AttentionVGG-Modified VGG-16 for Alzheimer Disease Classification

Alzheimer’s disease (AD) is a neurodegenerative disorder that inflicts irreversible damage on several brain regions, particularly the hippocampus, resulting in cognitive, functional, and behavioral impairments. Timely identification of the ailment will alleviate the suffering of sufferers and their family ones. Deep learning algorithms have been effectively utilized in recent years for several medical imaging applications, including the identification of Alzheimer’s disease. In this paper, we propose a modification of VGG-16 architecture, AttentionVGG-16, that outperforms current state-of-the-art results on both the ADNI and OASIS datasets. We have done an ablation study to understand the impact of adding attention layers to VGG-16 and zero-shot learning strategy to understand the model’s generalization capability. We have achieved accuracy and the macro f-1 score of 0.98 on the OASIS dataset and 0.97 on the ADNI dataset. Additionally, the results obtained are compared with analogous methods presented in the literature.