Acne Vulgaris Severity Analysis Application

Abstract

Abstract The most common skin problem, acne vulgaris, may have serious financial and psychological implications for individuals who have it, thus it's crucial to have an accurate grading system for effective treatment. Artificial intelligence (AI)-based skin image analysis has gained a lot of relevance in recent years, particularly for analyzing and assessing the skin images captured by mobile phones. The difficulty in accurately assessing the severity of acne lies in the similarity of lesion appearances and the challenge of counting lesions. The study suggested a mobile app that can identify different forms of acne to solve this problem by using photos of facial skin. This study employed the You Only Look Once (YOLO) deep learning algorithm to find and identify acne. Comedone, papule, pustule, and nodule are the four forms of acne vulgaris taken into consideration. The dataset used to train and test the model is taken from the ACNE04 dataset and a private dataset from the dermatology OPD of JSS Medical Hospital, Mysuru, Karnataka, India. The app showed positive outcomes in severity analysis, showing dermatologist-level diagnosis. This application could be a valuable tool for clinicians with a smart phone to assess acne severity quickly and conveniently, anywhere and at any time.