A Low Cost CNN-Based Method for Differential Diagnosis of Alzheimer’s and Frontotemporal Dementia

Alzheimer’s disease (AD) and Frontotemporal Dementia (FTD) are degenerative neurological diseases, affecting the cognitive abilities of large elderly populations throughout the world. Advancement in technology has brought about a surge in sophisticated computer-aided diagnostic tools involving Machine Learning (ML) approaches to classify different forms of dementia accurately. Despite this, our healthcare system is facing a scarcity of reliable methods that can accurately classify Alzheimer’s with FTD having similar clinical symptoms. In the proposed work, we introduce a Convolutional Neural Network (CNN) architecture-based model, designed specifically for multi-class classification tasks using traditionally extracted features. We evaluate the proposed model and demonstrate its effectiveness as compared to leading state-of-the-art models. Our method differentially diagnoses patients with neurological diseases (AD and FTD), and healthy patients with high accuracy and much lower computational cost than image-based methods. Our work achieves a higher accuracy of 95.39% on a publicly accessible EEG dataset, outperforming existing state-of-the-art methods.