Transforming Alzheimer’s Disease Diagnosis: Implementing Vision Transformer (ViT) for MRI Images Classification

Abstract

Listen

Translate article

Alzheimer’s disease (AD) is a progressive neurological disorder and the leading cause of dementia, accounting for 60-80% of cases. Early detection of AD is crucial for timely intervention, as the disease significantly impacts cognitive functions and daily activities. Diagnosing AD in clinical settings remains challenging due to subtle early symptoms, diverse presentations, lengthy diagnostic processes, and inconsistent criteria that heavily rely on medical expertise. Accurate and timely diagnosis during the early stages is essential for effective treatment and intervention. Modern imaging techniques, such as Magnetic Resonance Imaging (MRI), have become essential in diagnosing Alzheimer’s by providing detailed insights into structural brain changes. This study explores the application of the Vision Transformer (ViT) model for classifying MRI images of Alzheimer’s patients, focusing on enhancing accuracy and reliability through data augmentation during pre-processing. A dataset of 8,000 MRI images, categorised into four groups—non-demented, very mild demented, mild demented, and moderate demented—was used to evaluate the ViT model. The experiment achieved promising results, with an accuracy of 98.19%, sensitivity of 96.34%, specificity of 98.80%, and an F1-score of 96.37%. These findings underscore the model’s effectiveness in distinguishing between affected and unaffected individuals, minimising misdiagnosis and enabling timely clinical interventions. However, some challenges remain, particularly in the classification between “Non-Demented” and “Very Mild Demented” cases. Future research should focus on enhancing data augmentation techniques and increasing data diversity in these categories to improve the model’s performance further. The ViT model holds great potential for advancing Alzheimer’s diagnosis, offering a valuable tool for early detection and intervention in clinical settings.