Classification of Brain MRI Slices into Control/Alzheimer's using Deep-Learning Scheme

Alzheimer's Disease (AD) is a progressive neurological ailment which causes damage to memory as well as behaviour. It is the most general cause of dementia among older people. Appropriate detection and management is necessary and the clinical stage detection of AD is usually executed with brain image supported methods. Magnetic Resonance Imaging (MRI) is one of the common practice to analyse the AD. This study aims to develop a deep-learning scheme (DLS) to classify the chosen axial-plane MRI-slices into control/AD class. The stages of this DLS involve in; data collection and resizing feature extraction with a chosen DL-model, classification using 3-fold cross-validation with the help of chosen classifiers. This work is executed on the brain MRI slices of with and without skull-region. Further, the classification is performed using the individual-feature, and ensemble of deep-learning features (EODF). This study's result verifies that the DLS implemented on MRI slices without skull helps to achieve an accuracy of $>98 \%$ when FF based classification is executed with random-forest classifier. The overall result of this study confirms the merit of DLS on the chosen database.