Classification of Different Types of Psoriasis using CNN

Abstract

Psoriasis is a chronic disease in which excessive immune activity leads to excessive skin cell proliferation. The initiative aims to solve this critical issue Rapid skin cell turnover, which happens every 3 to 7 days as opposed to the typical 3 to 4 weeks, is the cause of this disorder. Its onset is influenced by various factors, including infections, stress, and heredity. It is not contagious. Elbows, knees, and scalp are the most common places for scaly, inflammatory skin patches to appear, but other body areas may also be affected. A rash with itchy, scaly patches is the result of the skin condition psoriasis. Over 125 million people worldwide—roughly 2% to 3% of the population—have psoriasis. Psoriasis comes in different forms, including nail psoriasis, erythrodermic psoriasis, pustular psoriasis, inverse psoriasis, and plaque psoriasis. Psoriasis is a skin disorder that affects 296 people; of these, 175 (59.1%) are Malay people, 82 (22.7%) are Indians, 37 (12.5%) are Chinese, and 2 (0.6%) are other people. The majority of clinical cases (89.9%) are caused by chronic plaque psoriasis, with the remaining cases being caused by erythrodermic psoriasis (4.7%), guttate psoriasis (3.0%), and pustular psoriasis (1.7%) in that order of prevalence [18]. Using machine learning approaches, this study aims to accurately classify several forms of psoriasis, such as plaque, guttate, nail, erythrodermic, pustular, inverse and normal psoriasis. The goal is to create a reliable classification technique that can use image data to discern between these different types of psoriasis. In order to do this, deep learning models called convolutional neural networks, or CNNs, are used. CNNs are particularly well-suited for picture categorization tasks. For efficient categorization across all classes in this study, the CNN algorithm is used. Data augmentation and transfer learning are the methods used. We take the affected image as input and determine the type of psoriasis from the output.