Deep Learning Approach for Early Diagnosis of Alzheimer’s Disease

Alzheimer’s disease (AD) is a widespread neurological condition that affects millions of individuals worldwide. As a leading cause of dementia, AD is primarily marked by cognitive deterioration and the presence of various behavioral challenges that disrupt everyday activities. Drug trials for Alzheimer’s disease (AD) fail most of the time, probably due to the difficulty in identifying patients at an early stage. Efforts are being made to advance the understanding of the causal molecular processes involved in Alzheimer’s disease through non-invasive imaging modalities. With the boost in the field of big data, deep learning, and the advancement in computational capabilities, diagnosis and classification of various diseases have become easier. Convolutional Neural Networks have been found to be better at diagnosing Alzheimer’s disease through MRI than machine learning approaches using neuropsychological data. In this study, a big data framework based on Hadoop and deep learning approach is used to identify early diagnostic biomarkers of Alzheimer’s disease by integrating MRI and neuropsychological test results. Neuropsychological scores, MRI, and neurochemical scores are used to extract the brain’s structural, neurochemical, and behavioral features. A combination rule using XceptionNet and SVM classifier is then applied for accurate final classification and clinician validation. This includes feature selection and ensemble-based classification.