Mobile Application for Skin Disease Classification Using CNN with User Privacy

Abstract

The human body faces numerous challenges, and the skin, in particular, proves to be difficult to analyze mechanically because of its unevenness, complexion, presence of hair, and other alleviated attachments. Consequently, there is a growing need for an accurate, automatic system for detecting skin diseases. Due to dataset imbalance and data security issues, classifying skin diseases using medical images is difficult. The primary reason for the lack of accessible datasets is privacy and confidentiality concerns related to medical data sharing. In Medical Image Analysis (MIA), the Convolution Neural Network (CNN) performance for classification and a federated learning strategy for data privacy protection is impressive. Our results show that CNN can achieve an accuracy score of 0.90. We suggest a mobile application for classifying skin diseases using CNN and a federated learning strategy. The analysis of human skin with this mobile app is outstanding while maintaining data security.