EfficientNet and Vision Transformers for Alzheimer’s Detection and Diagnosis

Alzheimer's disease (AD) is a serious mental disorder that affects memory and cognition. Effective treatment and intervention of AD depend on early diagnosis. In this paper, we propose an in-depth study on the detection of AD using data from MRI and PET scans. The system uses two efficient algorithms for image classification: Vision Transformers and EfficientNet. ViT and EfficientNet architectures are primarily used to extract features after normalizing clinical data in the pipeline. Using publicly available data to train and evaluate the two models, the results show that ViT outperforms EfficientNet with 94% accuracy, while EfficientNet has 91% accuracy. Scanning is very efficient because it uses a wide range of image qualities. However, EfficientNet balances efficiency and accuracy, making it suitable for resource-constrained environments. Although both models are successful in detecting AD, the comparative analysis showed that ViT outperforms in terms of accuracy. This study contributes to cognitive behavioral therapy research and shows how these models can help doctors identify and diagnose Alzheimer’s disease at an early stage.