A BRIEF OVERVIEW OF DEEP LEARNING'S POTENTIAL FOR MRI IMAGE BASED ALZHEIMER'S DISEASE PREDICTION

Cognitive capacities gradually deteriorate with the onset of Alzheimer's Disease (AD), an irreversible neurological disease. As the disease progresses, it gradually robs people of their ability to control their movements and other brain functions. Initiating therapeutic measures that can impede the advancement of the disease is made possible by early detection. To detect structural and functional alterations linked to AD, Magnetic Resonance Imaging (MRI) offers high-resolution pictures of the brain. However, owing to typical aging overlap and brain architecture diversity, MRI picture interpretation might be difficult. In order to conduct a longitudinal study, sophisticated analytic methods and highly trained radiologists are required. This problem has been addressed in recent decades with the emergence of Deep Learning (DL) systems that automatically detect and classify AD using MRI scans. DL models are able to learn from complicated patterns and enormous datasets, allowing for early diagnosis and the detection of minor changes in brain structure prior to clinical signs. This study provides a comprehensive analysis of various DL frameworks that have been created for the purpose of AD picture recognition and classification using MRI. A number of AD classification systems built using DL algorithms have been briefly reviewed at the outset. In order to regulate the global morality rates, it is necessary to predict and classify MRI-based AD effectively. A comparative analysis is then performed to analyse the shortcomings of such algorithms and propose a new approach.