Skin Cancer Risk Checker

CLARITY: Finding SKIN Cancer by AI Image Processing Technique

(Comparing Lesions and Assessing Risks with an Integrated Tentative Diagnosis Yield)

By Pranayaa Jeyaraman

Upload Image

The rising incidence and mortality rates associated with malignant skin tumors present a significant public health challenge, particularly as early detection remains crucial for effective treatment. Current diagnostic methods heavily rely on physicians' expertise, which can sometimes lead to misdiagnoses or unnecessary biopsies. This research explores the role of Artificial Intelligence (AI) in improving skin cancer diagnosis, with a focus on machine learning (ML) and deep learning (DL) techniques. The study offers a comprehensive review of these AI methods, comparing them across widely used datasets and prevalent literature, while also highlighting the strengths and limitations of historical diagnostic approaches.

Special attention is given to the PAD-UFES-20 dataset, which comprises 2,298 clinical images of six different skin lesions, including three types of skin cancer (Basal Cell Carcinoma, Squamous Cell Carcinoma, and Melanoma) and three benign lesions (Actinic Keratosis, Seborrheic Keratosis, and Nevus). The dataset, collected by the Dermatological and Surgical Assistance Program at the Federal University of Espírito Santo, Brazil, is enriched with clinical metadata, offering an opportunity to assess AI's diagnostic accuracy. This survey discusses the current challenges in automated skin cancer detection, outlines the progress made in AI-based diagnostic systems, and presents future directions for research. Ultimately, the aim is to enhance early detection, reduce healthcare costs, and improve accessibility, particularly in resource-limited settings where skin cancer specialists are scarce.