ADNet: deep learning based model for Alzheimer’s disease classification

Alzheimer’s Disease (AD) is the most prevalent form of dementia, which results in a progressive decline in cognitive abilities in elderly individuals. Imaging data from neuroimaging technique such as Magnetic Resonance Imaging (MRI) allow for the classification on the basis of the structural and functional alterations induced by AD in the brain. In medical environments, the accurate diagnosis of Alzheimer’s disease is essential, as it facilitates timely intervention and treatment strategizing, as well as improves our understanding of the complexities of the disease in the field of neuroscience. In this paper, we propose ADNet, a new Deep Learning (DL) architecture with the objective of improving the accuracy and performance of AD classification. It leverages the increasingly widespread use of DL methods. Three key additions to our CNNs are the implementation of an inverted bottleneck, the use of independent downsampling layers (DoL), and the use of depth-wise convolutions to enrich the spatial dimension with information. The research findings show that the proposed model ADNet surpasses all other modern approaches with a remarkable 99.81% accuracy. The evaluation metrics highlight the strong performance of the model by showcasing its precision and recall for different binary classifications. © The Author(s), under exclusive licence to Springer Nature Switzerland AG 2025.