Alzheimer’s Disease Detection Using Deep Learning and Federated Learning

Abstract

Appropriate and precise diagnosis of brain diseases is crucial as many forms of Alzheimer’s disease display similar indications in their initial stages. Most automatic detection or classification systems based on deep learning present confidentiality concerns due to their use of integrated computing and local storage data requirements for training. The goal of this paper is to defend sensitive patient data by proposing a deep learning algorithm that utilizes Federated learning techniques in an IoT-based edge computing framework. It is shown that the secrecy of the patient data can be maintained while still preserving accuracy and efficiency. This method provides a secure edge data protection model where there is no requirement for centralized storage. This paper discusses the strategy and implementation of the federated structure, taking into account the number of devices in the network, memory, and processing capabilities. The success and accuracy of the proposed algorithm are established with empirically defined metrics such as accuracy and defined thresholds.