Privacy-Preserving Medical Image Classification using Federated Learning

Abstract

The field of medical image classification has made significant progress due to the increasing adoption of medical imaging technologies and the digitization of healthcare data. However, challenges related to data privacy and security hinder the full potential of these vast medical image datasets. In this thesis, we propose an innovative approach to address these concerns by using federated learning, a distributed machine learning paradigm that facilitates collaborative model training without sharing raw data. The primary goal of this research is to develop a privacy-preserving framework for medical image classification that ensures the confidentiality of sensitive medical images while achieving accurate classification results. To achieve this objective, we adopt a federated learning setup where multiple healthcare institutions contribute their local medical image datasets while maintaining patient information privacy. Our framework incorporates advanced cryptographic techniques and decentralized learning algorithms, enabling collaborative model training without the need to exchange raw data. The study's findings can have a significant impact on privacy-preserving machine learning systems in healthcare. They promote trust and collaboration among healthcare providers and researchers, benefiting diagnostic accuracy and patient care. The framework for privacy preserving medical image classification using federated learning ensures data confidentiality and encourages responsible data usage. Overall, this research contributes to advancements in healthcare technology.