From Pixels to Progression-A Deep Dive into Alzheimer's Staging with MRI and CNN

With the number of cases projected to reach nearly 13 million by 2050, Alzheimer's disease is a pressing global health threat. The economic burden is expected to surpass USD 1 trillion by then. The key to successful intervention is early detection, yet obtaining large and diverse datasets for research is challenging due to privacy concerns. This study addresses the imperative need for accurate Alzheimer's disease (AD) detection utilizing a deep learning model trained on the OASIS Alzheimer's Detection dataset obtained from Kaggle. The dataset, encompassing 80,000 brain MRI images systematically categorized into distinct Alzheimer's progression stages, serves as a crucial resource. Data pre-processing involves normalization, resizing, and augmentation, optimizing images for effective training of a Convolutional Neural Network (CNN) model developed from scratch. The developed model attained an accuracy of around 98.71%, showcasing its reliability and potential as a decision support system for Alzheimer's detection. Given the escalating prevalence of disease and the related financial and healthcare burdens, the proposed study is crucial for developing intelligent tools that can aid medical practitioners in accurate and timely diagnoses.