.	.	PUNCT
fcis-20320	189	1	there	there	PRON
fcis-20320	189	2	are	be	VERB
fcis-20320	189	3	two	two	NUM
fcis-20320	189	4	main	main	ADJ
fcis-20320	189	5	methods	method	NOUN
fcis-20320	189	6	for	for	ADP
fcis-20320	189	7	calculating	calculate	VERB
fcis-20320	189	8	auc	auc	NOUN
fcis-20320	189	9	.	.	PUNCT
fcis-20320	190	1	one	one	NUM
fcis-20320	190	2	is	be	AUX
fcis-20320	190	3	to	to	PART
fcis-20320	190	4	directly	directly	ADV
fcis-20320	190	5	calculate	calculate	VERB
fcis-20320	190	6	the	the	DET
fcis-20320	190	7	area	area	NOUN
fcis-20320	190	8	under	under	ADP
fcis-20320	190	9	the	the	DET
fcis-20320	190	10	roc	roc	PROPN
fcis-20320	190	11	curve	curve	NOUN
fcis-20320	190	12	,	,	PUNCT
fcis-20320	190	13	which	which	PRON
fcis-20320	190	14	is	be	AUX
fcis-20320	190	15	obtained	obtain	VERB
fcis-20320	190	16	by	by	ADP
fcis-20320	190	17	approximately	approximately	ADV
fcis-20320	190	18	calculating	calculate	VERB
fcis-20320	190	19	the	the	DET
fcis-20320	190	20	area	area	NOUN
fcis-20320	190	21	of	of	ADP
fcis-20320	190	22	each	each	DET
fcis-20320	190	23	small	small	ADJ
fcis-20320	190	24	trapezoid	trapezoid	NOUN
fcis-20320	190	25	under	under	ADP
fcis-20320	190	26	the	the	DET
fcis-20320	190	27	roc	roc	PROPN
fcis-20320	190	28	curve	curve	NOUN
fcis-20320	190	29	;	;	PUNCT
fcis-20320	190	30	the	the	DET
fcis-20320	190	31	other	other	ADJ
fcis-20320	190	32	is	be	AUX
fcis-20320	190	33	to	to	PART
fcis-20320	190	34	count	count	VERB
fcis-20320	190	35	the	the	DET
fcis-20320	190	36	proportion	proportion	NOUN
fcis-20320	190	37	of	of	ADP
fcis-20320	190	38	positive	positive	ADJ
fcis-20320	190	39	and	and	CCONJ
fcis-20320	190	40	negative	negative	ADJ
fcis-20320	190	41	sample	sample	NOUN
fcis-20320	190	42	pairs	pair	NOUN
fcis-20320	190	43	.	.	PUNCT
fcis-20320	191	1	the	the	DET
fcis-20320	191	2	idea	idea	NOUN
fcis-20320	191	3	of	of	ADP
fcis-20320	191	4	this	this	DET
fcis-20320	191	5	method	method	NOUN
fcis-20320	191	6	is	be	AUX
fcis-20320	191	7	to	to	PART
fcis-20320	191	8	randomly	randomly	VERB
fcis-20320	191	9	select	select	VERB
fcis-20320	191	10	one	one	NUM
fcis-20320	191	11	sample	sample	NOUN
fcis-20320	191	12	from	from	ADP
fcis-20320	191	13	each	each	PRON
fcis-20320	191	14	of	of	ADP
fcis-20320	191	15	the	the	DET
fcis-20320	191	16	positive	positive	ADJ
fcis-20320	191	17	and	and	CCONJ
fcis-20320	191	18	negative	negative	ADJ
fcis-20320	191	19	samples	sample	NOUN
fcis-20320	191	20	,	,	PUNCT
fcis-20320	191	21	and	and	CCONJ
fcis-20320	191	22	calculate	calculate	VERB
fcis-20320	191	23	the	the	DET
fcis-20320	191	24	probability	probability	NOUN
fcis-20320	191	25	that	that	SCONJ
fcis-20320	191	26	the	the	DET
fcis-20320	191	27	predicted	predict	VERB
fcis-20320	191	28	value	value	NOUN
fcis-20320	191	29	of	of	ADP
fcis-20320	191	30	the	the	DET
fcis-20320	191	31	positive	positive	ADJ
fcis-20320	191	32	sample	sample	NOUN
fcis-20320	191	33	is	be	AUX
fcis-20320	191	34	greater	great	ADJ
fcis-20320	191	35	than	than	ADP
fcis-20320	191	36	the	the	DET
fcis-20320	191	37	negative	negative	ADJ
fcis-20320	191	38	sample	sample	NOUN
fcis-20320	191	39	.	.	PUNCT
fcis-20320	192	1	the	the	DET
fcis-20320	192	2	computational	computational	ADJ
fcis-20320	192	3	complexity	complexity	NOUN
fcis-20320	192	4	of	of	ADP
fcis-20320	192	5	these	these	DET
fcis-20320	192	6	two	two	NUM
fcis-20320	192	7	methods	method	NOUN
fcis-20320	192	8	differs	differ	NOUN
fcis-20320	192	9	.	.	PUNCT
fcis-20320	193	1	the	the	DET
fcis-20320	193	2	time	time	NOUN
fcis-20320	193	3	complexity	complexity	NOUN
fcis-20320	193	4	of	of	ADP
fcis-20320	193	5	directly	directly	ADV
fcis-20320	193	6	calculating	calculate	VERB
fcis-20320	193	7	the	the	DET
fcis-20320	193	8	area	area	NOUN
fcis-20320	193	9	under	under	ADP
fcis-20320	193	10	the	the	DET
fcis-20320	193	11	roc	roc	PROPN
fcis-20320	193	12	curve	curve	NOUN
fcis-20320	193	13	is	be	AUX
fcis-20320	193	14			NOUN
fcis-20320	193	15	o	o	PROPN
fcis-20320	193	16	p	p	NOUN
fcis-20320	193	17	n	n	NOUN
fcis-20320	193	18	,	,	PUNCT
fcis-20320	193	19	while	while	SCONJ
fcis-20320	193	20	the	the	DET
fcis-20320	193	21	time	time	NOUN
fcis-20320	193	22	complexity	complexity	NOUN
fcis-20320	193	23	of	of	ADP
fcis-20320	193	24	counting	count	VERB
fcis-20320	193	25	the	the	DET
fcis-20320	193	26	proportion	proportion	NOUN
fcis-20320	193	27	of	of	ADP
fcis-20320	193	28	positive	positive	ADJ
fcis-20320	193	29	and	and	CCONJ
fcis-20320	193	30	negative	negative	ADJ
fcis-20320	193	31	sample	sample	NOUN
fcis-20320	193	32	pairs	pair	NOUN
fcis-20320	193	33	is	be	AUX
fcis-20320	193	34			NOUN
fcis-20320	193	35			NUM
fcis-20320	193	36			NOUN
fcis-20320	193	37			NOUN
fcis-20320	194	1	logo	logo	PROPN
fcis-20320	194	2	p	p	X
fcis-20320	195	1	n	n	PRON
fcis-20320	195	2	p	p	ADJ
fcis-20320	195	3	n	n	PROPN
fcis-20320	195	4			PROPN
fcis-20320	195	5			ADV
fcis-20320	195	6	,	,	PUNCT
fcis-20320	195	7	where	where	SCONJ
fcis-20320	195	8	p	p	NOUN
fcis-20320	195	9	and	and	CCONJ
fcis-20320	195	10	n	n	NOUN
fcis-20320	195	11	are	be	AUX
fcis-20320	195	12	the	the	DET
fcis-20320	195	13	numbers	number	NOUN
fcis-20320	195	14	of	of	ADP
fcis-20320	195	15	positive	positive	ADJ
fcis-20320	195	16	and	and	CCONJ
fcis-20320	195	17	negative	negative	ADJ
fcis-20320	195	18	samples	sample	NOUN
fcis-20320	195	19	respectively	respectively	ADV
fcis-20320	195	20	.	.	PUNCT
fcis-20320	196	1	this	this	DET
fcis-20320	196	2	section	section	NOUN
fcis-20320	196	3	mainly	mainly	ADV
fcis-20320	196	4	uses	use	VERB
fcis-20320	196	5	the	the	DET
fcis-20320	196	6	second	second	ADJ
fcis-20320	196	7	method	method	NOUN
fcis-20320	196	8	to	to	PART
fcis-20320	196	9	calculate	calculate	VERB
fcis-20320	196	10	auc	auc	NOUN
fcis-20320	196	11	:	:	PUNCT
fcis-20320	196	12	first	first	ADV
fcis-20320	196	13	,	,	PUNCT
fcis-20320	196	14	sort	sort	ADV
fcis-20320	196	15	all	all	DET
fcis-20320	196	16	samples	sample	NOUN
fcis-20320	196	17	from	from	ADP
fcis-20320	196	18	low	low	ADJ
fcis-20320	196	19	to	to	ADP
fcis-20320	196	20	high	high	ADJ
fcis-20320	196	21	according	accord	VERB
fcis-20320	196	22	to	to	ADP
fcis-20320	196	23	the	the	DET
fcis-20320	196	24	predicted	predict	VERB
fcis-20320	196	25	value	value	NOUN
fcis-20320	196	26	.	.	PUNCT
fcis-20320	197	1	for	for	ADP
fcis-20320	197	2	each	each	DET
fcis-20320	197	3	positive	positive	ADJ
fcis-20320	197	4	sample	sample	NOUN
fcis-20320	197	5	,	,	PUNCT
fcis-20320	197	6	calculate	calculate	VERB
fcis-20320	197	7	its	its	PRON
fcis-20320	197	8	position	position	NOUN
fcis-20320	197	9	in	in	ADP
fcis-20320	197	10	the	the	DET
fcis-20320	197	11	sorting	sorting	NOUN
fcis-20320	197	12	,	,	PUNCT
fcis-20320	197	13	recorded	record	VERB
fcis-20320	197	14	as	as	ADP
fcis-20320	197	15	ir	ir	PROPN
fcis-20320	197	16	.	.	PUNCT
fcis-20320	198	1	then	then	ADV
fcis-20320	198	2	calculate	calculate	VERB
fcis-20320	198	3	the	the	DET
fcis-20320	198	4	position	position	NOUN
fcis-20320	198	5	sum	sum	NOUN
fcis-20320	198	6	of	of	ADP
fcis-20320	198	7	all	all	DET
fcis-20320	198	8	positive	positive	ADJ
fcis-20320	198	9	samples	sample	NOUN
fcis-20320	198	10	,	,	PUNCT
fcis-20320	198	11	denoted	denote	VERB
fcis-20320	198	12	as	as	ADP
fcis-20320	198	13	ir	ir	NOUN
fcis-20320	198	14	.	.	PUNCT
fcis-20320	199	1	calculate	calculate	VERB
fcis-20320	199	2	the	the	DET
fcis-20320	199	3	number	number	NOUN
fcis-20320	199	4	of	of	ADP
fcis-20320	199	5	all	all	DET
fcis-20320	199	6	positive	positive	ADJ
fcis-20320	199	7	samples	sample	NOUN
fcis-20320	199	8	,	,	PUNCT
fcis-20320	199	9	denoted	denote	VERB
fcis-20320	199	10	as	as	ADP
fcis-20320	199	11	p	p	NOUN
fcis-20320	199	12	,	,	PUNCT
fcis-20320	199	13	and	and	CCONJ
fcis-20320	199	14	the	the	DET
fcis-20320	199	15	number	number	NOUN
fcis-20320	199	16	of	of	ADP
fcis-20320	199	17	negative	negative	ADJ
fcis-20320	199	18	samples	sample	NOUN
fcis-20320	199	19	,	,	PUNCT
fcis-20320	199	20	denoted	denote	VERB
fcis-20320	199	21	as	as	ADP
fcis-20320	199	22	n	n	X
fcis-20320	199	23	.	.	PUNCT
fcis-20320	200	1	the	the	DET
fcis-20320	200	2	final	final	ADJ
fcis-20320	200	3	auc	auc	NOUN
fcis-20320	200	4	calculation	calculation	NOUN
fcis-20320	200	5	formula	formula	NOUN
fcis-20320	200	6	is	be	AUX
fcis-20320	200	7	:	:	PUNCT
fcis-20320	200	8			PROPN
fcis-20320	200	9	1	1	ADJ
fcis-20320	200	10	2ir	2ir	ADJ
fcis-20320	201	1	p	p	X
fcis-20320	201	2	p	p	X
fcis-20320	201	3	auc	auc	NOUN
fcis-20320	201	4	p	p	NOUN
fcis-20320	201	5	n	n	X
fcis-20320	201	6			PROPN
fcis-20320	201	7			PROPN
fcis-20320	201	8			ADV
fcis-20320	201	9			PROPN
fcis-20320	201	10			PROPN
fcis-20320	201	11			X
fcis-20320	201	12	(	(	PUNCT
fcis-20320	201	13	8)	8)	NUM
fcis-20320	201	14	among	among	ADP
fcis-20320	201	15	them	they	PRON
fcis-20320	201	16	,	,	PUNCT
fcis-20320	201	17			PROPN
fcis-20320	201	18	1	1	ADP
fcis-20320	201	19	2ir	2ir	ADJ
fcis-20320	201	20	p	p	ADJ
fcis-20320	201	21	p	p	NOUN
fcis-20320	201	22			PROPN
fcis-20320	201	23			NOUN
fcis-20320	201	24	represents	represent	VERB
fcis-20320	201	25	the	the	DET
fcis-20320	201	26	number	number	NOUN
fcis-20320	201	27	of	of	ADP
fcis-20320	201	28	positive	positive	ADJ
fcis-20320	201	29	and	and	CCONJ
fcis-20320	201	30	negative	negative	ADJ
fcis-20320	201	31	sample	sample	NOUN
fcis-20320	201	32	pairs	pair	NOUN
fcis-20320	201	33	whose	whose	DET
fcis-20320	201	34	predicted	predict	VERB
fcis-20320	201	35	values	value	NOUN
fcis-20320	201	36	of	of	ADP
fcis-20320	201	37	all	all	DET
fcis-20320	201	38	positive	positive	ADJ
fcis-20320	201	39	samples	sample	NOUN
fcis-20320	201	40	are	be	AUX
fcis-20320	201	41	greater	great	ADJ
fcis-20320	201	42	than	than	ADP
fcis-20320	201	43	the	the	DET
fcis-20320	201	44	predicted	predict	VERB
fcis-20320	201	45	values	value	NOUN
fcis-20320	201	46	of	of	ADP
fcis-20320	201	47	negative	negative	ADJ
fcis-20320	201	48	samples	sample	NOUN
fcis-20320	201	49	.	.	PUNCT
fcis-20320	202	1	4.3	4.3	NUM
fcis-20320	202	2	.	.	PUNCT
fcis-20320	202	3	proposed	propose	VERB
fcis-20320	202	4	architecture	architecture	NOUN
fcis-20320	202	5	the	the	DET
fcis-20320	202	6	model	model	NOUN
fcis-20320	202	7	res	re	NOUN
fcis-20320	202	8	-	-	PUNCT
fcis-20320	202	9	vgg	vgg	NOUN
fcis-20320	202	10	proposed	propose	VERB
fcis-20320	202	11	in	in	ADP
fcis-20320	202	12	this	this	DET
fcis-20320	202	13	chapter	chapter	NOUN
fcis-20320	202	14	combines	combine	VERB
fcis-20320	202	15	two	two	NUM
fcis-20320	202	16	different	different	ADJ
fcis-20320	202	17	cnn	cnn	PROPN
fcis-20320	202	18	models	model	NOUN
fcis-20320	202	19	,	,	PUNCT
fcis-20320	202	20	namely	namely	ADV
fcis-20320	202	21	vgg16	vgg16	NOUN
fcis-20320	202	22	and	and	CCONJ
fcis-20320	202	23	resnet	resnet	NOUN
fcis-20320	202	24	.	.	PUNCT
fcis-20320	203	1	the	the	DET
fcis-20320	203	2	model	model	NOUN
fcis-20320	203	3	uses	use	VERB
fcis-20320	203	4	10	10	NUM
fcis-20320	203	5	convolutional	convolutional	ADJ
fcis-20320	203	6	layers	layer	NOUN
fcis-20320	203	7	,	,	PUNCT
fcis-20320	203	8	2	2	NUM
fcis-20320	203	9	fully	fully	ADV
fcis-20320	203	10	connected	connected	ADJ
fcis-20320	203	11	layers	layer	NOUN
fcis-20320	203	12	,	,	PUNCT
fcis-20320	203	13	and	and	CCONJ
fcis-20320	203	14	1	1	NUM
fcis-20320	203	15	softmax	softmax	NOUN
fcis-20320	203	16	layer	layer	NOUN
fcis-20320	203	17	,	,	PUNCT
fcis-20320	203	18	for	for	ADP
fcis-20320	203	19	a	a	DET
fcis-20320	203	20	total	total	NOUN
fcis-20320	203	21	of	of	ADP
fcis-20320	203	22	13	13	NUM
fcis-20320	203	23	layers	layer	NOUN
fcis-20320	203	24	.	.	PUNCT
fcis-20320	204	1	additionally	additionally	ADV
fcis-20320	204	2	,	,	PUNCT
fcis-20320	204	3	max	max	PROPN
fcis-20320	204	4	pooling	pooling	NOUN
fcis-20320	204	5	and	and	CCONJ
fcis-20320	204	6	average	average	ADJ
fcis-20320	204	7	pooling	pooling	NOUN
fcis-20320	204	8	functions	function	NOUN
fcis-20320	204	9	are	be	AUX
fcis-20320	204	10	used	use	VERB
fcis-20320	204	11	,	,	PUNCT
fcis-20320	204	12	as	as	ADV
fcis-20320	204	13	well	well	ADV
fcis-20320	204	14	as	as	ADP
fcis-20320	204	15	different	different	ADJ
fcis-20320	204	16	convolution	convolution	NOUN
fcis-20320	204	17	kernel	kernel	PROPN
fcis-20320	204	18	sizes	size	NOUN
fcis-20320	204	19	,	,	PUNCT
fcis-20320	204	20	and	and	CCONJ
fcis-20320	204	21	pooling	pool	VERB
fcis-20320	204	22	layers	layer	NOUN
fcis-20320	204	23	are	be	AUX
fcis-20320	204	24	implemented	implement	VERB
fcis-20320	204	25	.	.	PUNCT
fcis-20320	205	1	figure	figure	NOUN
fcis-20320	205	2	4	4	NUM
fcis-20320	205	3	is	be	AUX
fcis-20320	205	4	the	the	DET
fcis-20320	205	5	structure	structure	NOUN
fcis-20320	205	6	diagram	diagram	NOUN
fcis-20320	205	7	of	of	ADP
fcis-20320	205	8	res	re	NOUN
fcis-20320	205	9	-	-	PUNCT
fcis-20320	205	10	vgg	vgg	NOUN
fcis-20320	205	11	.	.	PUNCT
fcis-20320	206	1	this	this	DET
fcis-20320	206	2	study	study	NOUN
fcis-20320	206	3	is	be	AUX
fcis-20320	206	4	improved	improve	VERB
fcis-20320	206	5	based	base	VERB
fcis-20320	206	6	on	on	ADP
fcis-20320	206	7	vgg16	vgg16	PROPN
fcis-20320	206	8	,	,	PUNCT
fcis-20320	206	9	which	which	PRON
fcis-20320	206	10	contains	contain	VERB
fcis-20320	206	11	13	13	NUM
fcis-20320	206	12	convolutional	convolutional	ADJ
fcis-20320	206	13	layers	layer	NOUN
fcis-20320	206	14	and	and	CCONJ
fcis-20320	206	15	3	3	NUM
fcis-20320	206	16	fully	fully	ADV
fcis-20320	206	17	connected	connected	ADJ
fcis-20320	206	18	layers	layer	NOUN
fcis-20320	206	19	.	.	PUNCT
fcis-20320	207	1	first	first	ADV
fcis-20320	207	2	,	,	PUNCT
fcis-20320	207	3	the	the	DET
fcis-20320	207	4	convolutional	convolutional	ADJ
fcis-20320	207	5	layers	layer	NOUN
fcis-20320	207	6	of	of	ADP
fcis-20320	207	7	vgg16	vgg16	PROPN
fcis-20320	207	8	are	be	AUX
fcis-20320	207	9	reduced	reduce	VERB
fcis-20320	207	10	to	to	ADP
fcis-20320	207	11	10	10	NUM
fcis-20320	207	12	,	,	PUNCT
fcis-20320	207	13	which	which	PRON
fcis-20320	207	14	is	be	AUX
fcis-20320	207	15	intended	intend	VERB
fcis-20320	207	16	to	to	PART
fcis-20320	207	17	reduce	reduce	VERB
fcis-20320	207	18	the	the	DET
fcis-20320	207	19	complexity	complexity	NOUN
fcis-20320	207	20	of	of	ADP
fcis-20320	207	21	the	the	DET
fcis-20320	207	22	model	model	NOUN
fcis-20320	207	23	and	and	CCONJ
fcis-20320	207	24	improve	improve	VERB
fcis-20320	207	25	computational	computational	ADJ
fcis-20320	207	26	efficiency	efficiency	NOUN
fcis-20320	207	27	.	.	PUNCT
fcis-20320	208	1	every	every	DET
fcis-20320	208	2	two	two	NUM
fcis-20320	208	3	convolutional	convolutional	ADJ
fcis-20320	208	4	layers	layer	NOUN
fcis-20320	208	5	form	form	VERB
fcis-20320	208	6	a	a	DET
fcis-20320	208	7	pair	pair	NOUN
fcis-20320	208	8	,	,	PUNCT
fcis-20320	208	9	and	and	CCONJ
fcis-20320	208	10	except	except	SCONJ
fcis-20320	208	11	for	for	ADP
fcis-20320	208	12	the	the	DET
fcis-20320	208	13	second	second	ADJ
fcis-20320	208	14	pooling	pool	VERB
fcis-20320	208	15	layer	layer	NOUN
fcis-20320	208	16	,	,	PUNCT
fcis-20320	208	17	all	all	DET
fcis-20320	208	18	other	other	ADJ
fcis-20320	208	19	layers	layer	NOUN
fcis-20320	208	20	use	use	VERB
fcis-20320	208	21	average	average	ADJ
fcis-20320	208	22	pooling	pooling	NOUN
fcis-20320	208	23	.	.	PUNCT
fcis-20320	209	1	this	this	DET
fcis-20320	209	2	design	design	NOUN
fcis-20320	209	3	helps	help	VERB
fcis-20320	209	4	to	to	PART
fcis-20320	209	5	better	well	ADV
fcis-20320	209	6	retain	retain	VERB
fcis-20320	209	7	feature	feature	NOUN
fcis-20320	209	8	information	information	NOUN
fcis-20320	209	9	while	while	SCONJ
fcis-20320	209	10	reducing	reduce	VERB
fcis-20320	209	11	the	the	DET
fcis-20320	209	12	number	number	NOUN
fcis-20320	209	13	of	of	ADP
fcis-20320	209	14	parameters	parameter	NOUN
fcis-20320	209	15	of	of	ADP
fcis-20320	209	16	the	the	DET
fcis-20320	209	17	model	model	NOUN
fcis-20320	209	18	.	.	PUNCT
fcis-20320	210	1	in	in	ADP
fcis-20320	210	2	addition	addition	NOUN
fcis-20320	210	3	,	,	PUNCT
fcis-20320	210	4	by	by	ADP
fcis-20320	210	5	reducing	reduce	VERB
fcis-20320	210	6	one	one	NUM
fcis-20320	210	7	pooling	pool	VERB
fcis-20320	210	8	layer	layer	NOUN
fcis-20320	210	9	and	and	CCONJ
fcis-20320	210	10	one	one	NUM
fcis-20320	210	11	fully	fully	ADV
fcis-20320	210	12	connected	connected	ADJ
fcis-20320	210	13	layer	layer	NOUN
fcis-20320	210	14	,	,	PUNCT
fcis-20320	210	15	the	the	DET
fcis-20320	210	16	complexity	complexity	NOUN
fcis-20320	210	17	of	of	ADP
fcis-20320	210	18	the	the	DET
fcis-20320	210	19	model	model	NOUN
fcis-20320	210	20	is	be	AUX
fcis-20320	210	21	further	far	ADV
fcis-20320	210	22	reduced	reduce	VERB
fcis-20320	210	23	and	and	CCONJ
fcis-20320	210	24	helps	helps	AUX
fcis-20320	210	25	reduce	reduce	VERB
fcis-20320	210	26	the	the	DET
fcis-20320	210	27	possibility	possibility	NOUN
fcis-20320	210	28	of	of	ADP
fcis-20320	210	29	overfitting	overfitte	VERB
fcis-20320	210	30	.	.	PUNCT
fcis-20320	211	1	next	next	ADV
fcis-20320	211	2	,	,	PUNCT
fcis-20320	211	3	the	the	DET
fcis-20320	211	4	kernel	kernel	NOUN
fcis-20320	211	5	size	size	NOUN
fcis-20320	211	6	of	of	ADP
fcis-20320	211	7	the	the	DET
fcis-20320	211	8	second	second	ADJ
fcis-20320	211	9	convolutional	convolutional	ADJ
fcis-20320	211	10	layer	layer	NOUN
fcis-20320	211	11	was	be	AUX
fcis-20320	211	12	modified	modify	VERB
fcis-20320	211	13	to	to	ADP
fcis-20320	211	14	5×5	5×5	NUM
fcis-20320	211	15	to	to	PART
fcis-20320	211	16	better	well	ADV
fcis-20320	211	17	capture	capture	VERB
fcis-20320	211	18	local	local	ADJ
fcis-20320	211	19	features	feature	NOUN
fcis-20320	211	20	,	,	PUNCT
fcis-20320	211	21	resulting	result	VERB
fcis-20320	211	22	in	in	ADP
fcis-20320	211	23	an	an	DET
fcis-20320	211	24	improved	improved	ADJ
fcis-20320	211	25	13	13	NUM
fcis-20320	211	26	-	-	PUNCT
fcis-20320	211	27	layer	layer	NOUN
fcis-20320	211	28	vgg16	vgg16	NOUN
fcis-20320	211	29	network	network	NOUN
fcis-20320	211	30	.	.	PUNCT
fcis-20320	212	1	on	on	ADP
fcis-20320	212	2	this	this	DET
fcis-20320	212	3	basis	basis	NOUN
fcis-20320	212	4	,	,	PUNCT
fcis-20320	212	5	in	in	ADP
fcis-20320	212	6	order	order	NOUN
fcis-20320	212	7	to	to	PART
fcis-20320	212	8	solve	solve	VERB
fcis-20320	212	9	the	the	DET
fcis-20320	212	10	vanishing	vanish	VERB
fcis-20320	212	11	gradient	gradient	NOUN
fcis-20320	212	12	problem	problem	NOUN
fcis-20320	212	13	in	in	ADP
fcis-20320	212	14	deep	deep	ADJ
fcis-20320	212	15	networks	network	NOUN
fcis-20320	212	16	and	and	CCONJ
fcis-20320	212	17	improve	improve	VERB
fcis-20320	212	18	the	the	DET
fcis-20320	212	19	training	training	NOUN
fcis-20320	212	20	efficiency	efficiency	NOUN
fcis-20320	212	21	of	of	ADP
fcis-20320	212	22	the	the	DET
fcis-20320	212	23	model	model	NOUN
fcis-20320	212	24	,	,	PUNCT
fcis-20320	212	25	a	a	DET
fcis-20320	212	26	residual	residual	ADJ
fcis-20320	212	27	connection	connection	NOUN
fcis-20320	212	28	is	be	AUX
fcis-20320	212	29	added	add	VERB
fcis-20320	212	30	between	between	ADP
fcis-20320	212	31	the	the	DET
fcis-20320	212	32	input	input	NOUN
fcis-20320	212	33	and	and	CCONJ
fcis-20320	212	34	output	output	NOUN
fcis-20320	212	35	of	of	ADP
fcis-20320	212	36	each	each	DET
fcis-20320	212	37	pair	pair	NOUN
fcis-20320	212	38	of	of	ADP
fcis-20320	212	39	convolutional	convolutional	ADJ
fcis-20320	212	40	layers	layer	NOUN
fcis-20320	212	41	.	.	PUNCT
fcis-20320	213	1	when	when	SCONJ
fcis-20320	213	2	the	the	DET
fcis-20320	213	3	input	input	NOUN
fcis-20320	213	4	and	and	CCONJ
fcis-20320	213	5	output	output	NOUN
fcis-20320	213	6	dimensions	dimension	NOUN
fcis-20320	213	7	are	be	AUX
fcis-20320	213	8	the	the	DET
fcis-20320	213	9	same	same	ADJ
fcis-20320	213	10	,	,	PUNCT
fcis-20320	213	11	the	the	DET
fcis-20320	213	12	connection	connection	NOUN
fcis-20320	213	13	method	method	NOUN
fcis-20320	213	14	of	of	ADP
fcis-20320	213	15	equation	equation	NOUN
fcis-20320	213	16	(	(	PUNCT
fcis-20320	213	17	1	1	X
fcis-20320	213	18	)	)	PUNCT
fcis-20320	213	19	is	be	AUX
fcis-20320	213	20	used	use	VERB
fcis-20320	213	21	.	.	PUNCT
fcis-20320	214	1	when	when	SCONJ
fcis-20320	214	2	the	the	DET
fcis-20320	214	3	dimensions	dimension	NOUN
fcis-20320	214	4	increase	increase	VERB
fcis-20320	214	5	,	,	PUNCT
fcis-20320	214	6	the	the	DET
fcis-20320	214	7	connection	connection	NOUN
fcis-20320	214	8	method	method	NOUN
fcis-20320	214	9	of	of	ADP
fcis-20320	214	10	equation	equation	NOUN
fcis-20320	214	11	(	(	PUNCT
fcis-20320	214	12	2	2	X
fcis-20320	214	13	)	)	PUNCT
fcis-20320	214	14	is	be	AUX
fcis-20320	214	15	used	use	VERB
fcis-20320	214	16	to	to	PART
fcis-20320	214	17	match	match	VERB
fcis-20320	214	18	the	the	DET
fcis-20320	214	19	size	size	NOUN
fcis-20320	214	20	.	.	PUNCT
fcis-20320	215	1	fully	fully	ADV
fcis-20320	215	2	connected	connected	ADJ
fcis-20320	215	3	layers	layer	NOUN
fcis-20320	215	4	are	be	AUX
fcis-20320	215	5	used	use	VERB
fcis-20320	215	6	to	to	PART
fcis-20320	215	7	extract	extract	VERB
fcis-20320	215	8	information	information	NOUN
fcis-20320	215	9	,	,	PUNCT
fcis-20320	215	10	while	while	SCONJ
fcis-20320	215	11	the	the	DET
fcis-20320	215	12	last	last	ADJ
fcis-20320	215	13	layer	layer	NOUN
fcis-20320	215	14	of	of	ADP
fcis-20320	215	15	the	the	DET
fcis-20320	215	16	network	network	NOUN
fcis-20320	215	17	is	be	AUX
fcis-20320	215	18	a	a	DET
fcis-20320	215	19	softmax	softmax	NOUN
fcis-20320	215	20	layer	layer	NOUN
fcis-20320	215	21	for	for	ADP
fcis-20320	215	22	predicting	predict	VERB
fcis-20320	215	23	two	two	NUM
fcis-20320	215	24	categories	category	NOUN
fcis-20320	215	25	of	of	ADP
fcis-20320	215	26	lung	lung	NOUN
fcis-20320	215	27	nodules	nodule	NOUN
fcis-20320	215	28	(	(	PUNCT
fcis-20320	215	29	benign	benign	ADJ
fcis-20320	215	30	and	and	CCONJ
fcis-20320	215	31	malignant	malignant	ADJ
fcis-20320	215	32	)	)	PUNCT
fcis-20320	215	33	.	.	PUNCT
fcis-20320	216	1	fig	fig	NOUN
fcis-20320	216	2	4	4	NUM
fcis-20320	216	3	.	.	PUNCT
fcis-20320	217	1	the	the	DET
fcis-20320	217	2	structure	structure	NOUN
fcis-20320	217	3	of	of	ADP
fcis-20320	217	4	the	the	DET
fcis-20320	217	5	res	re	NOUN
fcis-20320	217	6	-	-	PUNCT
fcis-20320	217	7	vgg	vgg	ADJ
fcis-20320	217	8	11	11	NUM
fcis-20320	217	9	4.4	4.4	NUM
fcis-20320	217	10	.	.	PUNCT
fcis-20320	218	1	results	result	NOUN
fcis-20320	218	2	and	and	CCONJ
fcis-20320	218	3	analysis	analysis	NOUN
fcis-20320	218	4	during	during	ADP
fcis-20320	218	5	the	the	DET
fcis-20320	218	6	training	training	NOUN
fcis-20320	218	7	process	process	NOUN
fcis-20320	218	8	,	,	PUNCT
fcis-20320	218	9	data	datum	NOUN
fcis-20320	218	10	sets	set	NOUN
fcis-20320	218	11	with	with	ADP
fcis-20320	218	12	a	a	DET
fcis-20320	218	13	benign	benign	ADJ
fcis-20320	218	14	and	and	CCONJ
fcis-20320	218	15	malignant	malignant	ADJ
fcis-20320	218	16	degree	degree	NOUN
fcis-20320	218	17	of	of	ADP
fcis-20320	218	18	3	3	NUM
fcis-20320	218	19	(	(	PUNCT
fcis-20320	218	20	indicating	indicate	VERB
fcis-20320	218	21	uncertainty	uncertainty	NOUN
fcis-20320	218	22	)	)	PUNCT
fcis-20320	218	23	were	be	AUX
fcis-20320	218	24	discarded	discard	VERB
fcis-20320	218	25	.	.	PUNCT
fcis-20320	219	1	it	it	PRON
fcis-20320	219	2	is	be	AUX
fcis-20320	219	3	only	only	ADV
fcis-20320	219	4	necessary	necessary	ADJ
fcis-20320	219	5	to	to	PART
fcis-20320	219	6	judge	judge	VERB
fcis-20320	219	7	whether	whether	SCONJ
fcis-20320	219	8	benign	benign	ADJ
fcis-20320	219	9	or	or	CCONJ
fcis-20320	219	10	malignant	malignant	ADJ
fcis-20320	219	11	,	,	PUNCT
fcis-20320	219	12	so	so	CCONJ
fcis-20320	219	13	the	the	DET
fcis-20320	219	14	benign	benign	ADJ
fcis-20320	219	15	and	and	CCONJ
fcis-20320	219	16	malignant	malignant	ADJ
fcis-20320	219	17	degrees	degree	NOUN
fcis-20320	219	18	of	of	ADP
fcis-20320	219	19	1	1	NUM
fcis-20320	219	20	and	and	CCONJ
fcis-20320	219	21	2	2	NUM
fcis-20320	219	22	are	be	AUX
fcis-20320	219	23	classified	classify	VERB
fcis-20320	219	24	as	as	ADP
fcis-20320	219	25	benign	benign	ADJ
fcis-20320	219	26	(	(	PUNCT
fcis-20320	219	27	malignant	malignant	ADJ
fcis-20320	219	28	)	)	PUNCT
fcis-20320	219	29	,	,	PUNCT
fcis-20320	219	30	and	and	CCONJ
fcis-20320	219	31	the	the	DET
fcis-20320	219	32	benign	benign	ADJ
fcis-20320	219	33	and	and	CCONJ
fcis-20320	219	34	malignant	malignant	ADJ
fcis-20320	219	35	degrees	degree	NOUN
fcis-20320	219	36	of	of	ADP
fcis-20320	219	37	4	4	NUM
fcis-20320	219	38	and	and	CCONJ
fcis-20320	219	39	5	5	NUM
fcis-20320	219	40	are	be	AUX
fcis-20320	219	41	classified	classify	VERB
fcis-20320	219	42	as	as	ADP
fcis-20320	219	43	malignant	malignant	ADJ
fcis-20320	219	44	(	(	PUNCT
fcis-20320	219	45	benign	benign	ADJ
fcis-20320	219	46	)	)	PUNCT
fcis-20320	219	47	.	.	PUNCT
fcis-20320	220	1	through	through	ADP
fcis-20320	220	2	the	the	DET
fcis-20320	220	3	data	datum	NOUN
fcis-20320	220	4	preprocessing	preprocesse	VERB
fcis-20320	220	5	in	in	ADP
fcis-20320	220	6	the	the	DET
fcis-20320	220	7	previous	previous	ADJ
fcis-20320	220	8	section	section	NOUN
fcis-20320	220	9	,	,	PUNCT
fcis-20320	220	10	450	450	NUM
fcis-20320	220	11	malignant	malignant	ADJ
fcis-20320	220	12	nodules	nodule	NOUN
fcis-20320	220	13	and	and	CCONJ
fcis-20320	220	14	554	554	NUM
fcis-20320	220	15	benign	benign	ADJ
fcis-20320	220	16	nodules	nodule	NOUN
fcis-20320	220	17	were	be	AUX
fcis-20320	220	18	finally	finally	ADV
fcis-20320	220	19	obtained	obtain	VERB
fcis-20320	220	20	.	.	PUNCT
fcis-20320	221	1	the	the	DET
fcis-20320	221	2	res	re	NOUN
fcis-20320	221	3	-	-	PUNCT
fcis-20320	221	4	vgg	vgg	NOUN
fcis-20320	221	5	proposed	propose	VERB
fcis-20320	221	6	in	in	ADP
fcis-20320	221	7	this	this	DET
fcis-20320	221	8	chapter	chapter	NOUN
fcis-20320	221	9	was	be	AUX
fcis-20320	221	10	evaluated	evaluate	VERB
fcis-20320	221	11	on	on	ADP
fcis-20320	221	12	the	the	DET
fcis-20320	221	13	luna16	luna16	ADJ
fcis-20320	221	14	database	database	NOUN
fcis-20320	221	15	containing	contain	VERB
fcis-20320	221	16	1004	1004	NUM
fcis-20320	221	17	nodules	nodule	NOUN
fcis-20320	221	18	(	(	PUNCT
fcis-20320	221	19	554	554	NUM
fcis-20320	221	20	benign	benign	ADJ
fcis-20320	221	21	and	and	CCONJ
fcis-20320	221	22	450	450	NUM
fcis-20320	221	23	malignant	malignant	NOUN
fcis-20320	221	24	)	)	PUNCT
fcis-20320	221	25	.	.	PUNCT
fcis-20320	222	1	all	all	DET
fcis-20320	222	2	image	image	NOUN
fcis-20320	222	3	data	datum	NOUN
fcis-20320	222	4	were	be	AUX
fcis-20320	222	5	randomly	randomly	ADV
fcis-20320	222	6	divided	divide	VERB
fcis-20320	222	7	into	into	ADP
fcis-20320	222	8	10	10	NUM
fcis-20320	222	9	subsets	subset	NOUN
fcis-20320	222	10	for	for	ADP
fcis-20320	222	11	ten	ten	ADJ
fcis-20320	222	12	-	-	ADJ
fcis-20320	222	13	fold	fold	ADJ
fcis-20320	222	14	cross	cross	NOUN
fcis-20320	222	15	validation	validation	NOUN
fcis-20320	222	16	.	.	PUNCT
fcis-20320	223	1	for	for	ADP
fcis-20320	223	2	each	each	DET
fcis-20320	223	3	fold	fold	NOUN
fcis-20320	223	4	,	,	PUNCT
fcis-20320	223	5	nine	nine	NUM
fcis-20320	223	6	subsets	subset	NOUN
fcis-20320	223	7	were	be	AUX
fcis-20320	223	8	selected	select	VERB
fcis-20320	223	9	for	for	ADP
fcis-20320	223	10	training	training	NOUN
fcis-20320	223	11	and	and	CCONJ
fcis-20320	223	12	one	one	NUM
fcis-20320	223	13	subset	subset	NOUN
fcis-20320	223	14	for	for	ADP
fcis-20320	223	15	testing	testing	NOUN
fcis-20320	223	16	.	.	PUNCT
fcis-20320	224	1	this	this	DET
fcis-20320	224	2	experiment	experiment	NOUN
fcis-20320	224	3	was	be	AUX
fcis-20320	224	4	conducted	conduct	VERB
fcis-20320	224	5	under	under	ADP
fcis-20320	224	6	the	the	DET
fcis-20320	224	7	windows	window	NOUN
fcis-20320	224	8	10	10	NUM
fcis-20320	224	9	operating	operating	NOUN
fcis-20320	224	10	system	system	NOUN
fcis-20320	224	11	with	with	ADP
fcis-20320	224	12	nvidia	nvidia	PROPN
fcis-20320	224	13	geforce	geforce	NOUN
fcis-20320	224	14	rtx	rtx	PROPN
fcis-20320	224	15	4070	4070	NUM
fcis-20320	224	16	12	12	NUM
fcis-20320	224	17	gb	gb	PROPN
fcis-20320	224	18	gpu	gpu	PROPN
fcis-20320	224	19	.	.	PUNCT
fcis-20320	225	1	the	the	DET
fcis-20320	225	2	learning	learning	NOUN
fcis-20320	225	3	process	process	NOUN
fcis-20320	225	4	is	be	AUX
fcis-20320	225	5	optimized	optimize	VERB
fcis-20320	225	6	using	use	VERB
fcis-20320	225	7	the	the	DET
fcis-20320	225	8	stochastic	stochastic	ADJ
fcis-20320	225	9	gradient	gradient	ADJ
fcis-20320	225	10	descent	descent	NOUN
fcis-20320	225	11	(	(	PUNCT
fcis-20320	225	12	sgd	sgd	NOUN
fcis-20320	225	13	)	)	PUNCT
fcis-20320	225	14	algorithm	algorithm	NOUN
fcis-20320	225	15	with	with	ADP
fcis-20320	225	16	a	a	DET
fcis-20320	225	17	momentum	momentum	NOUN
fcis-20320	225	18	of	of	ADP
fcis-20320	225	19	0.9	0.9	NUM
fcis-20320	225	20	.	.	PUNCT
fcis-20320	226	1	in	in	ADP
fcis-20320	226	2	order	order	NOUN
fcis-20320	226	3	to	to	PART
fcis-20320	226	4	reduce	reduce	VERB
fcis-20320	226	5	overfitting	overfitting	NOUN
fcis-20320	226	6	,	,	PUNCT
fcis-20320	226	7	the	the	DET
fcis-20320	226	8	network	network	NOUN
fcis-20320	226	9	uses	use	VERB
fcis-20320	226	10	regularization	regularization	NOUN
fcis-20320	226	11	technology	technology	NOUN
fcis-20320	226	12	and	and	CCONJ
fcis-20320	226	13	dropout	dropout	NOUN
fcis-20320	226	14	.	.	PUNCT
fcis-20320	227	1	additionally	additionally	ADV
fcis-20320	227	2	,	,	PUNCT
fcis-20320	227	3	due	due	ADP
fcis-20320	227	4	to	to	ADP
fcis-20320	227	5	the	the	DET
fcis-20320	227	6	12	12	NUM
fcis-20320	227	7	gb	gb	PROPN
fcis-20320	227	8	gpu	gpu	NOUN
fcis-20320	227	9	memory	memory	NOUN
fcis-20320	227	10	limit	limit	NOUN
fcis-20320	227	11	,	,	PUNCT
fcis-20320	227	12	the	the	DET
fcis-20320	227	13	batch	batch	NOUN
fcis-20320	227	14	size	size	NOUN
fcis-20320	227	15	is	be	AUX
fcis-20320	227	16	limited	limit	VERB
fcis-20320	227	17	to	to	ADP
fcis-20320	227	18	8	8	NUM
fcis-20320	227	19	.	.	PUNCT
fcis-20320	228	1	the	the	DET
fcis-20320	228	2	learning	learning	NOUN
fcis-20320	228	3	model	model	NOUN
fcis-20320	228	4	was	be	AUX
fcis-20320	228	5	trained	train	VERB
fcis-20320	228	6	for	for	ADP
fcis-20320	228	7	200	200	NUM
fcis-20320	228	8	epochs	epoch	NOUN
fcis-20320	228	9	.	.	PUNCT
fcis-20320	229	1	the	the	DET
fcis-20320	229	2	initial	initial	ADJ
fcis-20320	229	3	learning	learning	NOUN
fcis-20320	229	4	rate	rate	NOUN
fcis-20320	229	5	is	be	AUX
fcis-20320	229	6	0.01	0.01	NUM
fcis-20320	229	7	,	,	PUNCT
fcis-20320	229	8	which	which	PRON
fcis-20320	229	9	decreases	decrease	VERB
fcis-20320	229	10	to	to	ADP
fcis-20320	229	11	0.001	0.001	NUM
fcis-20320	229	12	after	after	ADP
fcis-20320	229	13	60	60	NUM
fcis-20320	229	14	epochs	epoch	NOUN
fcis-20320	229	15	and	and	CCONJ
fcis-20320	229	16	to	to	ADP
fcis-20320	229	17	0.0001	0.0001	NUM
fcis-20320	229	18	after	after	ADP
fcis-20320	229	19	120	120	NUM
fcis-20320	229	20	epochs	epoch	NOUN
fcis-20320	229	21	.	.	PUNCT
fcis-20320	230	1	the	the	DET
fcis-20320	230	2	weight	weight	NOUN
fcis-20320	230	3	decay	decay	NOUN
fcis-20320	230	4	is	be	AUX
fcis-20320	230	5	set	set	VERB
fcis-20320	230	6	to	to	ADP
fcis-20320	230	7	1×10	1×10	NUM
fcis-20320	230	8	-	-	SYM
fcis-20320	230	9	4	4	NUM
fcis-20320	230	10	.	.	PUNCT
fcis-20320	231	1	during	during	ADP
fcis-20320	231	2	the	the	DET
fcis-20320	231	3	optimization	optimization	NOUN
fcis-20320	231	4	process	process	NOUN
fcis-20320	231	5	,	,	PUNCT
fcis-20320	231	6	it	it	PRON
fcis-20320	231	7	needs	need	VERB
fcis-20320	231	8	to	to	PART
fcis-20320	231	9	use	use	VERB
fcis-20320	231	10	a	a	DET
fcis-20320	231	11	higher	high	ADJ
fcis-20320	231	12	learning	learning	NOUN
fcis-20320	231	13	rate	rate	NOUN
fcis-20320	231	14	to	to	PART
fcis-20320	231	15	move	move	VERB
fcis-20320	231	16	quickly	quickly	ADV
fcis-20320	231	17	,	,	PUNCT
fcis-20320	231	18	but	but	CCONJ
fcis-20320	231	19	when	when	SCONJ
fcis-20320	231	20	it	it	PRON
fcis-20320	231	21	reaches	reach	VERB
fcis-20320	231	22	a	a	DET
fcis-20320	231	23	point	point	NOUN
fcis-20320	231	24	close	close	ADV
fcis-20320	231	25	to	to	ADP
fcis-20320	231	26	the	the	DET
fcis-20320	231	27	relative	relative	ADJ
fcis-20320	231	28	optimal	optimal	ADJ
fcis-20320	231	29	point	point	NOUN
fcis-20320	231	30	,	,	PUNCT
fcis-20320	231	31	the	the	DET
fcis-20320	231	32	learning	learning	NOUN
fcis-20320	231	33	rate	rate	NOUN
fcis-20320	231	34	must	must	AUX
fcis-20320	231	35	be	be	AUX
fcis-20320	231	36	reduced	reduce	VERB
fcis-20320	231	37	so	so	ADV
fcis-20320	231	38	as	as	SCONJ
fcis-20320	231	39	not	not	PART
fcis-20320	231	40	to	to	PART
fcis-20320	231	41	miss	miss	VERB
fcis-20320	231	42	the	the	DET
fcis-20320	231	43	optimal	optimal	ADJ
fcis-20320	231	44	point	point	NOUN
fcis-20320	231	45	.	.	PUNCT
fcis-20320	232	1	therefore	therefore	ADV
fcis-20320	232	2	,	,	PUNCT
fcis-20320	232	3	the	the	DET
fcis-20320	232	4	learning	learning	NOUN
fcis-20320	232	5	rate	rate	NOUN
fcis-20320	232	6	is	be	AUX
fcis-20320	232	7	reduced	reduce	VERB
fcis-20320	232	8	by	by	ADP
fcis-20320	232	9	a	a	DET
fcis-20320	232	10	decay	decay	NOUN
fcis-20320	232	11	factor	factor	NOUN
fcis-20320	232	12	in	in	ADP
fcis-20320	232	13	each	each	DET
fcis-20320	232	14	epoch	epoch	NOUN
fcis-20320	232	15	.	.	PUNCT
fcis-20320	233	1	the	the	DET
fcis-20320	233	2	parameter	parameter	NOUN
fcis-20320	233	3	settings	setting	NOUN
fcis-20320	233	4	of	of	ADP
fcis-20320	233	5	this	this	DET
fcis-20320	233	6	experiment	experiment	NOUN
fcis-20320	233	7	are	be	AUX
fcis-20320	233	8	shown	show	VERB
fcis-20320	233	9	in	in	ADP
fcis-20320	233	10	table	table	NOUN
fcis-20320	233	11	1	1	NUM
fcis-20320	233	12	.	.	PUNCT
fcis-20320	233	13	table	table	NOUN
fcis-20320	233	14	1	1	NUM
fcis-20320	233	15	.	.	PUNCT
fcis-20320	233	16	experimental	experimental	ADJ
fcis-20320	233	17	parameter	parameter	NOUN
fcis-20320	233	18	configuration	configuration	NOUN
fcis-20320	233	19	parameters	parameter	NOUN
fcis-20320	233	20	batch	batch	VERB
fcis-20320	233	21	size	size	NOUN
fcis-20320	233	22	epoch	epoch	NOUN
fcis-20320	233	23	learning	learning	NOUN
fcis-20320	233	24	rate	rate	NOUN
fcis-20320	233	25	weight	weight	NOUN
fcis-20320	233	26	decay	decay	NOUN
fcis-20320	233	27	value	value	NOUN
fcis-20320	233	28	8	8	NUM
fcis-20320	233	29	200	200	NUM
fcis-20320	233	30	0.01	0.01	NUM
fcis-20320	233	31	0.0001	0.0001	NUM
fcis-20320	233	32	the	the	DET
fcis-20320	233	33	final	final	ADJ
fcis-20320	233	34	performance	performance	NOUN
fcis-20320	233	35	of	of	ADP
fcis-20320	233	36	the	the	DET
fcis-20320	233	37	method	method	NOUN
fcis-20320	233	38	is	be	AUX
fcis-20320	233	39	the	the	DET
fcis-20320	233	40	average	average	ADJ
fcis-20320	233	41	performance	performance	NOUN
fcis-20320	233	42	of	of	ADP
fcis-20320	233	43	ten	ten	NUM
fcis-20320	233	44	test	test	NOUN
fcis-20320	233	45	folds	fold	NOUN
fcis-20320	233	46	,	,	PUNCT
fcis-20320	233	47	evaluated	evaluate	VERB
fcis-20320	233	48	by	by	ADP
fcis-20320	233	49	measuring	measure	VERB
fcis-20320	233	50	accuracy	accuracy	NOUN
fcis-20320	233	51	,	,	PUNCT
fcis-20320	233	52	sensitivity	sensitivity	NOUN
fcis-20320	233	53	,	,	PUNCT
fcis-20320	233	54	specificity	specificity	NOUN
fcis-20320	233	55	,	,	PUNCT
fcis-20320	233	56	and	and	CCONJ
fcis-20320	233	57	auc	auc	NOUN
fcis-20320	233	58	.	.	PUNCT
fcis-20320	234	1	the	the	DET
fcis-20320	234	2	proposed	propose	VERB
fcis-20320	234	3	benign	benign	ADJ
fcis-20320	234	4	and	and	CCONJ
fcis-20320	234	5	malignant	malignant	ADJ
fcis-20320	234	6	pulmonary	pulmonary	ADJ
fcis-20320	234	7	nodule	nodule	NOUN
fcis-20320	234	8	classification	classification	NOUN
fcis-20320	234	9	network	network	NOUN
fcis-20320	234	10	res	re	NOUN
fcis-20320	234	11	-	-	PUNCT
fcis-20320	234	12	vgg	vgg	NOUN
fcis-20320	234	13	was	be	AUX
fcis-20320	234	14	compared	compare	VERB
fcis-20320	234	15	with	with	ADP
fcis-20320	234	16	other	other	ADJ
fcis-20320	234	17	networks	network	NOUN
fcis-20320	234	18	vgg16	vgg16	PROPN
fcis-20320	234	19	,	,	PUNCT
fcis-20320	234	20	resnet	resnet	NOUN
fcis-20320	234	21	,	,	PUNCT
fcis-20320	234	22	and	and	CCONJ
fcis-20320	234	23	densenet	densenet	NOUN
fcis-20320	234	24	used	use	VERB
fcis-20320	234	25	for	for	ADP
fcis-20320	234	26	pulmonary	pulmonary	ADJ
fcis-20320	234	27	nodule	nodule	NOUN
fcis-20320	234	28	classification	classification	NOUN
fcis-20320	234	29	.	.	PUNCT
fcis-20320	235	1	the	the	DET
fcis-20320	235	2	experimental	experimental	ADJ
fcis-20320	235	3	results	result	NOUN
fcis-20320	235	4	are	be	AUX
fcis-20320	235	5	shown	show	VERB
fcis-20320	235	6	in	in	ADP
fcis-20320	235	7	table	table	NOUN
fcis-20320	235	8	2	2	NUM
fcis-20320	235	9	.	.	PUNCT
fcis-20320	235	10	table	table	NOUN
fcis-20320	235	11	2	2	NUM
fcis-20320	235	12	.	.	PUNCT
fcis-20320	235	13	comparison	comparison	NOUN
fcis-20320	235	14	experiment	experiment	NOUN
fcis-20320	235	15	models	model	NOUN
fcis-20320	235	16	accuracy/%	accuracy/%	ADP
fcis-20320	235	17	sensitivity/%	sensitivity/%	ADV
fcis-20320	235	18	specificity/%	specificity/%	ADJ
fcis-20320	235	19	auc	auc	X
fcis-20320	235	20	res	re	NOUN
fcis-20320	235	21	-	-	PUNCT
fcis-20320	235	22	vgg	vgg	NOUN
fcis-20320	235	23	84.35	84.35	NUM
fcis-20320	235	24	83.86	83.86	NUM
fcis-20320	235	25	84.52	84.52	NUM
fcis-20320	235	26	0.920	0.920	NUM
fcis-20320	235	27	vgg16	vgg16	PROPN
fcis-20320	235	28	83.36	83.36	NUM
fcis-20320	235	29	82.53	82.53	NUM
fcis-20320	235	30	82.88	82.88	NUM
fcis-20320	235	31	0.895	0.895	NUM
fcis-20320	235	32	resnet	resnet	VERB
fcis-20320	235	33	82.35	82.35	NUM
fcis-20320	235	34	80.67	80.67	NUM
fcis-20320	235	35	83.80	83.80	NUM
fcis-20320	235	36	0.876	0.876	NUM
fcis-20320	235	37	densenet	densenet	NOUN
fcis-20320	235	38	82.58	82.58	NUM
fcis-20320	235	39	83.96	83.96	NUM
fcis-20320	235	40	81.45	81.45	NUM
fcis-20320	235	41	0.882	0.882	NUM
fcis-20320	235	42	as	as	SCONJ
fcis-20320	235	43	shown	show	VERB
fcis-20320	235	44	in	in	ADP
fcis-20320	235	45	table	table	NOUN
fcis-20320	235	46	2	2	NUM
fcis-20320	235	47	,	,	PUNCT
fcis-20320	235	48	compared	compare	VERB
fcis-20320	235	49	with	with	ADP
fcis-20320	235	50	densenet	densenet	NOUN
fcis-20320	235	51	,	,	PUNCT
fcis-20320	235	52	resnet	resnet	NOUN
fcis-20320	235	53	is	be	AUX
fcis-20320	235	54	slightly	slightly	ADV
fcis-20320	235	55	lower	low	ADJ
fcis-20320	235	56	in	in	ADP
fcis-20320	235	57	both	both	DET
fcis-20320	235	58	accuracy	accuracy	NOUN
fcis-20320	235	59	and	and	CCONJ
fcis-20320	235	60	sensitivity	sensitivity	NOUN
fcis-20320	235	61	,	,	PUNCT
fcis-20320	235	62	but	but	CCONJ
fcis-20320	235	63	has	have	VERB
fcis-20320	235	64	a	a	DET
fcis-20320	235	65	better	well	ADJ
fcis-20320	235	66	impact	impact	NOUN
fcis-20320	235	67	on	on	ADP
fcis-20320	235	68	specificity	specificity	NOUN
fcis-20320	235	69	,	,	PUNCT
fcis-20320	235	70	which	which	PRON
fcis-20320	235	71	is	be	AUX
fcis-20320	235	72	83.80	83.80	NUM
fcis-20320	235	73	%	%	NOUN
fcis-20320	235	74	.	.	PUNCT
fcis-20320	236	1	good	good	ADJ
fcis-20320	236	2	specificity	specificity	NOUN
fcis-20320	236	3	means	mean	VERB
fcis-20320	236	4	that	that	SCONJ
fcis-20320	236	5	more	more	ADV
fcis-20320	236	6	malignant	malignant	ADJ
fcis-20320	236	7	pulmonary	pulmonary	ADJ
fcis-20320	236	8	nodules	nodule	NOUN
fcis-20320	236	9	can	can	AUX
fcis-20320	236	10	be	be	AUX
fcis-20320	236	11	detected	detect	VERB
fcis-20320	236	12	in	in	ADP
fcis-20320	236	13	the	the	DET
fcis-20320	236	14	same	same	ADJ
fcis-20320	236	15	data	datum	NOUN
fcis-20320	236	16	set	set	VERB
fcis-20320	236	17	,	,	PUNCT
fcis-20320	236	18	which	which	PRON
fcis-20320	236	19	may	may	AUX
fcis-20320	236	20	be	be	AUX
fcis-20320	236	21	more	more	ADV
fcis-20320	236	22	helpful	helpful	ADJ
fcis-20320	236	23	for	for	ADP
fcis-20320	236	24	the	the	DET
fcis-20320	236	25	early	early	ADJ
fcis-20320	236	26	diagnosis	diagnosis	NOUN
fcis-20320	236	27	of	of	ADP
fcis-20320	236	28	pulmonary	pulmonary	ADJ
fcis-20320	236	29	nodules	nodule	NOUN
fcis-20320	236	30	.	.	PUNCT
fcis-20320	237	1	res	re	NOUN
fcis-20320	237	2	-	-	PUNCT
fcis-20320	237	3	vgg	vgg	NOUN
fcis-20320	237	4	has	have	VERB
fcis-20320	237	5	the	the	DET
fcis-20320	237	6	best	good	ADJ
fcis-20320	237	7	accuracy	accuracy	NOUN
fcis-20320	237	8	,	,	PUNCT
fcis-20320	237	9	with	with	ADP
fcis-20320	237	10	an	an	DET
fcis-20320	237	11	accuracy	accuracy	NOUN
fcis-20320	237	12	rate	rate	NOUN
fcis-20320	237	13	of	of	ADP
fcis-20320	237	14	84.35	84.35	NUM
fcis-20320	237	15	%	%	NOUN
fcis-20320	237	16	,	,	PUNCT
fcis-20320	237	17	a	a	DET
fcis-20320	237	18	sensitivity	sensitivity	NOUN
fcis-20320	237	19	of	of	ADP
fcis-20320	237	20	83.86	83.86	NUM
fcis-20320	237	21	%	%	NOUN
fcis-20320	237	22	,	,	PUNCT
fcis-20320	237	23	and	and	CCONJ
fcis-20320	237	24	a	a	DET
fcis-20320	237	25	specificity	specificity	NOUN
fcis-20320	237	26	of	of	ADP
fcis-20320	237	27	84.52	84.52	NUM
fcis-20320	237	28	%	%	NOUN
fcis-20320	237	29	.	.	PUNCT
fcis-20320	238	1	the	the	DET
fcis-20320	238	2	accuracy	accuracy	NOUN
fcis-20320	238	3	of	of	ADP
fcis-20320	238	4	vgg16	vgg16	PROPN
fcis-20320	238	5	was	be	AUX
fcis-20320	238	6	83.36	83.36	NUM
fcis-20320	238	7	%	%	NOUN
fcis-20320	238	8	,	,	PUNCT
fcis-20320	238	9	the	the	DET
fcis-20320	238	10	sensitivity	sensitivity	NOUN
fcis-20320	238	11	was	be	AUX
fcis-20320	238	12	82.53	82.53	NUM
fcis-20320	238	13	%	%	NOUN
fcis-20320	238	14	,	,	PUNCT
fcis-20320	238	15	and	and	CCONJ
fcis-20320	238	16	the	the	DET
fcis-20320	238	17	specificity	specificity	NOUN
fcis-20320	238	18	was	be	AUX
fcis-20320	238	19	82.88	82.88	NUM
fcis-20320	238	20	%	%	NOUN
fcis-20320	238	21	.	.	PUNCT
fcis-20320	239	1	the	the	DET
fcis-20320	239	2	reason	reason	NOUN
fcis-20320	239	3	why	why	SCONJ
fcis-20320	239	4	the	the	DET
fcis-20320	239	5	res	re	NOUN
fcis-20320	239	6	-	-	PUNCT
fcis-20320	239	7	vgg	vgg	NOUN
fcis-20320	239	8	model	model	NOUN
fcis-20320	239	9	can	can	AUX
fcis-20320	239	10	achieve	achieve	VERB
fcis-20320	239	11	the	the	DET
fcis-20320	239	12	best	good	ADJ
fcis-20320	239	13	results	result	NOUN
fcis-20320	239	14	is	be	AUX
fcis-20320	239	15	mainly	mainly	ADV
fcis-20320	239	16	because	because	SCONJ
fcis-20320	239	17	it	it	PRON
fcis-20320	239	18	combines	combine	VERB
fcis-20320	239	19	the	the	DET
fcis-20320	239	20	advantages	advantage	NOUN
fcis-20320	239	21	of	of	ADP
fcis-20320	239	22	vgg	vgg	NOUN
fcis-20320	239	23	and	and	CCONJ
fcis-20320	239	24	resnet	resnet	NOUN
fcis-20320	239	25	.	.	PUNCT
fcis-20320	240	1	it	it	PRON
fcis-20320	240	2	not	not	PART
fcis-20320	240	3	only	only	ADV
fcis-20320	240	4	retains	retain	VERB
fcis-20320	240	5	the	the	DET
fcis-20320	240	6	characteristics	characteristic	NOUN
fcis-20320	240	7	of	of	ADP
fcis-20320	240	8	the	the	DET
fcis-20320	240	9	vgg	vgg	ADJ
fcis-20320	240	10	deep	deep	ADJ
fcis-20320	240	11	convolution	convolution	NOUN
fcis-20320	240	12	network	network	NOUN
fcis-20320	240	13	,	,	PUNCT
fcis-20320	240	14	but	but	CCONJ
fcis-20320	240	15	also	also	ADV
fcis-20320	240	16	introduces	introduce	VERB
fcis-20320	240	17	the	the	DET
fcis-20320	240	18	residual	residual	ADJ
fcis-20320	240	19	connection	connection	NOUN
fcis-20320	240	20	of	of	ADP
fcis-20320	240	21	resnet	resnet	NOUN
fcis-20320	240	22	,	,	PUNCT
fcis-20320	240	23	effectively	effectively	ADV
fcis-20320	240	24	solving	solve	VERB
fcis-20320	240	25	the	the	DET
fcis-20320	240	26	depth	depth	NOUN
fcis-20320	240	27	problem	problem	NOUN
fcis-20320	240	28	.	.	PUNCT
fcis-20320	241	1	vanishing	vanish	VERB
fcis-20320	241	2	gradient	gradient	NOUN
fcis-20320	241	3	and	and	CCONJ
fcis-20320	241	4	representation	representation	NOUN
fcis-20320	241	5	bottleneck	bottleneck	NOUN
fcis-20320	241	6	problems	problem	NOUN
fcis-20320	241	7	in	in	ADP
fcis-20320	241	8	networks	network	NOUN
fcis-20320	241	9	.	.	PUNCT
fcis-20320	242	1	in	in	ADP
fcis-20320	242	2	addition	addition	NOUN
fcis-20320	242	3	,	,	PUNCT
fcis-20320	242	4	res	re	NOUN
fcis-20320	242	5	-	-	PUNCT
fcis-20320	242	6	vgg	vgg	NOUN
fcis-20320	242	7	improves	improve	VERB
fcis-20320	242	8	the	the	DET
fcis-20320	242	9	expression	expression	NOUN
fcis-20320	242	10	ability	ability	NOUN
fcis-20320	242	11	of	of	ADP
fcis-20320	242	12	the	the	DET
fcis-20320	242	13	model	model	NOUN
fcis-20320	242	14	by	by	ADP
fcis-20320	242	15	increasing	increase	VERB
fcis-20320	242	16	the	the	DET
fcis-20320	242	17	depth	depth	NOUN
fcis-20320	242	18	and	and	CCONJ
fcis-20320	242	19	width	width	NOUN
fcis-20320	242	20	of	of	ADP
fcis-20320	242	21	the	the	DET
fcis-20320	242	22	network	network	NOUN
fcis-20320	242	23	,	,	PUNCT
fcis-20320	242	24	making	make	VERB
fcis-20320	242	25	it	it	PRON
fcis-20320	242	26	perform	perform	VERB
fcis-20320	242	27	well	well	ADV
fcis-20320	242	28	in	in	ADP
fcis-20320	242	29	the	the	DET
fcis-20320	242	30	task	task	NOUN
fcis-20320	242	31	of	of	ADP
fcis-20320	242	32	classifying	classify	VERB
fcis-20320	242	33	benign	benign	ADJ
fcis-20320	242	34	and	and	CCONJ
fcis-20320	242	35	malignant	malignant	ADJ
fcis-20320	242	36	pulmonary	pulmonary	ADJ
fcis-20320	242	37	nodules	nodule	NOUN
fcis-20320	242	38	.	.	PUNCT
fcis-20320	243	1	the	the	DET
fcis-20320	243	2	network	network	NOUN
fcis-20320	243	3	performance	performance	NOUN
fcis-20320	243	4	of	of	ADP
fcis-20320	243	5	the	the	DET
fcis-20320	243	6	proposed	propose	VERB
fcis-20320	243	7	method	method	NOUN
fcis-20320	243	8	is	be	AUX
fcis-20320	243	9	measured	measure	VERB
fcis-20320	243	10	using	use	VERB
fcis-20320	243	11	the	the	DET
fcis-20320	243	12	roc	roc	PROPN
fcis-20320	243	13	curve	curve	NOUN
fcis-20320	243	14	,	,	PUNCT
fcis-20320	243	15	and	and	CCONJ
fcis-20320	243	16	the	the	DET
fcis-20320	243	17	roc	roc	PROPN
fcis-20320	243	18	curve	curve	NOUN
fcis-20320	243	19	of	of	ADP
fcis-20320	243	20	resvgg	resvgg	NOUN
fcis-20320	243	21	is	be	AUX
fcis-20320	243	22	shown	show	VERB
fcis-20320	243	23	in	in	ADP
fcis-20320	243	24	figure	figure	NOUN
fcis-20320	243	25	5	5	NUM
fcis-20320	243	26	.	.	PUNCT
fcis-20320	243	27	fig	fig	NOUN
fcis-20320	243	28	5	5	NUM
fcis-20320	243	29	.	.	PUNCT
fcis-20320	244	1	the	the	DET
fcis-20320	244	2	structure	structure	NOUN
fcis-20320	244	3	of	of	ADP
fcis-20320	244	4	the	the	DET
fcis-20320	244	5	res	re	NOUN
fcis-20320	244	6	-	-	PUNCT
fcis-20320	244	7	vgg	vgg	ADJ
fcis-20320	244	8	figure	figure	NOUN
fcis-20320	244	9	6	6	NUM
fcis-20320	244	10	provides	provide	VERB
fcis-20320	244	11	an	an	DET
fcis-20320	244	12	example	example	NOUN
fcis-20320	244	13	of	of	ADP
fcis-20320	244	14	benign	benign	ADJ
fcis-20320	244	15	and	and	CCONJ
fcis-20320	244	16	malignant	malignant	ADJ
fcis-20320	244	17	nodule	nodule	NOUN
fcis-20320	244	18	classification	classification	NOUN
fcis-20320	244	19	results	result	NOUN
fcis-20320	244	20	.	.	PUNCT
fcis-20320	245	1	malignant	malignant	ADJ
fcis-20320	245	2	and	and	CCONJ
fcis-20320	245	3	benign	benign	ADJ
fcis-20320	245	4	nodules	nodule	NOUN
fcis-20320	245	5	correctly	correctly	ADV
fcis-20320	245	6	classified	classify	VERB
fcis-20320	245	7	by	by	ADP
fcis-20320	245	8	the	the	DET
fcis-20320	245	9	proposed	propose	VERB
fcis-20320	245	10	method	method	NOUN
fcis-20320	245	11	are	be	AUX
fcis-20320	245	12	shown	show	VERB
fcis-20320	245	13	in	in	ADP
fcis-20320	245	14	the	the	DET
fcis-20320	245	15	first	first	ADJ
fcis-20320	245	16	and	and	CCONJ
fcis-20320	245	17	second	second	ADJ
fcis-20320	245	18	rows	row	NOUN
fcis-20320	245	19	,	,	PUNCT
fcis-20320	245	20	respectively	respectively	ADV
fcis-20320	245	21	,	,	PUNCT
fcis-20320	245	22	and	and	CCONJ
fcis-20320	245	23	the	the	DET
fcis-20320	245	24	numbers	number	NOUN
fcis-20320	245	25	below	below	ADP
fcis-20320	245	26	the	the	DET
fcis-20320	245	27	nodules	nodule	NOUN
fcis-20320	245	28	are	be	AUX
fcis-20320	245	29	the	the	DET
fcis-20320	245	30	predicted	predict	VERB
fcis-20320	245	31	malignant	malignant	ADJ
fcis-20320	245	32	probabilities	probability	NOUN
fcis-20320	245	33	.	.	PUNCT
fcis-20320	246	1	the	the	DET
fcis-20320	246	2	average	average	ADJ
fcis-20320	246	3	degree	degree	NOUN
fcis-20320	246	4	of	of	ADP
fcis-20320	246	5	malignancy	malignancy	NOUN
fcis-20320	246	6	among	among	ADP
fcis-20320	246	7	four	four	NUM
fcis-20320	246	8	expert	expert	NOUN
fcis-20320	246	9	radiologists	radiologist	NOUN
fcis-20320	246	10	was	be	AUX
fcis-20320	246	11	compared	compare	VERB
fcis-20320	246	12	with	with	ADP
fcis-20320	246	13	the	the	DET
fcis-20320	246	14	predicted	predict	VERB
fcis-20320	246	15	probability	probability	NOUN
fcis-20320	246	16	of	of	ADP
fcis-20320	246	17	malignancy	malignancy	NOUN
fcis-20320	246	18	for	for	ADP
fcis-20320	246	19	each	each	DET
fcis-20320	246	20	nodule	nodule	NOUN
fcis-20320	246	21	.	.	PUNCT
fcis-20320	247	1	it	it	PRON
fcis-20320	247	2	was	be	AUX
fcis-20320	247	3	observed	observe	VERB
fcis-20320	247	4	that	that	SCONJ
fcis-20320	247	5	the	the	DET
fcis-20320	247	6	predicted	predict	VERB
fcis-20320	247	7	malignancy	malignancy	NOUN
fcis-20320	247	8	probability	probability	NOUN
fcis-20320	247	9	of	of	ADP
fcis-20320	247	10	each	each	DET
fcis-20320	247	11	nodule	nodule	NOUN
fcis-20320	247	12	was	be	AUX
fcis-20320	247	13	consistent	consistent	ADJ
fcis-20320	247	14	with	with	ADP
fcis-20320	247	15	the	the	DET
fcis-20320	247	16	average	average	ADJ
fcis-20320	247	17	diagnostic	diagnostic	ADJ
fcis-20320	247	18	level	level	NOUN
fcis-20320	247	19	.	.	PUNCT
fcis-20320	248	1	12	12	NUM
fcis-20320	248	2	fig	fig	NOUN
fcis-20320	248	3	6	6	NUM
fcis-20320	248	4	.	.	PUNCT
fcis-20320	248	5	example	example	NOUN
fcis-20320	248	6	of	of	ADP
fcis-20320	248	7	classification	classification	NOUN
fcis-20320	248	8	results	result	NOUN
fcis-20320	248	9	5	5	NUM
fcis-20320	248	10	.	.	X
fcis-20320	248	11	conclusion	conclusion	NOUN
fcis-20320	248	12	this	this	DET
fcis-20320	248	13	paper	paper	NOUN
fcis-20320	248	14	proposes	propose	VERB
fcis-20320	248	15	a	a	DET
fcis-20320	248	16	new	new	ADJ
fcis-20320	248	17	classification	classification	NOUN
fcis-20320	248	18	model	model	NOUN
fcis-20320	248	19	,	,	PUNCT
fcis-20320	248	20	res	re	NOUN
fcis-20320	248	21	-	-	PUNCT
fcis-20320	248	22	vgg	vgg	NOUN
fcis-20320	248	23	,	,	PUNCT
fcis-20320	248	24	which	which	PRON
fcis-20320	248	25	combines	combine	VERB
fcis-20320	248	26	two	two	NUM
fcis-20320	248	27	different	different	ADJ
fcis-20320	248	28	cnn	cnn	PROPN
fcis-20320	248	29	models	model	NOUN
fcis-20320	248	30	to	to	PART
fcis-20320	248	31	classify	classify	VERB
fcis-20320	248	32	benign	benign	ADJ
fcis-20320	248	33	and	and	CCONJ
fcis-20320	248	34	malignant	malignant	ADJ
fcis-20320	248	35	pulmonary	pulmonary	ADJ
fcis-20320	248	36	nodules	nodule	NOUN
fcis-20320	248	37	.	.	PUNCT
fcis-20320	249	1	the	the	DET
fcis-20320	249	2	experimental	experimental	ADJ
fcis-20320	249	3	content	content	NOUN
fcis-20320	249	4	is	be	AUX
fcis-20320	249	5	introduced	introduce	VERB
fcis-20320	249	6	in	in	ADP
fcis-20320	249	7	detail	detail	NOUN
fcis-20320	249	8	,	,	PUNCT
fcis-20320	249	9	including	include	VERB
fcis-20320	249	10	steps	step	NOUN
fcis-20320	249	11	such	such	ADJ
fcis-20320	249	12	as	as	ADP
fcis-20320	249	13	data	data	NOUN
fcis-20320	249	14	preprocessing	preprocessing	NOUN
fcis-20320	249	15	,	,	PUNCT
fcis-20320	249	16	network	network	NOUN
fcis-20320	249	17	training	training	NOUN
fcis-20320	249	18	,	,	PUNCT
fcis-20320	249	19	and	and	CCONJ
fcis-20320	249	20	performance	performance	NOUN
fcis-20320	249	21	evaluation	evaluation	NOUN
fcis-20320	249	22	.	.	PUNCT
fcis-20320	250	1	all	all	DET
fcis-20320	250	2	experiments	experiment	NOUN
fcis-20320	250	3	were	be	AUX
fcis-20320	250	4	conducted	conduct	VERB
fcis-20320	250	5	on	on	ADP
fcis-20320	250	6	the	the	DET
fcis-20320	250	7	luna16	luna16	ADJ
fcis-20320	250	8	database	database	NOUN
fcis-20320	250	9	,	,	PUNCT
fcis-20320	250	10	and	and	CCONJ
fcis-20320	250	11	a	a	DET
fcis-20320	250	12	ten	ten	ADJ
fcis-20320	250	13	-	-	ADJ
fcis-20320	250	14	fold	fold	ADJ
fcis-20320	250	15	cross	cross	ADJ
fcis-20320	250	16	-	-	ADJ
fcis-20320	250	17	validation	validation	ADJ
fcis-20320	250	18	method	method	NOUN
fcis-20320	250	19	was	be	AUX
fcis-20320	250	20	used	use	VERB
fcis-20320	250	21	to	to	PART
fcis-20320	250	22	evaluate	evaluate	VERB
fcis-20320	250	23	the	the	DET
fcis-20320	250	24	performance	performance	NOUN
fcis-20320	250	25	of	of	ADP
fcis-20320	250	26	the	the	DET
fcis-20320	250	27	model	model	NOUN
fcis-20320	250	28	.	.	PUNCT
fcis-20320	251	1	experimental	experimental	ADJ
fcis-20320	251	2	results	result	NOUN
fcis-20320	251	3	show	show	VERB
fcis-20320	251	4	that	that	SCONJ
fcis-20320	251	5	the	the	DET
fcis-20320	251	6	res	re	NOUN
fcis-20320	251	7	-	-	PUNCT
fcis-20320	251	8	vgg	vgg	NOUN
fcis-20320	251	9	model	model	NOUN
fcis-20320	251	10	achieves	achieve	VERB
fcis-20320	251	11	good	good	ADJ
fcis-20320	251	12	performance	performance	NOUN
fcis-20320	251	13	on	on	ADP
fcis-20320	251	14	the	the	DET
fcis-20320	251	15	luna16	luna16	ADJ
fcis-20320	251	16	database	database	NOUN
fcis-20320	251	17	,	,	PUNCT
fcis-20320	251	18	proving	prove	VERB
fcis-20320	251	19	its	its	PRON
fcis-20320	251	20	effectiveness	effectiveness	NOUN
fcis-20320	251	21	in	in	ADP
fcis-20320	251	22	the	the	DET
fcis-20320	251	23	classification	classification	NOUN
fcis-20320	251	24	task	task	NOUN
fcis-20320	251	25	of	of	ADP
fcis-20320	251	26	benign	benign	ADJ
fcis-20320	251	27	and	and	CCONJ
fcis-20320	251	28	malignant	malignant	ADJ
fcis-20320	251	29	pulmonary	pulmonary	ADJ
fcis-20320	251	30	nodules	nodule	NOUN
fcis-20320	251	31	.	.	PUNCT
fcis-20320	252	1	compared	compare	VERB
fcis-20320	252	2	with	with	ADP
fcis-20320	252	3	current	current	ADJ
fcis-20320	252	4	methods	method	NOUN
fcis-20320	252	5	based	base	VERB
fcis-20320	252	6	on	on	ADP
fcis-20320	252	7	traditional	traditional	ADJ
fcis-20320	252	8	features	feature	NOUN
fcis-20320	252	9	and	and	CCONJ
fcis-20320	252	10	deep	deep	ADJ
fcis-20320	252	11	features	feature	NOUN
fcis-20320	252	12	,	,	PUNCT
fcis-20320	252	13	the	the	DET
fcis-20320	252	14	accuracy	accuracy	NOUN
fcis-20320	252	15	of	of	ADP
fcis-20320	252	16	the	the	DET
fcis-20320	252	17	res	re	NOUN
fcis-20320	252	18	-	-	PUNCT
fcis-20320	252	19	vgg	vgg	NOUN
fcis-20320	252	20	model	model	NOUN
fcis-20320	252	21	has	have	AUX
fcis-20320	252	22	been	be	AUX
fcis-20320	252	23	significantly	significantly	ADV
fcis-20320	252	24	improved	improve	VERB
fcis-20320	252	25	.	.	PUNCT
fcis-20320	253	1	in	in	ADP
fcis-20320	253	2	addition	addition	NOUN
fcis-20320	253	3	,	,	PUNCT
fcis-20320	253	4	the	the	DET
fcis-20320	253	5	proposed	propose	VERB
fcis-20320	253	6	model	model	NOUN
fcis-20320	253	7	is	be	AUX
fcis-20320	253	8	compared	compare	VERB
fcis-20320	253	9	with	with	ADP
fcis-20320	253	10	three	three	NUM
fcis-20320	253	11	other	other	ADJ
fcis-20320	253	12	common	common	ADJ
fcis-20320	253	13	classification	classification	NOUN
fcis-20320	253	14	networks	network	NOUN
fcis-20320	253	15	,	,	PUNCT
fcis-20320	253	16	and	and	CCONJ
fcis-20320	253	17	the	the	DET
fcis-20320	253	18	results	result	NOUN
fcis-20320	253	19	show	show	VERB
fcis-20320	253	20	that	that	SCONJ
fcis-20320	253	21	the	the	DET
fcis-20320	253	22	resvgg	resvgg	NOUN
fcis-20320	253	23	model	model	NOUN
fcis-20320	253	24	outperforms	outperform	VERB
fcis-20320	253	25	other	other	ADJ
fcis-20320	253	26	models	model	NOUN
fcis-20320	253	27	due	due	ADP
fcis-20320	253	28	to	to	ADP
fcis-20320	253	29	the	the	DET
fcis-20320	253	30	use	use	NOUN
fcis-20320	253	31	of	of	ADP
fcis-20320	253	32	different	different	ADJ
fcis-20320	253	33	sizes	size	NOUN
fcis-20320	253	34	of	of	ADP
fcis-20320	253	35	convolution	convolution	NOUN
fcis-20320	253	36	kernels	kernel	NOUN
fcis-20320	253	37	and	and	CCONJ
fcis-20320	253	38	maximum	maximum	ADJ
fcis-20320	253	39	pooling	pooling	NOUN
fcis-20320	253	40	and	and	CCONJ
fcis-20320	253	41	average	average	ADJ
fcis-20320	253	42	pooling	pooling	NOUN
fcis-20320	253	43	.	.	PUNCT
fcis-20320	254	1	references	reference	NOUN
fcis-20320	254	2	[	[	X
fcis-20320	254	3	1	1	NUM
fcis-20320	254	4	]	]	PUNCT
fcis-20320	254	5	m.	m.	NOUN
fcis-20320	254	6	marianna	marianna	PROPN
fcis-20320	254	7	sockrider	sockrider	PROPN
fcis-20320	254	8	,	,	PUNCT
fcis-20320	254	9	what	what	PRON
fcis-20320	254	10	is	be	AUX
fcis-20320	254	11	a	a	DET
fcis-20320	254	12	lung	lung	NOUN
fcis-20320	254	13	nodule	nodule	NOUN
fcis-20320	254	14	?	?	PUNCT
fcis-20320	254	15	,	,	PUNCT
fcis-20320	254	16	american	american	ADJ
fcis-20320	254	17	journal	journal	NOUN
fcis-20320	254	18	of	of	ADP
fcis-20320	254	19	respiratory	respiratory	ADJ
fcis-20320	254	20	and	and	CCONJ
fcis-20320	254	21	critical	critical	ADJ
fcis-20320	254	22	care	care	NOUN
fcis-20320	254	23	medicine	medicine	NOUN
fcis-20320	254	24	,	,	PUNCT
fcis-20320	254	25	193	193	NUM
fcis-20320	254	26	(	(	PUNCT
fcis-20320	254	27	2016	2016	NUM
fcis-20320	254	28	)	)	PUNCT
fcis-20320	254	29	i.	i.	NOUN
fcis-20320	255	1	[	[	X
fcis-20320	255	2	2	2	X
fcis-20320	255	3	]	]	X
fcis-20320	255	4	y.	y.	PROPN
fcis-20320	255	5	xie	xie	PROPN
fcis-20320	255	6	,	,	PUNCT
fcis-20320	255	7	y.	y.	PROPN
fcis-20320	255	8	xia	xia	PROPN
fcis-20320	255	9	,	,	PUNCT
fcis-20320	255	10	j.	j.	PROPN
fcis-20320	255	11	zhang	zhang	PROPN
fcis-20320	255	12	,	,	PUNCT
fcis-20320	255	13	y.	y.	PROPN
fcis-20320	255	14	song	song	PROPN
fcis-20320	255	15	,	,	PUNCT
fcis-20320	255	16	d.	d.	PROPN
fcis-20320	255	17	feng	feng	PROPN
fcis-20320	255	18	,	,	PUNCT
fcis-20320	255	19	m.	m.	PROPN
fcis-20320	255	20	fulham	fulham	PROPN
fcis-20320	255	21	,	,	PUNCT
fcis-20320	255	22	w.	w.	PROPN
fcis-20320	255	23	cai	cai	PROPN
fcis-20320	255	24	,	,	PUNCT
fcis-20320	255	25	knowledge	knowledge	NOUN
fcis-20320	255	26	-	-	PUNCT
fcis-20320	255	27	based	base	VERB
fcis-20320	255	28	collaborative	collaborative	ADJ
fcis-20320	255	29	deep	deep	ADJ
fcis-20320	255	30	learning	learning	NOUN
fcis-20320	255	31	for	for	ADP
fcis-20320	255	32	benignmalignant	benignmalignant	ADJ
fcis-20320	255	33	lung	lung	NOUN
fcis-20320	255	34	nodule	nodule	NOUN
fcis-20320	255	35	classification	classification	NOUN
fcis-20320	255	36	on	on	ADP
fcis-20320	255	37	chest	chest	NOUN
fcis-20320	255	38	ct	ct	PROPN
fcis-20320	255	39	,	,	PUNCT
fcis-20320	255	40	ieee	ieee	NOUN
fcis-20320	255	41	transactions	transaction	NOUN
fcis-20320	255	42	on	on	ADP
fcis-20320	255	43	medical	medical	ADJ
fcis-20320	255	44	imaging	imaging	NOUN
fcis-20320	255	45	,	,	PUNCT
fcis-20320	255	46	38	38	NUM
fcis-20320	255	47	(	(	PUNCT
fcis-20320	255	48	2018	2018	NUM
fcis-20320	255	49	)	)	PUNCT
fcis-20320	255	50	991	991	NUM
fcis-20320	255	51	-	-	SYM
fcis-20320	255	52	1004	1004	NUM
fcis-20320	255	53	.	.	PUNCT
fcis-20320	256	1	[	[	X
fcis-20320	256	2	3	3	X
fcis-20320	256	3	]	]	PUNCT
fcis-20320	256	4	s.	s.	PROPN
fcis-20320	256	5	liu	liu	PROPN
fcis-20320	256	6	,	,	PUNCT
fcis-20320	256	7	y.	y.	PROPN
fcis-20320	256	8	xie	xie	PROPN
fcis-20320	256	9	,	,	PUNCT
fcis-20320	256	10	a.	a.	PROPN
fcis-20320	256	11	jirapatnakul	jirapatnakul	PROPN
fcis-20320	256	12	,	,	PUNCT
fcis-20320	256	13	a.p	a.p	PROPN
fcis-20320	256	14	.	.	PROPN
fcis-20320	256	15	reeves	reeves	PROPN
fcis-20320	256	16	,	,	PUNCT
fcis-20320	256	17	pulmonary	pulmonary	ADJ
fcis-20320	256	18	nodule	nodule	NOUN
fcis-20320	256	19	classification	classification	NOUN
fcis-20320	256	20	in	in	ADP
fcis-20320	256	21	lung	lung	NOUN
fcis-20320	256	22	cancer	cancer	NOUN
fcis-20320	256	23	screening	screening	NOUN
fcis-20320	256	24	with	with	ADP
fcis-20320	256	25	three	three	NUM
fcis-20320	256	26	-	-	PUNCT
fcis-20320	256	27	dimensional	dimensional	ADJ
fcis-20320	256	28	convolutional	convolutional	ADJ
fcis-20320	256	29	neural	neural	ADJ
fcis-20320	256	30	networks	network	NOUN
fcis-20320	256	31	,	,	PUNCT
fcis-20320	256	32	journal	journal	NOUN
fcis-20320	256	33	of	of	ADP
fcis-20320	256	34	medical	medical	ADJ
fcis-20320	256	35	imaging	imaging	NOUN
fcis-20320	256	36	,	,	PUNCT
fcis-20320	256	37	4	4	NUM
fcis-20320	256	38	(	(	PUNCT
fcis-20320	256	39	2017	2017	NUM
fcis-20320	256	40	)	)	PUNCT
fcis-20320	256	41	041308	041308	NUM
fcis-20320	256	42	-	-	SYM
fcis-20320	256	43	041308	041308	NUM
fcis-20320	256	44	.	.	PUNCT
fcis-20320	257	1	[	[	X
fcis-20320	257	2	4	4	X
fcis-20320	257	3	]	]	X
fcis-20320	257	4	k.-l	k.-l	PROPN
fcis-20320	257	5	.	.	PUNCT
fcis-20320	258	1	hua	hua	PROPN
fcis-20320	258	2	,	,	PUNCT
fcis-20320	258	3	c.-h	c.-h	PROPN
fcis-20320	258	4	.	.	PUNCT
fcis-20320	259	1	hsu	hsu	PROPN
fcis-20320	259	2	,	,	PUNCT
fcis-20320	259	3	s.c	s.c	PROPN
fcis-20320	259	4	.	.	PROPN
fcis-20320	259	5	hidayati	hidayati	PROPN
fcis-20320	259	6	,	,	PUNCT
fcis-20320	259	7	w.-h	w.-h	NOUN
fcis-20320	259	8	.	.	PUNCT
fcis-20320	260	1	cheng	cheng	PROPN
fcis-20320	260	2	,	,	PUNCT
fcis-20320	260	3	y.-j	y.-j	PROPN
fcis-20320	260	4	.	.	PUNCT
fcis-20320	261	1	chen	chen	PROPN
fcis-20320	261	2	,	,	PUNCT
fcis-20320	261	3	computer	computer	NOUN
fcis-20320	261	4	-	-	PUNCT
fcis-20320	261	5	aided	aid	VERB
fcis-20320	261	6	classification	classification	NOUN
fcis-20320	261	7	of	of	ADP
fcis-20320	261	8	lung	lung	NOUN
fcis-20320	261	9	nodules	nodule	NOUN
fcis-20320	261	10	on	on	ADP
fcis-20320	261	11	computed	compute	VERB
fcis-20320	261	12	tomography	tomography	NOUN
fcis-20320	261	13	images	image	NOUN
fcis-20320	261	14	via	via	ADP
fcis-20320	261	15	deep	deep	ADJ
fcis-20320	261	16	learning	learning	NOUN
fcis-20320	261	17	technique	technique	NOUN
fcis-20320	261	18	,	,	PUNCT
fcis-20320	261	19	oncotargets	oncotarget	NOUN
fcis-20320	261	20	and	and	CCONJ
fcis-20320	261	21	therapy	therapy	NOUN
fcis-20320	261	22	,	,	PUNCT
fcis-20320	261	23	(	(	PUNCT
fcis-20320	261	24	2015	2015	NUM
fcis-20320	261	25	)	)	PUNCT
fcis-20320	261	26	2015	2015	NUM
fcis-20320	261	27	-	-	SYM
fcis-20320	261	28	2022	2022	NUM
fcis-20320	261	29	.	.	PUNCT
fcis-20320	262	1	[	[	X
fcis-20320	262	2	5	5	X
fcis-20320	262	3	]	]	PUNCT
fcis-20320	262	4	w.	w.	PROPN
fcis-20320	262	5	li	li	PROPN
fcis-20320	262	6	,	,	PUNCT
fcis-20320	262	7	p.	p.	PROPN
fcis-20320	262	8	cao	cao	PROPN
fcis-20320	262	9	,	,	PUNCT
fcis-20320	262	10	d.	d.	PROPN
fcis-20320	262	11	zhao	zhao	PROPN
fcis-20320	262	12	,	,	PUNCT
fcis-20320	262	13	j.	j.	PROPN
fcis-20320	262	14	wang	wang	PROPN
fcis-20320	262	15	,	,	PUNCT
fcis-20320	262	16	pulmonary	pulmonary	ADJ
fcis-20320	262	17	nodule	nodule	NOUN
fcis-20320	262	18	classification	classification	NOUN
fcis-20320	262	19	with	with	ADP
fcis-20320	262	20	deep	deep	ADJ
fcis-20320	262	21	convolutional	convolutional	ADJ
fcis-20320	262	22	neural	neural	ADJ
fcis-20320	262	23	networks	network	NOUN
fcis-20320	262	24	on	on	ADP
fcis-20320	262	25	computed	compute	VERB
fcis-20320	262	26	tomography	tomography	NOUN
fcis-20320	262	27	images	image	NOUN
fcis-20320	262	28	,	,	PUNCT
fcis-20320	262	29	computational	computational	ADJ
fcis-20320	262	30	and	and	CCONJ
fcis-20320	262	31	mathematical	mathematical	ADJ
fcis-20320	262	32	methods	method	NOUN
fcis-20320	262	33	in	in	ADP
fcis-20320	262	34	medicine	medicine	NOUN
fcis-20320	262	35	,	,	PUNCT
fcis-20320	262	36	2016	2016	NUM
fcis-20320	262	37	(	(	PUNCT
fcis-20320	262	38	2016	2016	NUM
fcis-20320	262	39	)	)	PUNCT
fcis-20320	262	40	.	.	PUNCT
fcis-20320	263	1	[	[	X
fcis-20320	263	2	6	6	NUM
fcis-20320	263	3	]	]	PUNCT
fcis-20320	263	4	p.	p.	PROPN
fcis-20320	263	5	sahu	sahu	PROPN
fcis-20320	263	6	,	,	PUNCT
fcis-20320	263	7	d.	d.	PROPN
fcis-20320	263	8	yu	yu	PROPN
fcis-20320	263	9	,	,	PUNCT
fcis-20320	263	10	m.	m.	NOUN
fcis-20320	263	11	dasari	dasari	PROPN
fcis-20320	263	12	,	,	PUNCT
fcis-20320	263	13	f.	f.	PROPN
fcis-20320	263	14	hou	hou	PROPN
fcis-20320	263	15	,	,	PUNCT
fcis-20320	263	16	h.	h.	PROPN
fcis-20320	263	17	qin	qin	PROPN
fcis-20320	263	18	,	,	PUNCT
fcis-20320	263	19	a	a	DET
fcis-20320	263	20	lightweight	lightweight	ADJ
fcis-20320	263	21	multisection	multisection	NOUN
fcis-20320	263	22	cnn	cnn	PROPN
fcis-20320	263	23	for	for	ADP
fcis-20320	263	24	lung	lung	NOUN
fcis-20320	263	25	nodule	nodule	NOUN
fcis-20320	263	26	classification	classification	NOUN
fcis-20320	263	27	and	and	CCONJ
fcis-20320	263	28	malignancy	malignancy	NOUN
fcis-20320	263	29	estimation	estimation	NOUN
fcis-20320	263	30	,	,	PUNCT
fcis-20320	263	31	ieee	ieee	NOUN
fcis-20320	263	32	journal	journal	PROPN
fcis-20320	263	33	of	of	ADP
fcis-20320	263	34	biomedical	biomedical	ADJ
fcis-20320	263	35	and	and	CCONJ
fcis-20320	263	36	health	health	NOUN
fcis-20320	263	37	informatics	informatic	NOUN
fcis-20320	263	38	,	,	PUNCT
fcis-20320	263	39	23	23	NUM
fcis-20320	263	40	(	(	PUNCT
fcis-20320	263	41	2018	2018	NUM
fcis-20320	263	42	)	)	PUNCT
fcis-20320	263	43	960	960	NUM
fcis-20320	263	44	-	-	SYM
fcis-20320	263	45	968	968	NUM
fcis-20320	263	46	.	.	PUNCT
fcis-20320	264	1	[	[	X
fcis-20320	264	2	7	7	X
fcis-20320	264	3	]	]	PUNCT
fcis-20320	264	4	k.	k.	NOUN
fcis-20320	265	1	he	he	PROPN
fcis-20320	265	2	,	,	PUNCT
fcis-20320	265	3	x.	x.	PROPN
fcis-20320	265	4	zhang	zhang	PROPN
fcis-20320	265	5	,	,	PUNCT
fcis-20320	265	6	s.	s.	PROPN
fcis-20320	265	7	ren	ren	PROPN
fcis-20320	265	8	,	,	PUNCT
fcis-20320	265	9	j.	j.	PROPN
fcis-20320	265	10	sun	sun	PROPN
fcis-20320	265	11	,	,	PUNCT
fcis-20320	265	12	deep	deep	ADJ
fcis-20320	265	13	residual	residual	ADJ
fcis-20320	265	14	learning	learning	NOUN
fcis-20320	265	15	for	for	ADP
fcis-20320	265	16	image	image	NOUN
fcis-20320	265	17	recognition	recognition	NOUN
fcis-20320	265	18	,	,	PUNCT
fcis-20320	265	19	proceedings	proceeding	NOUN
fcis-20320	265	20	of	of	ADP
fcis-20320	265	21	the	the	DET
fcis-20320	265	22	ieee	ieee	NOUN
fcis-20320	265	23	conference	conference	NOUN
fcis-20320	265	24	on	on	ADP
fcis-20320	265	25	computer	computer	NOUN
fcis-20320	265	26	vision	vision	NOUN
fcis-20320	265	27	and	and	CCONJ
fcis-20320	265	28	pattern	pattern	NOUN
fcis-20320	265	29	recognition	recognition	NOUN
fcis-20320	265	30	,	,	PUNCT
fcis-20320	265	31	2016	2016	NUM
fcis-20320	265	32	,	,	PUNCT
fcis-20320	265	33	pp	pp	ADJ
fcis-20320	265	34	.	.	PUNCT
fcis-20320	266	1	770	770	NUM
fcis-20320	266	2	-	-	SYM
fcis-20320	266	3	778	778	NUM
fcis-20320	266	4	.	.	PUNCT
fcis-20320	267	1	[	[	X
fcis-20320	267	2	8	8	X
fcis-20320	267	3	]	]	PUNCT
fcis-20320	267	4	j.	j.	PROPN
fcis-20320	267	5	gong	gong	PROPN
fcis-20320	267	6	,	,	PUNCT
fcis-20320	267	7	j.	j.	PROPN
fcis-20320	267	8	liu	liu	PROPN
fcis-20320	267	9	,	,	PUNCT
fcis-20320	267	10	w.	w.	PROPN
fcis-20320	267	11	hao	hao	PROPN
fcis-20320	267	12	,	,	PUNCT
fcis-20320	267	13	s.	s.	PROPN
fcis-20320	267	14	nie	nie	PROPN
fcis-20320	267	15	,	,	PUNCT
fcis-20320	267	16	b.	b.	PROPN
fcis-20320	267	17	zheng	zheng	PROPN
fcis-20320	267	18	,	,	PUNCT
fcis-20320	267	19	s.	s.	PROPN
fcis-20320	267	20	wang	wang	PROPN
fcis-20320	267	21	,	,	PUNCT
fcis-20320	267	22	w.	w.	PROPN
fcis-20320	267	23	peng	peng	PROPN
fcis-20320	267	24	,	,	PUNCT
fcis-20320	267	25	a	a	DET
fcis-20320	267	26	deep	deep	ADJ
fcis-20320	267	27	residual	residual	ADJ
fcis-20320	267	28	learning	learning	NOUN
fcis-20320	267	29	network	network	NOUN
fcis-20320	267	30	for	for	ADP
fcis-20320	267	31	predicting	predict	VERB
fcis-20320	267	32	lung	lung	NOUN
fcis-20320	267	33	adenocarcinoma	adenocarcinoma	NOUN
fcis-20320	267	34	manifesting	manifesting	NOUN
fcis-20320	267	35	as	as	ADP
fcis-20320	267	36	ground	ground	NOUN
fcis-20320	267	37	-	-	PUNCT
fcis-20320	267	38	glass	glass	NOUN
fcis-20320	267	39	nodule	nodule	NOUN
fcis-20320	267	40	on	on	ADP
fcis-20320	267	41	ct	ct	NUM
fcis-20320	267	42	images	image	NOUN
fcis-20320	267	43	,	,	PUNCT
fcis-20320	267	44	european	european	ADJ
fcis-20320	267	45	radiology	radiology	NOUN
fcis-20320	267	46	,	,	PUNCT
fcis-20320	267	47	30	30	NUM
fcis-20320	267	48	(	(	PUNCT
fcis-20320	267	49	2020	2020	NUM
fcis-20320	267	50	)	)	PUNCT
fcis-20320	267	51	1847	1847	NUM
fcis-20320	267	52	-	-	SYM
fcis-20320	267	53	1855	1855	NUM
fcis-20320	267	54	.	.	PUNCT
fcis-20320	268	1	[	[	X
fcis-20320	268	2	9	9	NUM
fcis-20320	268	3	]	]	PUNCT
fcis-20320	268	4	x.	x.	NOUN
fcis-20320	268	5	xia	xia	PROPN
fcis-20320	268	6	,	,	PUNCT
fcis-20320	268	7	j.	j.	PROPN
fcis-20320	268	8	gong	gong	PROPN
fcis-20320	268	9	,	,	PUNCT
fcis-20320	268	10	w.	w.	PROPN
fcis-20320	268	11	hao	hao	PROPN
fcis-20320	268	12	,	,	PUNCT
fcis-20320	268	13	t.	t.	PROPN
fcis-20320	268	14	yang	yang	PROPN
fcis-20320	268	15	,	,	PUNCT
fcis-20320	268	16	y.	y.	PROPN
fcis-20320	268	17	lin	lin	PROPN
fcis-20320	268	18	,	,	PUNCT
fcis-20320	268	19	s.	s.	PROPN
fcis-20320	268	20	wang	wang	PROPN
fcis-20320	268	21	,	,	PUNCT
fcis-20320	268	22	w.	w.	PROPN
fcis-20320	268	23	peng	peng	PROPN
fcis-20320	268	24	,	,	PUNCT
fcis-20320	268	25	comparison	comparison	NOUN
fcis-20320	268	26	and	and	CCONJ
fcis-20320	268	27	fusion	fusion	NOUN
fcis-20320	268	28	of	of	ADP
fcis-20320	268	29	deep	deep	ADJ
fcis-20320	268	30	learning	learning	NOUN
fcis-20320	268	31	and	and	CCONJ
fcis-20320	268	32	radiomics	radiomic	NOUN
fcis-20320	268	33	features	feature	NOUN
fcis-20320	268	34	of	of	ADP
fcis-20320	268	35	ground	ground	NOUN
fcis-20320	268	36	-	-	PUNCT
fcis-20320	268	37	glass	glass	NOUN
fcis-20320	268	38	nodules	nodule	NOUN
fcis-20320	268	39	to	to	PART
fcis-20320	268	40	predict	predict	VERB
fcis-20320	268	41	the	the	DET
fcis-20320	268	42	invasiveness	invasiveness	ADJ
fcis-20320	268	43	risk	risk	NOUN
fcis-20320	268	44	of	of	ADP
fcis-20320	268	45	stage	stage	NOUN
fcis-20320	268	46	-	-	PUNCT
fcis-20320	268	47	i	i	PRON
fcis-20320	268	48	lung	lung	NOUN
fcis-20320	268	49	adenocarcinomas	adenocarcinoma	NOUN
fcis-20320	268	50	in	in	ADP
fcis-20320	268	51	ct	ct	PROPN
fcis-20320	268	52	scan	scan	NOUN
fcis-20320	268	53	,	,	PUNCT
fcis-20320	268	54	frontiers	frontier	NOUN
fcis-20320	268	55	in	in	ADP
fcis-20320	268	56	oncology	oncology	NOUN
fcis-20320	268	57	,	,	PUNCT
fcis-20320	268	58	10	10	NUM
fcis-20320	268	59	(	(	PUNCT
fcis-20320	268	60	2020	2020	NUM
fcis-20320	268	61	)	)	PUNCT
fcis-20320	268	62	418	418	NUM
fcis-20320	268	63	.	.	PUNCT
fcis-20320	269	1	[	[	X
fcis-20320	269	2	10	10	NUM
fcis-20320	269	3	]	]	PUNCT
fcis-20320	269	4	p.	p.	PROPN
fcis-20320	269	5	wu	wu	PROPN
fcis-20320	269	6	,	,	PUNCT
fcis-20320	269	7	x.	x.	PROPN
fcis-20320	269	8	sun	sun	PROPN
fcis-20320	269	9	,	,	PUNCT
fcis-20320	269	10	z.	z.	PROPN
fcis-20320	269	11	zhao	zhao	PROPN
fcis-20320	269	12	,	,	PUNCT
fcis-20320	269	13	h.	h.	PROPN
fcis-20320	269	14	wang	wang	PROPN
fcis-20320	269	15	,	,	PUNCT
fcis-20320	269	16	s.	s.	PROPN
fcis-20320	269	17	pan	pan	PROPN
fcis-20320	269	18	,	,	PUNCT
fcis-20320	269	19	b.	b.	PROPN
fcis-20320	269	20	schuller	schuller	PROPN
fcis-20320	269	21	,	,	PUNCT
fcis-20320	269	22	classification	classification	NOUN
fcis-20320	269	23	of	of	ADP
fcis-20320	269	24	lung	lung	NOUN
fcis-20320	269	25	nodules	nodule	NOUN
fcis-20320	269	26	based	base	VERB
fcis-20320	269	27	on	on	ADP
fcis-20320	269	28	deep	deep	ADJ
fcis-20320	269	29	residual	residual	ADJ
fcis-20320	269	30	networks	network	NOUN
fcis-20320	269	31	and	and	CCONJ
fcis-20320	269	32	migration	migration	NOUN
fcis-20320	269	33	learning	learning	NOUN
fcis-20320	269	34	,	,	PUNCT
fcis-20320	269	35	computational	computational	ADJ
fcis-20320	269	36	intelligence	intelligence	NOUN
fcis-20320	269	37	and	and	CCONJ
fcis-20320	269	38	neuroscience	neuroscience	NOUN
fcis-20320	269	39	,	,	PUNCT
fcis-20320	269	40	2020	2020	NUM
fcis-20320	269	41	(	(	PUNCT
fcis-20320	269	42	2020	2020	NUM
fcis-20320	269	43	)	)	PUNCT
fcis-20320	269	44	.	.	PUNCT
fcis-20320	270	1	[	[	X
fcis-20320	270	2	11	11	NUM
fcis-20320	270	3	]	]	PUNCT
fcis-20320	270	4	x.	x.	NOUN
fcis-20320	270	5	li	li	PROPN
fcis-20320	270	6	,	,	PUNCT
fcis-20320	270	7	b.	b.	PROPN
fcis-20320	270	8	hu	hu	PROPN
fcis-20320	270	9	,	,	PUNCT
fcis-20320	270	10	h.	h.	PROPN
fcis-20320	270	11	li	li	PROPN
fcis-20320	270	12	,	,	PUNCT
fcis-20320	270	13	b.	b.	PROPN
fcis-20320	271	1	you	you	PRON
fcis-20320	271	2	,	,	PUNCT
fcis-20320	271	3	application	application	NOUN
fcis-20320	271	4	of	of	ADP
fcis-20320	271	5	artificial	artificial	ADJ
fcis-20320	271	6	intelligence	intelligence	NOUN
fcis-20320	271	7	in	in	ADP
fcis-20320	271	8	the	the	DET
fcis-20320	271	9	diagnosis	diagnosis	NOUN
fcis-20320	271	10	of	of	ADP
fcis-20320	271	11	multiple	multiple	ADJ
fcis-20320	271	12	primary	primary	ADJ
fcis-20320	271	13	lung	lung	NOUN
fcis-20320	271	14	cancer	cancer	NOUN
fcis-20320	271	15	,	,	PUNCT
fcis-20320	271	16	thoracic	thoracic	NOUN
fcis-20320	271	17	cancer	cancer	NOUN
fcis-20320	271	18	,	,	PUNCT
fcis-20320	271	19	10	10	NUM
fcis-20320	271	20	(	(	PUNCT
fcis-20320	271	21	2019	2019	NUM
fcis-20320	271	22	)	)	PUNCT
fcis-20320	271	23	2168	2168	NUM
fcis-20320	271	24	-	-	SYM
fcis-20320	271	25	2174	2174	NUM
fcis-20320	271	26	.	.	PUNCT
fcis-20320	272	1	[	[	X
fcis-20320	272	2	12	12	NUM
fcis-20320	272	3	]	]	X
fcis-20320	272	4	d.r	d.r	PROPN
fcis-20320	272	5	.	.	PROPN
fcis-20320	272	6	baldwin	baldwin	PROPN
fcis-20320	272	7	,	,	PUNCT
fcis-20320	272	8	j.	j.	PROPN
fcis-20320	272	9	gustafson	gustafson	PROPN
fcis-20320	272	10	,	,	PUNCT
fcis-20320	272	11	l.	l.	PROPN
fcis-20320	272	12	pickup	pickup	PROPN
fcis-20320	272	13	,	,	PUNCT
fcis-20320	272	14	c.	c.	PROPN
fcis-20320	272	15	arteta	arteta	PROPN
fcis-20320	272	16	,	,	PUNCT
fcis-20320	272	17	p.	p.	PROPN
fcis-20320	272	18	novotny	novotny	PROPN
fcis-20320	272	19	,	,	PUNCT
fcis-20320	272	20	j.	j.	PROPN
fcis-20320	272	21	declerck	declerck	PROPN
fcis-20320	272	22	,	,	PUNCT
fcis-20320	272	23	t.	t.	PROPN
fcis-20320	272	24	kadir	kadir	PROPN
fcis-20320	272	25	,	,	PUNCT
fcis-20320	272	26	c.	c.	PROPN
fcis-20320	272	27	figueiras	figueiras	PROPN
fcis-20320	272	28	,	,	PUNCT
fcis-20320	272	29	a.	a.	NOUN
fcis-20320	272	30	sterba	sterba	NOUN
fcis-20320	272	31	,	,	PUNCT
fcis-20320	272	32	a.	a.	NOUN
fcis-20320	272	33	exell	exell	PROPN
fcis-20320	272	34	,	,	PUNCT
fcis-20320	272	35	external	external	ADJ
fcis-20320	272	36	validation	validation	NOUN
fcis-20320	272	37	of	of	ADP
fcis-20320	272	38	a	a	DET
fcis-20320	272	39	convolutional	convolutional	ADJ
fcis-20320	272	40	neural	neural	ADJ
fcis-20320	272	41	network	network	NOUN
fcis-20320	272	42	artificial	artificial	ADJ
fcis-20320	272	43	intelligence	intelligence	NOUN
fcis-20320	272	44	tool	tool	NOUN
fcis-20320	272	45	to	to	PART
fcis-20320	272	46	predict	predict	VERB
fcis-20320	272	47	malignancy	malignancy	NOUN
fcis-20320	272	48	in	in	ADP
fcis-20320	272	49	pulmonary	pulmonary	ADJ
fcis-20320	272	50	nodules	nodule	NOUN
fcis-20320	272	51	,	,	PUNCT
fcis-20320	272	52	thorax	thorax	NOUN
fcis-20320	272	53	,	,	PUNCT
fcis-20320	272	54	75	75	NUM
fcis-20320	272	55	(	(	PUNCT
fcis-20320	272	56	2020	2020	NUM
fcis-20320	272	57	)	)	PUNCT
fcis-20320	272	58	306	306	NUM
fcis-20320	272	59	-	-	SYM
fcis-20320	272	60	312	312	NUM
fcis-20320	272	61	.	.	PUNCT
fcis-20320	273	1	[	[	X
fcis-20320	273	2	13	13	NUM
fcis-20320	273	3	]	]	X
fcis-20320	273	4	y.	y.	PROPN
fcis-20320	273	5	ren	ren	PROPN
fcis-20320	273	6	,	,	PUNCT
fcis-20320	273	7	m.-y	m.-y	PROPN
fcis-20320	273	8	.	.	PUNCT
fcis-20320	274	1	tsai	tsai	PROPN
fcis-20320	274	2	,	,	PUNCT
fcis-20320	274	3	l.	l.	PROPN
fcis-20320	274	4	chen	chen	PROPN
fcis-20320	274	5	,	,	PUNCT
fcis-20320	274	6	j.	j.	PROPN
fcis-20320	274	7	wang	wang	PROPN
fcis-20320	274	8	,	,	PUNCT
fcis-20320	274	9	s.	s.	PROPN
fcis-20320	274	10	li	li	PROPN
fcis-20320	274	11	,	,	PUNCT
fcis-20320	274	12	y.	y.	PROPN
fcis-20320	274	13	liu	liu	PROPN
fcis-20320	274	14	,	,	PUNCT
fcis-20320	274	15	x.	x.	PROPN
fcis-20320	274	16	jia	jia	PROPN
fcis-20320	274	17	,	,	PUNCT
fcis-20320	274	18	c.	c.	PROPN
fcis-20320	274	19	shen	shen	PROPN
fcis-20320	274	20	,	,	PUNCT
fcis-20320	274	21	a	a	DET
fcis-20320	274	22	manifold	manifold	ADJ
fcis-20320	274	23	learning	learn	VERB
fcis-20320	274	24	regularization	regularization	NOUN
fcis-20320	274	25	approach	approach	NOUN
fcis-20320	274	26	to	to	PART
fcis-20320	274	27	enhance	enhance	VERB
fcis-20320	274	28	3d	3d	NUM
fcis-20320	274	29	ct	ct	NUM
fcis-20320	274	30	image	image	NOUN
fcis-20320	274	31	-	-	PUNCT
fcis-20320	274	32	based	base	VERB
fcis-20320	274	33	lung	lung	NOUN
fcis-20320	274	34	nodule	nodule	NOUN
fcis-20320	274	35	classification	classification	NOUN
fcis-20320	274	36	,	,	PUNCT
fcis-20320	274	37	international	international	ADJ
fcis-20320	274	38	journal	journal	NOUN
fcis-20320	274	39	of	of	ADP
fcis-20320	274	40	computer	computer	NOUN
fcis-20320	274	41	assisted	assist	VERB
fcis-20320	274	42	radiology	radiology	NOUN
fcis-20320	274	43	and	and	CCONJ
fcis-20320	274	44	surgery	surgery	NOUN
fcis-20320	274	45	,	,	PUNCT
fcis-20320	274	46	15	15	NUM
fcis-20320	274	47	(	(	PUNCT
fcis-20320	274	48	2020	2020	NUM
fcis-20320	274	49	)	)	PUNCT
fcis-20320	274	50	287	287	NUM
fcis-20320	274	51	-	-	SYM
fcis-20320	274	52	295	295	NUM
fcis-20320	274	53	.	.	PUNCT
fcis-20320	275	1	[	[	X
fcis-20320	275	2	14	14	NUM
fcis-20320	275	3	]	]	X
fcis-20320	275	4	s.	s.	PROPN
fcis-20320	275	5	lakshmanaprabu	lakshmanaprabu	PROPN
fcis-20320	275	6	,	,	PUNCT
fcis-20320	275	7	s.n	s.n	PROPN
fcis-20320	275	8	.	.	PROPN
fcis-20320	275	9	mohanty	mohanty	PROPN
fcis-20320	275	10	,	,	PUNCT
fcis-20320	275	11	k.	k.	PROPN
fcis-20320	275	12	shankar	shankar	PROPN
fcis-20320	275	13	,	,	PUNCT
fcis-20320	275	14	n.	n.	PROPN
fcis-20320	275	15	arunkumar	arunkumar	PROPN
fcis-20320	275	16	,	,	PUNCT
fcis-20320	275	17	g.	g.	PROPN
fcis-20320	275	18	ramirez	ramirez	PROPN
fcis-20320	275	19	,	,	PUNCT
fcis-20320	275	20	optimal	optimal	ADJ
fcis-20320	275	21	deep	deep	ADJ
fcis-20320	275	22	learning	learning	NOUN
fcis-20320	275	23	model	model	NOUN
fcis-20320	275	24	for	for	ADP
fcis-20320	275	25	classification	classification	NOUN
fcis-20320	275	26	of	of	ADP
fcis-20320	275	27	lung	lung	NOUN
fcis-20320	275	28	cancer	cancer	NOUN
fcis-20320	275	29	on	on	ADP
fcis-20320	275	30	ct	ct	NUM
fcis-20320	275	31	images	image	NOUN
fcis-20320	275	32	,	,	PUNCT
fcis-20320	275	33	future	future	ADJ
fcis-20320	275	34	generation	generation	NOUN
fcis-20320	275	35	computer	computer	NOUN
fcis-20320	275	36	systems	system	NOUN
fcis-20320	275	37	,	,	PUNCT
fcis-20320	275	38	92	92	NUM
fcis-20320	275	39	(	(	PUNCT
fcis-20320	275	40	2019	2019	NUM
fcis-20320	275	41	)	)	PUNCT
fcis-20320	275	42	374	374	NUM
fcis-20320	275	43	-	-	SYM
fcis-20320	275	44	382	382	NUM
fcis-20320	275	45	.	.	PUNCT
fcis-20320	276	1	[	[	X
fcis-20320	276	2	15	15	NUM
fcis-20320	276	3	]	]	PUNCT
fcis-20320	276	4	q.	q.	NOUN
fcis-20320	276	5	song	song	PROPN
fcis-20320	276	6	,	,	PUNCT
fcis-20320	276	7	l.	l.	PROPN
fcis-20320	276	8	zhao	zhao	PROPN
fcis-20320	276	9	,	,	PUNCT
fcis-20320	276	10	x.	x.	PROPN
fcis-20320	276	11	luo	luo	PROPN
fcis-20320	276	12	,	,	PUNCT
fcis-20320	276	13	x.	x.	PROPN
fcis-20320	276	14	dou	dou	PROPN
fcis-20320	276	15	,	,	PUNCT
fcis-20320	276	16	using	use	VERB
fcis-20320	276	17	deep	deep	ADJ
fcis-20320	276	18	learning	learning	NOUN
fcis-20320	276	19	for	for	ADP
fcis-20320	276	20	classification	classification	NOUN
fcis-20320	276	21	of	of	ADP
fcis-20320	276	22	lung	lung	NOUN
fcis-20320	276	23	nodules	nodule	NOUN
fcis-20320	276	24	on	on	ADP
fcis-20320	276	25	computed	compute	VERB
fcis-20320	276	26	tomography	tomography	NOUN
fcis-20320	276	27	images	image	NOUN
fcis-20320	276	28	,	,	PUNCT
fcis-20320	276	29	journal	journal	NOUN
fcis-20320	276	30	of	of	ADP
fcis-20320	276	31	healthcare	healthcare	PROPN
fcis-20320	276	32	engineering	engineering	PROPN
fcis-20320	276	33	,	,	PUNCT
fcis-20320	276	34	2017	2017	NUM
fcis-20320	276	35	(	(	PUNCT
fcis-20320	276	36	2017	2017	NUM
fcis-20320	276	37	)	)	PUNCT
fcis-20320	276	38	.	.	PUNCT
fcis-20320	277	1	[	[	X
fcis-20320	277	2	16	16	NUM
fcis-20320	277	3	]	]	X
fcis-20320	277	4	v.	v.	PROPN
fcis-20320	277	5	nair	nair	PROPN
fcis-20320	277	6	,	,	PUNCT
fcis-20320	277	7	g.e	g.e	PROPN
fcis-20320	277	8	.	.	PROPN
fcis-20320	277	9	hinton	hinton	PROPN
fcis-20320	277	10	,	,	PUNCT
fcis-20320	277	11	rectified	rectify	VERB
fcis-20320	277	12	linear	linear	ADJ
fcis-20320	277	13	units	unit	NOUN
fcis-20320	277	14	improve	improve	VERB
fcis-20320	277	15	restricted	restrict	VERB
fcis-20320	277	16	boltzmann	boltzmann	PROPN
fcis-20320	277	17	machines	machine	NOUN
fcis-20320	277	18	,	,	PUNCT
fcis-20320	277	19	proceedings	proceeding	NOUN
fcis-20320	277	20	of	of	ADP
fcis-20320	277	21	the	the	DET
fcis-20320	277	22	27th	27th	ADJ
fcis-20320	277	23	international	international	ADJ
fcis-20320	277	24	conference	conference	NOUN
fcis-20320	277	25	on	on	ADP
fcis-20320	277	26	machine	machine	NOUN
fcis-20320	277	27	learning	learning	NOUN
fcis-20320	277	28	(	(	PUNCT
fcis-20320	277	29	icml-10	icml-10	PROPN
fcis-20320	277	30	)	)	PUNCT
fcis-20320	277	31	,	,	PUNCT
fcis-20320	277	32	2010	2010	NUM
fcis-20320	277	33	,	,	PUNCT
fcis-20320	277	34	pp	pp	ADJ
fcis-20320	277	35	.	.	PUNCT
fcis-20320	278	1	807	807	NUM
fcis-20320	278	2	-	-	SYM
fcis-20320	278	3	814	814	NUM
fcis-20320	278	4	.	.	PUNCT
fcis-20320	279	1	[	[	X
fcis-20320	279	2	17	17	NUM
fcis-20320	279	3	]	]	PUNCT
fcis-20320	279	4	k.	k.	PROPN
fcis-20320	279	5	simonyan	simonyan	PROPN
fcis-20320	279	6	,	,	PUNCT
fcis-20320	279	7	a.	a.	NOUN
fcis-20320	279	8	zisserman	zisserman	PROPN
fcis-20320	279	9	,	,	PUNCT
fcis-20320	279	10	very	very	ADV
fcis-20320	279	11	deep	deep	ADJ
fcis-20320	279	12	convolutional	convolutional	ADJ
fcis-20320	279	13	networks	network	NOUN
fcis-20320	279	14	for	for	ADP
fcis-20320	279	15	large	large	ADJ
fcis-20320	279	16	-	-	PUNCT
fcis-20320	279	17	scale	scale	NOUN
fcis-20320	279	18	image	image	NOUN
fcis-20320	279	19	recognition	recognition	NOUN
fcis-20320	279	20	,	,	PUNCT
fcis-20320	279	21	arxiv	arxiv	PROPN
fcis-20320	279	22	preprint	preprint	NOUN
fcis-20320	279	23	arxiv:1409.1556	arxiv:1409.1556	NOUN
fcis-20320	279	24	,	,	PUNCT
fcis-20320	279	25	(	(	PUNCT
fcis-20320	279	26	2014	2014	NUM
fcis-20320	279	27	)	)	PUNCT
fcis-20320	279	28	.	.	PUNCT
fcis-20320	280	1	[	[	X
fcis-20320	280	2	18	18	NUM
fcis-20320	280	3	]	]	X
fcis-20320	280	4	s.g	s.g	PROPN
fcis-20320	280	5	.	.	PROPN
fcis-20320	280	6	armato	armato	PROPN
fcis-20320	280	7	iii	iii	PROPN
fcis-20320	280	8	,	,	PUNCT
fcis-20320	280	9	g.	g.	PROPN
fcis-20320	280	10	mclennan	mclennan	PROPN
fcis-20320	280	11	,	,	PUNCT
fcis-20320	280	12	l.	l.	PROPN
fcis-20320	280	13	bidaut	bidaut	PROPN
fcis-20320	280	14	,	,	PUNCT
fcis-20320	280	15	m.f	m.f	PROPN
fcis-20320	280	16	.	.	PROPN
fcis-20320	280	17	mcnitt‐gray	mcnitt‐gray	PROPN
fcis-20320	280	18	,	,	PUNCT
fcis-20320	280	19	c.r	c.r	PROPN
fcis-20320	280	20	.	.	PROPN
fcis-20320	280	21	meyer	meyer	PROPN
fcis-20320	280	22	,	,	PUNCT
fcis-20320	280	23	a.p	a.p	PROPN
fcis-20320	280	24	.	.	PROPN
fcis-20320	280	25	reeves	reeves	PROPN
fcis-20320	280	26	,	,	PUNCT
fcis-20320	280	27	b.	b.	PROPN
fcis-20320	280	28	zhao	zhao	PROPN
fcis-20320	280	29	,	,	PUNCT
fcis-20320	280	30	d.r	d.r	PROPN
fcis-20320	280	31	.	.	PROPN
fcis-20320	280	32	aberle	aberle	PROPN
fcis-20320	280	33	,	,	PUNCT
fcis-20320	280	34	c.i	c.i	PROPN
fcis-20320	280	35	.	.	PROPN
fcis-20320	280	36	henschke	henschke	PROPN
fcis-20320	280	37	,	,	PUNCT
fcis-20320	280	38	e.a	e.a	PROPN
fcis-20320	280	39	.	.	PROPN
fcis-20320	280	40	hoffman	hoffman	PROPN
fcis-20320	280	41	,	,	PUNCT
fcis-20320	280	42	the	the	DET
fcis-20320	280	43	lung	lung	NOUN
fcis-20320	280	44	image	image	NOUN
fcis-20320	280	45	database	database	NOUN
fcis-20320	280	46	consortium	consortium	PROPN
fcis-20320	280	47	(	(	PUNCT
fcis-20320	280	48	lidc	lidc	NOUN
fcis-20320	280	49	)	)	PUNCT
fcis-20320	280	50	and	and	CCONJ
fcis-20320	280	51	image	image	NOUN
fcis-20320	280	52	database	database	NOUN
fcis-20320	280	53	resource	resource	NOUN
fcis-20320	280	54	initiative	initiative	NOUN
fcis-20320	280	55	(	(	PUNCT
fcis-20320	280	56	idri	idri	ADJ
fcis-20320	280	57	):	):	PUNCT
fcis-20320	280	58	a	a	DET
fcis-20320	280	59	completed	complete	VERB
fcis-20320	280	60	reference	reference	NOUN
fcis-20320	280	61	database	database	NOUN
fcis-20320	280	62	of	of	ADP
fcis-20320	280	63	lung	lung	NOUN
fcis-20320	280	64	nodules	nodule	NOUN
fcis-20320	280	65	on	on	ADP
fcis-20320	280	66	ct	ct	PROPN
fcis-20320	280	67	scans	scan	NOUN
fcis-20320	280	68	,	,	PUNCT
fcis-20320	280	69	medical	medical	ADJ
fcis-20320	280	70	physics	physics	NOUN
fcis-20320	280	71	,	,	PUNCT
fcis-20320	280	72	38	38	NUM
fcis-20320	280	73	(	(	PUNCT
fcis-20320	280	74	2011	2011	NUM
fcis-20320	280	75	)	)	PUNCT
fcis-20320	280	76	915	915	NUM
fcis-20320	280	77	-	-	SYM
fcis-20320	280	78	931	931	NUM
fcis-20320	280	79	.	.	PUNCT
fcis-20320	281	1	[	[	X
fcis-20320	281	2	19	19	NUM
fcis-20320	281	3	]	]	PUNCT
fcis-20320	281	4	g.	g.	PROPN
fcis-20320	281	5	zhang	zhang	PROPN
fcis-20320	281	6	,	,	PUNCT
fcis-20320	281	7	z.	z.	PROPN
fcis-20320	281	8	yang	yang	PROPN
fcis-20320	281	9	,	,	PUNCT
fcis-20320	281	10	l.	l.	PROPN
fcis-20320	281	11	gong	gong	PROPN
fcis-20320	281	12	,	,	PUNCT
fcis-20320	281	13	s.	s.	PROPN
fcis-20320	281	14	jiang	jiang	PROPN
fcis-20320	281	15	,	,	PUNCT
fcis-20320	281	16	l.	l.	PROPN
fcis-20320	281	17	wang	wang	PROPN
fcis-20320	281	18	,	,	PUNCT
fcis-20320	281	19	h.	h.	PROPN
fcis-20320	281	20	zhang	zhang	PROPN
fcis-20320	281	21	,	,	PUNCT
fcis-20320	281	22	classification	classification	NOUN
fcis-20320	281	23	of	of	ADP
fcis-20320	281	24	lung	lung	NOUN
fcis-20320	281	25	nodules	nodule	NOUN
fcis-20320	281	26	based	base	VERB
fcis-20320	281	27	on	on	ADP
fcis-20320	281	28	ct	ct	NUM
fcis-20320	281	29	images	image	NOUN
fcis-20320	281	30	using	use	VERB
fcis-20320	281	31	squeeze	squeeze	NOUN
fcis-20320	281	32	-	-	PUNCT
fcis-20320	281	33	and	and	CCONJ
fcis-20320	281	34	-	-	PUNCT
fcis-20320	281	35	excitation	excitation	NOUN
fcis-20320	281	36	network	network	NOUN
fcis-20320	281	37	and	and	CCONJ
fcis-20320	281	38	aggregated	aggregate	VERB
fcis-20320	281	39	residual	residual	ADJ
fcis-20320	281	40	transformations	transformation	NOUN
fcis-20320	281	41	,	,	PUNCT
fcis-20320	281	42	la	la	X
fcis-20320	281	43	radiologia	radiologia	NOUN
fcis-20320	281	44	medica	medica	PROPN
fcis-20320	281	45	,	,	PUNCT
fcis-20320	281	46	125	125	NUM
fcis-20320	281	47	(	(	PUNCT
fcis-20320	281	48	2020	2020	NUM
fcis-20320	281	49	)	)	PUNCT
fcis-20320	281	50	374	374	NUM
fcis-20320	281	51	-	-	SYM
fcis-20320	281	52	383	383	NUM
fcis-20320	281	53	.	.	PUNCT