Classification of Alzheimer's Disease from MRI Data Using an Ensemble of Hybrid Deep Convolutional Neural Networks

Although there is no cure for Alzheimer's disease (AD), an accurate early diagnosis is extremely important for both the patient and social care, and it will become even more significant once disease-modifying agents are available to prevent, cure, or even slow down the progression of the disease. In recent years, classification of AD through deep learning techniques has been one of the most active research areas in the medical field. However, most of the existing techniques cannot leverage the entire spatial information; hence, they lose the inter-slice correlation. In this paper, we propose a novel classification algorithm to discriminate patients having AD, mild cognitive impairment (MCI), and cognitively normal (CN) using an ensemble of hybrid deep learning architectures to leverage a more complete spatial information from the MRI data. The experimental results obtained by applying the proposed algorithm on the OASIS dataset show that the performance of the proposed classification framework to be superior to that of the some conventional methods.