APPLICATION OF SEGMENTATION ALGORITHMS FOR FINDING DISEASE CONTOURS ON SKIN AREAS
DOI:
https://doi.org/10.32689/maup.it.2023.3.5Keywords:
segmentation, localization, watershed, data, processing, image, morphological processing, thresholdAbstract
This paper investigates the application of segmentation to identify and highlight the location of a disease on a skin area. The object of the study is to select the optimal image segmentation algorithm with clear separation of the disease area and contours, regardless of its shape. The relevance of the study is due to the fact that modern methods of segmentation and localization of diseases are widely used to improve the accuracy and clarity of training a neural network. The algorithms allow to identify and fix the exact area of the skin that is needed to be fed to the neural network. The purpose of the work is to develop an algorithm for segmentation and contour finding that can identify and highlight a local part of a disease on a skin image provided by the user. The algorithm should be accurate and efficient, regardless of the image's external factors. The paper demonstrates the application of image segmentation methods, such as threshold segmentation, morphological processing algorithm, and watershed algorithm. For experiments, an image of atypical nevus from the DermNet dataset was used. Image segmentation was performed using the Skimage library, which also includes contour finding algorithms. Based on the results of the set experiments, where all algorithms received the same image, the clarity of disease detection was demonstrated using watershed segmentation. Unlike others, it was able to determine the location of the disease, clearly separate it from the overall image, and significantly use attenuation, which does not harm further collaboration with the contour finding algorithm. The study found that this method is suitable for solving segmentation and image processing problems in dermatology. This is due to the fact that it effectively highlights areas of the skin affected by the disease and does not conflict with the Skimage library-based contour localization algorithm at standard parameters.
References
Haleem A., Javaid M., Singh R. P., and. Suman R. «Telemedicine for healthcare: Capabilities, features, barriers, and applications». Sensors International. Vol. 2. 2021. DOI: 10.1016/j.sintl.2021.100117.
Zhang C., «Smartphones and telemedicine for older people in China: Opportunities and challenges». Digital Health. Vol. 8. 2022. DOI: 10.1177/20552076221133695.
Allaert F.A., Legrand L., Abdoul Carime N., and Quantin C. «Will applications on smartphones allow a generalization of telemedicine?». BMC Medical Informatics and Decision Making, Vol. 20. № 1. 2020. DOI: 10.1186/s12911-020-1036-0.
Joseph S. and Olugbara O. O., «Preprocessing Effects on Performance of Skin Lesion Saliency Segmentation». Diagnostics, Vol. 12, № 2. Feb. 2022, DOI: 10.3390/diagnostics12020344.
Masoud Abdulhamid I.A., Sahiner A., and Rahebi J. «New Auxiliary Function with Properties in Nonsmooth Global Optimization for Melanoma Skin Cancer Segmentation». Biomed Res Int. Vol. 2020. DOI: 10.1155/2020/534592
Garg S. and Jindal B. «Skin Lesion Segmentation in Dermoscopy Imagery». International Arab Journal of Information Technology. Vol. 19. № 1. P. 29–37. Jan. 2022. DOI: 10.34028/iajit/19/1/4.
Wang H. et al. «Watershed segmentation of dermoscopy images using a watershed technique». Skin Research and Technology. 2010. Vol. 16. № 3. P. 378–384. DOI: 10.1111/j.1600-0846.2010.00445.x.
Moussaoui H., N. El Akkad, and Benslimane M. «A hybrid skin lesions segmentation approach based on image processing methods». Statistics, Optimization and Information Computing. Vol. 11. № 1. P. 95–105. DOI: 10.19139/soic-2310-5070-1549.
Shahabi F., Poorahangaryan F., Edalatpanah S. A., and Beheshti H. «A Multilevel Image Thresholding Approach Based on Crow Search Algorithm and Otsu Method». Int J Comput Intell Appl. 2020. Vol. 19. № 2. DOI: 10.1142/S1469026820500157.
Pitoy P. A. and Suputra I. P. G. H. «Dermoscopy Image Segmentation in Melanoma Skin Cancer using Otsu Thresholding Method». JELIKU (Jurnal Elektronik Ilmu Komputer Udayana). Vol. 9. № 3. P. 397. DOI: 10.24843/jlk.2021.v09. i03.p11.
Lumini A., L. Nanni A., Codogno A., and Berno F. «Learning morphological operators for skin detection». Journal of Artificial Intelligence and Systems. 2019. Vol. 1. № 1. DOI: 10.33969/ais.2019.11004.
Rew J., Kim H., and Hwang E. «Hybrid segmentation scheme for skin features extraction using dermoscopy images». Computers, Materials and Continua. 2021. Vol. 69. № 1. DOI: 10.32604/cmc.2021.017892.
Prabha Devi D. and Iniya Shree S. «Recognition and investigation of skin cancer using morphological operations». International Journal of Recent Technology and Engineering. 2019. Vol. 7. № 4.
Wang H. et al. «Modified watershed technique and post-processing for segmentation of skin lesions in dermoscopy images». Computerized Medical Imaging and Graphics. 2011. Vol. 35. № 2. DOI: 10.1016/j.compmedimag.2010.09.006.
Das A. and Ghoshal D. «Human Skin Region Segmentation Based on Chrominance Component Using Modified Watershed Algorithm». Procedia Computer Science. 2016. DOI: 10.1016/j.procs.2016.06.072.
Shalu and Kamboj A. «A Color-Based Approach for Melanoma Skin Cancer Detection». ICSCCC 2018 – 1st International Conference on Secure Cyber Computing and Communications. 2018. DOI: 10.1109/ICSCCC.2018.8703309.
Ashour A. S., Nagieb R. M., El-Khobby H. A., Abd Elnaby M. M., and Dey N. «Genetic algorithm-based initial contour optimization for skin lesion border detection» Multimed Tools Appl. Vol. 80. № 2. 2021. DOI: 10.1007/s11042-020-09792-8.
