Client-Server-Based Mobile Application for Skin Cancer Diagnosis Using Convolutional Neural Networks

Skin cancer is the most prevalent cancer worldwide. Similarly, melanoma is one of the most common types of skin cancer that is highly prone to metastasis, which complicates the treatment greatly. Therefore, early detection of skin cancer, especially melanoma, is of utmost importance. However, the current widely used approach to diagnose melanoma relies on visual examination by dermatologists, which can be subjective and can lead to neglected early signs of malignancy. Considering the time-consuming nature of this process, it is a costly task to have every single mole checked by a professional. This study develops and evaluates a client-server-based mobile application designed to detect and classify moles as benign or malignant using a comprehensive dataset and two machine learning models, facilitating the early detection of malignant moles. Evaluations yield that the model accurately detects 93% of malignant moles from real-time videos taken with mobile device cameras and images selected from the gallery.