Detection of Alzheimer’s Disease progression from Structural MRIs using ConvMixer based Deep Learning Model

Alzheimer’s Disease (AD) is a progressive neurodegenerative disorder that significantly impacts cognitive functions and quality of life. Affecting over 50 million people worldwide, early and accurate detection of AD progression is crucial for timely intervention and management. This paper presents a reliable method for classifying Alzheimer’s stages using Deep Learning based ConvMixer model on structural MRI data. The proposed deep learning based convmixer model significantly utilizes the patch based learning with depthwise separable convolution to achieve improved performance. The dataset used in this work has been extracted from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) and comprises of three classes: Alzheimer’s Disease (AD), Cognitive Impairment (CI), and Cognitively Normal (CN). The proposed approach demonstrated exceptional performance, achieving an accuracy of 99.88%, specificity of 99.87%, sensitivity of 99.90 and precision of 99.95%. The achieved results demonstrate the proposed model’s accuracy in distinguishing Alzheimer’s stages, making it a promising tool for early diagnosis and disease progression monitoring.