Deep Learning-Based MRI Image Classification for Early-Stage Alzheimer's Disease Diagnosis

Dementia, which causes cognitive impairment, poses significant challenges to healthcare systems globally. Alzheimer's disease (AD), the most common category of dementia, demands early diagnosis in order to initiate relevant preventive treatment. In this work, a deep learning-based approach for AD identification by using MRI images is proposed. Four deep learning models such as ResNet-lOl, MobileNet-v2, ResNet-50, and ResNet-18 to analyze a dataset of 33,948 MRI images were utilized. The dataset were divided into four categories: mild dementia, moderate dementia, non-dementia, and very mild dementia. The results show that the ResNet-lOl demonstrated superior performance as compared to the others, with validation accuracy of 99.39%. Our proposed technique aims to enhance patient patient treatment and quality of life by facilitating early detection and diagnosis.