Highlights in BioScience ISSN:2682-4043 DOI:10.36462/H.BioSci.202108 Research Article Open Access 1 Electronic Engineering Department, Federal University of Rio de Janeiro, Rio de Janeiro, Brazil. 2 Department of Electrical Engineering, Hakim Sabzevari University, Sabzevar, Iran. 3 City University Health Sciences Center, Federal University of Rio de Janeiro, Biochemistry, Rio de Janeiro, Brazil. 4 Department of Electrical Engineering, Azad University, Sabzevar, Iran. Contacts of authors * To whom correspondence should be addressed: h.r.khezri2@gmail.com Received: March 23, 2021 Accepted: June 15, 2021 Published: August 3, 2021 Citation: Khezri H, Farzaneh M, Ghasem- ishahrestani Z, Moghadam AP . ANN-based diagnosis method for skin cancers using dermo- scopic images. 2021 Aug 3;4:bs202108 Copyright: © 2021 Khezri et al.. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the paper and supplementary materials. Funding: The authors have no support or funding to report. Competing interests: The authors declare that they have no competing interests. ANN-based diagnosis method for skin cancers using dermoscopic images Hamidreza Khezri*1 >< , Mojtaba Farzaneh2 >< , Zeinab Ghasemishahrestani3 >< , Ali Pakizeh Moghadam4 >< Abstract Melanoma is one of the most dangerous skin cancers in the world. It accounts for 55% of all deaths associated with skin cancer. Researchers believe that skin cancer increases the risk of other cancers if not diagnosed early. Therefore, prompt and timely diagnosis of this disease is very important for the successful treatment of the patient. This system can detect melanoma lethal carcinoma from other skin lesions without the need for surgery, with a low cost, accuracy of about 98.88% and specificity 99%. In this article, a new, intelligent and accurate software (Delphi) system has been used to diagnose melanoma skin cancer. To detect malignant melanoma, the ABCDT rule, asymmetry (A), boundary (B), color (C), diameter (D) and textural variation (T) of the lesion are calculated and finally, an artificial neural network (ANN) is used to obtain an accurate result. The ANN with Multi-Layer Perceptron (MLP) contains the five extraction Characteristics (ABCDT) of lesions is used as inputs, two hidden layers, and two outputs. Very good results were obtained using this method. It was observed that for a dataset of 180 dermoscopic lesion images including 80 malignant melanomas, 20 benign melanomas and 80 nevus lesions. Due to its automatic recognition and ability to be installed on a computer, this system can be very useful for dermatologists as well as the general public. Keywords: Melanoma skin cancer, ABCDT rule, Feature extraction, Artificial neural network Introduction Unfortunately, skin cancer has become very popular these days. Melanoma is caused by ge- netic mutations in the pigment-producing cells called melanocytes [1]. Melanoma affects mostly both sexes of the Caucasian population [2], and the prognosis of the disease becomes very poor at the metastatic stage [3-4]. There are no effective treatments for metastatic melanoma [5], and has grown rapidly over the past 30 years [6]. According to clinical definitions, malignant lesions are not regularly [6]. Figure Figure 1 shows the symmetric and asymmetric lesions. Diagnosis of melanoma in the early stages of the disease can certainly prevent the death of patients. Usually for two reasons skin lesions turn from benign to malignant: First, lack of attention to the skin lesions on body surface. Second, high similarity of skin lesions features and inaccessibility to a dermatologist. For example, Figure 2 shows two very similar skin lesions, malignant melanoma and Clark’s nevus, which is a benign skin lesion. Melanoma is the most common skin disease that can lead to death [7], which often begins with malignant pigment cell tumors that cause more than 70% of deaths in patients with skin cancer [8-9]. It should be noted that if skin cancer is not diagnosed in the early stages, it can affect different parts of the body, including the liver, bones, lungs and brain, and makes the treatment process very difficult and complicated. The need for an automatic and accurate device for reduce unnecessary biopsy and rapid detection is quite clear. Therefore, a method for early de- tection of melanoma is very useful and valuable [10-11] so that dermatologists can use this system to diagnose skin lesions with high accuracy. In fact, melanoma is evaluated by clinical imaging. Dermoscopy is often used to assess melanoma lesions, a non-invasive type of image analysis. A new approach is shown here, which examines the skin lesion image by a trained neural network to analyze if it is benign or malignant. The article is arranged as follows. Highlights in BioScience Page 1 of 7 August 2021|Volume 4 https://doi.org/10.36462/H.BioSci.202108 https://creativecommons.org/licenses/by/4.0/ h.r.khezri2@gmail.com https://orcid.org/0000-0002-5839-3690 mojtaba.farzaneh@gmail.com https://orcid.org/0000-0002-6037-6486 farzaneh.ghasemi.sha@gmail.com https://orcid.org/0000-0002-2482-4296 a.pakizehmoghadam@gmail.com http://bioscience.highlightsin.org/ Khezri et al., 2021 ANN-based diagnosis method for skin cancers using dermoscopic images Figure 1. A:Symmetric and B: Asymmetric melanoma lesions. Figure 2. Clark nevus. The second section describes previous work on melanoma analysis in skin images. In the third section, the new contri- bution to this article describes the analysis, feature extraction, classification, ANN, details, tests and evaluations. Section IV presents the conclusions. As a basic step towards computer-aided skin cancers, auto- matic diagnosis, and image analysis have often been studied in the literature [12-19]. In the last few years, many studies have been done to detect melanoma from skin images with an accu- racy of 70% to 96%. Telemedicine techniques have been studied as a source for the diagnosis of skin lesions. Compared to physi- cians in-person diagnoses (face to face) and telemedicine diag- noses (remote detection), tests on skin diseases have shown that the use of tele- communications technology diagnosis (teleder- matology) is more effective and safer. These techniques include the benefits of easy access, low cost, and quick and accurate ac- cess to treatment results [20-22]. Melanocytic cutaneous lesions have been reported to be the deadliest among the three skin cancer outbreaks and the second most common among adults aged 15 to 29 years [2]. Melanoma is less common in Asia, Africa and Latin America than in Aus- tralia, Europe, North America and New Zealand. Melanomas sometimes change in appearance, including changes in size, ir- regular edges, and discoloration, itching or fracture of the skin [23]. In fact, melanoma can rarely occur in the mouth, intestines, or eyes, but is most commonly found on the skin. It is common in men and women in the back and legs, respectively [24]. The automatic detection of asymmetry in digital images was proposed by the Stolz technique based on ABCD rule [25]. A study on the asymmetry using imaging techniques to identify melanoma skin lesions was presented by Ravichandran et al [26]. To date, many researchers are working on image process- ing, visual techniques, and various melanoma parameters such as size, shape, asymmetry, border, color, and diameter to detect skin cancer [27-31]. One of the known methods is the ABCD rule. The algorithm for detection is divided into four steps: asym- metry, border, color, and diameter. ABCD is a fast learning, calculation and a reliable way to diagnose melanoma [32-34]. Lesion irregularity, borders, colors and diameters can be ana- lyzed and calculated by dividing the image of the lesion into sub-images and extracting the properties of each image [35]. Fi- nally, melanoma could be detected by a simple threshold for the values obtained from the extracted features (lesion irregularity, borders, colors and diameter) [36]. These features are used as inputs to the first layer of the ANN [37-38]. We’ve developed the novel ABCDT rule by improve the ABCD, for automatic diagnosis of skin cancer with greater accuracy and precision. In other words, in the current study, the extraction of features is done based on the ABCDT rule in dermoscopy. Materials and Methods This article aims to develop a new, intelligent and accurate soft- ware system for skin cancer diagnosis using neural network and ABCDT rule. The input of the device is images of skin le- sions. This system with Pre-processing, ABCDT rule, and sep- aration, extract appropriate features from the image. To get the total dermoscopic score (TDS), for each of the "asymmetries, boundary, colors, diameters and texture changes", a coefficient is determined by which the TDS can be calculated. In other words, to obtain TDS (Table 1, the score of each "ABCDT" is multiplied by a specific weight factor. Finally, us- Highlights in BioScience Page 2 of 7 August 2021|Volume 4 http://bioscience.highlightsin.org/ Khezri et al., 2021 ANN-based diagnosis method for skin cancers using dermoscopic images ing NN as a smart medical decision-making system based on TDS, the type of lesion is determined to be melanoma or benign. In other words, an NN has been used to implement the new au- tomated classification of melanocytic lesions. Table 1. The Proposed diagnostic criteria. Diagnostic Criteria Score Weight Factor Asymmetric (A) 0-5 1.3 Border (B) 0-8 0.5 Color variation (C) 1-6 1 Diameter (D) 0-1 0.1 Textural variation (T) 1-10 0.9 In this work, briefly, after the detection and elimination of noise and hair on the image, ABCDT rule, Bayes learning algo- rithm, and the neural network method (Feed Forward Back Prop- agation) were used to detect the lesion is classified as benign or malignant. AS shown in Figures 3 and 3, the architecture of the skin cancer smart system used in this study consists of the five steps of pre-processing, segmentation, feature extraction, classi- fication and diagnosis as follows: • Pre-processing: It involves filtering and contrast enhance- ment techniques using the Retinex algorithm. Resize Pic- tures, to analyze and compare data with the bank, all im- ages are trans- formed into the same size. This size is 450×350 selected. • Blurring and segmentation: The purpose of blurring is to reduce noise. If we use edge detection algorithms for high- resolution images, we will find many results that we are not interested in. Conversely, if we blur the images too much, we will lose data. So, we have to find the amount of blurring we want to use without destroying the desired edges. There are various techniques to achieve blurred effects, but Gaussian blur is used with a factor of 2 to re- move the noise and the hairs on the skin. So, first, the im- age is converted to binary format. Using filters, detect and eliminate hair and noise from the image. Then, the lesion image is completely separated from the background. In other words, the exact location of the lesion on the image is determined by calculating the threshold and statistical characteristics. • Feature extraction: ABCDT rule examines the character- istics of a lesion. These properties include asymmetry, border, color, diameter, and textural variation. This rule is a development of the known ABCD rule [39], commonly used to diagnose melanoma from images. The extracted properties are fed to the first layer of the NN. Properties used to characterize the asymmetry of the lesion The asymmetric feature of the lesion is one of the important features in diagnosis. Natural moles are usually symmetrical. Asymmetry is usually calculated in two ways: entropy and bi- fold. To calculate the asymmetry score, each lesion is examined by two 90-degree axes, and the ANN determines its score. If the lesion is properly symmetrical on both axes, this score is 0, and if it is only on one axis, it will be 2.5. In case of asymmetry in both axes, the score is 5. Finally, the asymmetry score must be multiplied by 1.3 as the weight factor. Features used for irregular characterization at the lesion border Uneven or irregularly shaped margins increase the likelihood of some kind of skin lesion. To calculate boundary irregularities, the lesion is divided into eight sections. If the entire border of the lesion has a severe incision, it is given a maximum score. Otherwise the minimum score of 0 is given. The minimum and maximum score of B are defined 0 and 8, respectively with the 0.5 weight factor. Features used to characterize lesion color variation Color properties are calculated between six colors, and each color represents 1 point. Properties used for diameter According to our information, if the lesion diameter is larger than 0.6 mm, the risk of cancer is higher. Diameter with a weight coefficient of 0.1 is measured by converting the total number of pixels in the largest diameter to millimeters (mm). Characteristics used to characterize the textural variation of the lesion Since healthy skin is reddish, and skin lesions have more textural variation and lower pixel intensity than healthy areas of the skin. For this purpose, first, a low pass filter was used to normalize the color and then extracted the correct features [40]. The textural variation values are considered from 1 to 10 with a weight factor of 0.9. Classification The different features extracted from the lesion surface have different weights. The weight of each group was drawn based on the experience of dermatologists. Table 1 shows the impor- tance of each group. The total dermoscopy score (TDS) can be calculated using eq. (1). T DS = 1.3 × A + 0.5 × B + 1 × C + 0.1 × D + 0.9 × T (1) Where: A, B, C, D, and T Scores are for the asymmetry feature, the border irregularity, the color feature, the diameter size feature, and the textural variation, respectively. Highlights in BioScience Page 3 of 7 August 2021|Volume 4 http://bioscience.highlightsin.org/ Khezri et al., 2021 ANN-based diagnosis method for skin cancers using dermoscopic images Figure 3. A flowchart illustrating the proposed machine learning system for detecting skin cancers using dermoscopic images. Figure 4. Clark nevus. Diagnosis If individual scores of asymmetry, boundary, color, diameter, and textural variation are multiplied by weight- factor of 1.3, 0.5, 1, 0.1, and 0.9, respectively, a precise distinction can be made between benign and malignant melanocytic lesions. The TDS greater than 5.40 means a cancerous lesion, otherwise, it is benign. Results and Discussion All of these steps and the final decision (benign or malignant) are performed by a trained neural network. Various classifications have been performed by researchers for the ANNs [43-46]. In this work, the ANN with a multi-layer perceptron (MLP) utilizes a supervised learning technique and includes features extracted as inputs to the input layer with two hidden layers containing 10 and 7 neurons for each layer (Figure 5). The MLP uses a region-oriented hybrid algorithm, a method called elliptical symmetry to determine asymmetry, a Gaussian smoothing to measure boundary irregularities, and a threshold method for the lesion segment. Therefore, the system designed to diagnose melanoma uses five features of the lesion. The images used in this work were taken from the interna- tional skin imaging collaboration (ISIC) [41]. Out of this dataset, 180 dermoscopic lesion images including 80 malignant mela- no- mas, 20 benign melanomas and 80 nevus lesions were ex- tracted and preprocessed for this research. In several studies [42-44] they used ABCD rule to detect and analyze pictures, whereas in our study we added one more factor (ABCDT) to increase the efficiency and get the better result. We found out The T factor has a huge impact on increasing accuracy. Highlights in BioScience Page 4 of 7 August 2021|Volume 4 http://bioscience.highlightsin.org/ Khezri et al., 2021 ANN-based diagnosis method for skin cancers using dermoscopic images Figure 5. The neural network diagram of the proposed MLP model. Given the weight factor (0.9) and values of the (0-10) change intended for T, the number obtained from T has a great effect on the result, and this is the most important difference between this work and other similar tasks. According to the accuracy (ACC) and specificity (SPEC) ob- tained and compared with Pennisi et al. [47], Fan et al. [48], Jahanifar et al. [49] and Sreelatha et al. [50] can be claimed that the average performance of the technique used in this study is better than previous techniques (Table 2). Table 2. The Average performance evaluation metrics (%). Algorithm ACC SPEC Pennisi [47] 89.40 97.10 Fan [48] 93.60 - Jahanifar [49] 97.90 98.20 Sreelatha [50] 98.64 99.22 Proposed MLP 98.88 99.00 The proposed method for the diagnosis of melanoma skin cancer by ABCDT method revealed 98.88% accuracy and speci- ficity 99%. This approach is safe, accessible, effective, non- invasive and based on the principles of telemedicine with high ac- curacy and reasonable price. With this software system, people can make an early diag- nosis of their skin lesions without referring to a physician and specialists can use it as an intelligent, fast and accurate assistant. In summary, the ABCDT rule, Bayes learning algorithm, and neural network method were used to detect the type of carcino- genic or non-cancerous lesions. According to TDS, it is a clear fact that the three features of asymmetry, color, and textural vari- ation of the lesion are crucial in the diagnosis of melanoma from benign lesions. A TDS value above 5.40 indicates melanoma. In ABCDT method, all steps perform by a fully automated neural network. References 1. Tolleson WH. Human melanocyte biology, toxicology, and pathol- ogy. Journal of Environmental Science and Health Part C. 2005 Jul 1;23(2):105-61. 2. Ferlay J, Soerjomataram I, Dikshit R, Eser S, Mathers C, Re- belo M, Parkin DM, Forman D, Bray F. Cancer incidence and mortality worldwide: sources, methods and major patterns in GLOBOCAN 2012. International journal of cancer. 2015 Mar 1;136(5):E359-86. 3. Bombelli FB, Webster CA, Moncrieff M, Sherwood V. The scope of nanoparticle therapies for future metastatic melanoma treat- ment. The lancet oncology. 2014 Jan 1; 15(1):e22-32. Highlights in BioScience Page 5 of 7 August 2021|Volume 4 http://bioscience.highlightsin.org/ Khezri et al., 2021 ANN-based diagnosis method for skin cancers using dermoscopic images 4. Batus M, Waheed S, Ruby C, Petersen L, Bines SD, Kaufman HL. Optimal management of metastatic melanoma: current strate- gies and future directions. American journal of clinical derma- tology. 2013 Jun; 14(3):179-94. 5. Kaufman HL, Margolin K, Sullivan R. Management of metastatic melanoma in 2018. JAMA oncology. 2018 Jun 1; 4(6):857-8. 6. Balch CM, Gershenwald JE, Soong SJ, Thompson JF, Atkins MB, Byrd DR, Buzaid AC, Cochran AJ, Coit DG, Ding S, Eg- germont AM. Final version of 2009 AJCC melanoma staging and classification. Journal of clinical oncology. 2009 Dec 20;27 (36):6199. 7. Celebi ME, Kingravi HA, Uddin B, Iyatomi H, Aslandogan YA, Stoecker WV, Moss RH. A methodological approach to the clas- sification of dermoscopy images. Computerized Medical imag- ing and graphics. 2007 Sep 1;31(6):362-73. 8. Lau HT, Al-Jumaily A. Automatically early detection of skin cancer: Study based on nueral netwok classification. In2009 International Conference of Soft Computing and Pattern Recog- nition 2009 Dec 4 (pp. 375-380). IEEE. 9. Yuan X, Yang Z, Zouridakis G, Mullani N. SVM-based texture classification and application to early melanoma detection. In2006 International Conference of the IEEE Engineering in Medicine and Biology Society 2006 Aug 30 (pp. 4775- 4778). IEEE. 10. Whited JD. Teledermatology research review. International jour- nal of dermatology. 2006 Mar;45(3):220-9. 11. Ganster H, Pinz P, Rohrer R, Wildling E, Binder M, Kittler H. Automated melanoma recognition. IEEE transactions on medi- cal imaging. 2001 Mar;20(3):233-9. 12. Rose VL. Cancer facts and figures. American Family Physician. 1999 Mar 15;59(6):1697. 13. Shrestha B, Bishop J, Kam K, Chen X, Moss RH, Stoecker WV, Umbaugh S, Stanley RJ, Celebi ME, Marghoob AA, Argenziano G. Detection of atypical texture features in early malignant melanoma. Skin Research and Technology. 2010 Feb ;16(1):60-5. 14. Sadeghi M, Razmara M, Lee TK, Atkins MS. A novel method for detection of pigment network in dermoscopic images using graphs. Computerized Medical Imaging and Graphics. 2011 Mar 1;35(2):137-43. 15. Sadeghi M, Razmara M, Wighton P, Lee TK, Atkins MS. Model- ing the dermoscopic structure pigment network using a clinically inspired feature set. InInternational Workshop on Medical Imag- ing and Virtual Reality 2010 Sep 19 (pp. 467-474). Springer, Berlin, Heidelberg. 16. Anantha M, Moss RH, Stoecker WV. Detection of pigment net- work in dermatoscopy images using texture analysis. Computer- ized Medical Imaging and Graphics. 2004 Jul 1;28(5):225-34. 17. Betta G, Di Leo G, Fabbrocini G, Paolillo A, Sommella P. Der- moscopic image-analysis system: estimation of atypical pigment network and atypical vascular pattern. InIEEE International Work- shop on Medical Measurement and Applications, 2006. MeMea 2006. 2006 Apr 20 (pp. 63-67). IEEE. 18. Celebi ME, Iyatomi H, Stoecker WV, Moss RH, Rabinovitz HS, Argenziano G, Soyer HP. Automatic detection of blue-white veil and related structures in dermoscopy images. Computerized Med- ical Imaging and Graphics. 2008 Dec 1;32(8):670-7. 19. Shitara D, Nascimento M, Ishioka P, Carrera C, Alos L, Malvehy J, Puig S. Dermoscopy of naevus-associated melanomas. Acta dermato-venereologica. 2015 Jun 1;95(6):671-5. 20. Barata C, Marques JS, Rozeira J. Detecting the pigment network in dermoscopy images: a directional approach. In2011 Annual International Conference of the IEEE Engineering in Medicine and Biology Society 2011 Sep (pp. 5120-5123). IEEE. 21. Korotkov K, Garcia R. Computerized analysis of pigmented skin lesions: a review. Artificial intelligence in medicine. 2012 Oct 1;56(2):69-90. 22. Balch CM, Buzaid AC, Soong SJ, Atkins MB, Cascinelli N, Coit DG, Fleming ID, Gershenwald JE, Houghton Jr A, Kirkwood JM, McMasters KM. Final version of the American Joint Com- mittee on Cancer staging system for cutaneous melanoma. Jour- nal of Clinical Oncology. 2001 Aug 15;19(16):3635-48. 23. Freedberg KA, Geller AC, Miller DR, Lew RA, Koh HK. Screen- ing for malignant melanoma: a cost-effectiveness analysis. Jour- nal of the American Academy of Dermatology. 1999 Nov 1;41(5) :738-45. 24. Rigel DS, Friedman RJ, Kopf AW. The incidence of malignant melanoma in the United States: issues as we approach the 21st century. Journal of the American Academy of Dermatology. 1996 May 1;34(5):839-47. 25. Intraocular B. Melanoma Treatment (PDQ): Health Professional Version. PDQ Cancer Information Summaries. 2015. 26. Wild C, Stewart BW. World cancer report 2014. Wild CP, Stew- art BW, editors. Geneva, Switzerland: World Health Organiza- tion; 2014. 27. Cudek P, Grzymaa-Busse JW, Hippe ZS. Further research on au- tomatic estimation of asymmetry of melanocytic skin lesions. In- HumanComputer Systems Interaction: Backgrounds and Appli- cations 2 2012 (pp. 125-129). Springer, Berlin, Heidelberg. 28. Premaladha J, Ravichandran KS. Asymmetry analysis of malig- nant melanoma using image processing: a survey. Journal of Artificial Intelligence. 2014 Apr 1;7(2):45. 29. Jain S, Pise N. Computer aided melanoma skin cancer detection using image processing. Procedia Computer Science. 2015 Jan 1;48:735-40. 30. Ananthi B, Balamohan S, Hemalatha M. Melanoma detection us- ing RGB color model in medical imaging. Middle-East Journal of Scientific Research. 2014;21(11):1982-7. 31. Iqbal S, Sophia M, Divyashree J, Mundas M, Vidya R. Imple- mentation of supervised learning for melanoma detection using image processing. International Journal of Research in Engineer- ing and Technology. 2015;4(6):325-9. 32. Grammatikopoulos G, Hatzigaidas A, Papastergiou A, Lazaridis P, Zaharis Z, Kampitaki D, Tryfon G. Automated malignant mela- noma detection using Matlab. InProc. Fifth Int. Conf. on Data Networks, Communications and Computers, Bucharest, Roma- nia 2006 Oct 16. 33. Iqbal S, Sophia M, Divyashree JA, Mundas M, Vidya R. Imple- mentation of Stolzs algorithm for melanoma detection. Interna- tional Advanced Research Journal in Science, Engineering and Technology. 2015;2(6):9-12. Highlights in BioScience Page 6 of 7 August 2021|Volume 4 http://bioscience.highlightsin.org/ Khezri et al., 2021 ANN-based diagnosis method for skin cancers using dermoscopic images 34. Stolz W. ABCD rule of dermatoscopy: a new practical method for early recognition of malignant melanoma. Eur. J. Dermatol.. 1994;4:521-7. 35. Ahnlide I, Bjellerup M, Nilsson F, Nielsen K. Validity of ABCD rule of dermoscopy in clinical practice. Acta dermato-venereologica. 2016 Mar 1;96(3):367-72. 36. Bareiro Paniagua LR, Leguizamón Correa DN, Pinto-Roa DP, Vázquez Noguera JL, Salgueiro Toledo LA. Computerized Med- ical Diagnosis of Melanocytic Lesions based on the ABCD ap- proach. CLEI Electronic Journal. 2016 Aug;19(2):6-. 37. Shih TY. The reversibility of six geometric color spaces. Pho- togrammetric Engineering and Remote Sensing. 1995 Oct;61(10) :1223-32. 38. Mirzaalian H, Lee TK, Hamarneh G. Learning features for streak detection in dermoscopic color images using localized radial flux of prin- cipal intensity curvature. Proc. IEEE Workshop Math. Methods Biomed. Image Anal, 2012. 39. Aswin RB, Jaleel JA, Salim S. Implementation of ann classifier using matlab for skin cancer detection. International Journal of Computer Science and Mobile Computing. 2013 Dec;1002:87- 94. 40. Messadi M, Cherifi H, Bessaid A. Segmentation and ABCD rule extraction for skin tumors classification. arXiv preprint arXiv :2106.04372. 2021 Jun 8. 41. Santiago-Montero R, Asael D, Hernandez G. Border and asym- metry measuring of skin lesion for diagnostic of melanoma us- ing a perimeter ratio. Asian J Comput Sci and Inf Technol. 2016;6(2). 42. Jaleel JA, Salim S, Aswin RB. Artificial neural network based detection of skin cancer. International Journal of Advanced Re- search in Electrical, Electronics and Instrumentation Engineer- ing. 2012 Sep;1(3). 43. Ahmed K, Jesmin T, Rahman MZ. Early prevention and detec- tion of skin cancer risk using data mining. International Journal of Computer Applications. 2013 Jan 1;62(4). 44. Thirumavalavann S, Jayaraman S. ANN based computer aided diagnosis and classification of skin cancers. power and comput- ing technologies. 2017;12(4).pp1137-1142. 45. Alam FI, Faruqui RU. Optimized calculations of haralick texture features. European Journal of Scientific Research. 2011 Mar 1;50(4):543-53. 46. Codella NC, Gutman D, Celebi ME, Helba B, Marchetti MA, Dusza SW, Kalloo A, Liopyris K, Mishra N, Kittler H, Halpern A. Skin lesion analysis toward melanoma detection: A challenge at the 2017 international symposium on biomedical imaging (isbi), hosted by the international skin imaging collaboration (isic). In2018 IEEE 15th international symposium on biomedical imaging (ISBI 2018) 2018 Apr 4 (pp. 168-172). IEEE. 47. Pennisi A, Bloisi DD, Nardi D, Giampetruzzi AR, Mondino C, Facchiano A. Skin lesion image segmentation using Delaunay Triangulation for melanoma detection. Computerized Medical Imaging and Graphics. 2016 Sep 1;52:89-103. 48. Fan H, Xie F, Li Y, Jiang Z, Liu J. Automatic segmentation of dermoscopy images using saliency combined with Otsu thresh- old. Computers in biology and medicine. 2017 Jun 1;85:75-85. 49. Jahanifar M, Tajeddin NZ, Asl BM, Gooya A. Supervised saliency map driven segmentation of lesions in dermoscopic images. IEEE journal of biomedical and health informatics. 2018 May 22; 23(2):509-18. 50. Sreelatha T, Subramanyam MV, Prasad MG. Early detection of skin cancer using melanoma segmentation technique. Journal of medical systems. 2019 Jul;43(7):1-7. Highlights in BioScience Page 7 of 7 August 2021|Volume 4 http://bioscience.highlightsin.org/ Abstract Introduction Materials and Methods Properties used to characterize the asymmetry of the lesion Features used for irregular characterization at the lesion border Features used to characterize lesion color variation Properties used for diameter Characteristics used to characterize the textural variation of the lesion Classification Diagnosis Results and Discussion References