Automated Alzheimer’s Detection Using Deep Learning And Mri Analysis

Abstract

A wide range of cognitive impairment symptoms are included in dementia, with Alzheimer's disease making up around two-thirds of cases. Alzheimer's disease currently has no known cure, therefore economic, financial, and medical repercussions can be avoided by early identification and aggressive care. To categorize the various phases of Alzheimer's disease, we provide a deep learning method in this work. First, we preprocess utilizing a pyramid example and bi-linear interpolation as part of our four-step technique. The feature extraction process then uses both the local binary pattern (LBP) and gray-level co-occurrence matrix (GLCM) approaches. After the characteristics are recovered, the VGG19 neural network is used to concatenate and classify them.
By using the Alzheimer's Disease Neuroimaging Initiative's (ADNI) MPRAGE structural MRI dataset, our approach outperformed current state-of-the-art technologies. With an overall accuracy of 89.80%, the suggested method successfully differentiated between gray matter (GM) and white matter (WM) across a number of diagnostic categories, including cognitively normal (CN), early mild cognitive impairment (EMCI), late mild cognitive impairment (LMCI), and Alzheimer's disease (AD). Both GM and WM demonstrated good accuracy in binary classification tests. In particular, GM distinguished between CN and EMCI with 96.43% accuracy, EMCI and AD with 90.91% accuracy, and LMCI and AD with 95.24% accuracy. The accuracy of WM's categorization between CN and LMCI and between EMCI and LMCI was 95.6%. In certain comparisons, GM and WM both obtained the same 96.15% accuracy.