Volumetric Convolutional Neural Network for Alzheimer Detection

Abstract

Alzheimer's disease is a progressive brain disorder, which over a period leads to loss of memory due to the formation of mainly two types of lesions being senile plaques and neurofibrillary tangles. Alzheimer's detection at an early stage thus becomes of paramount importance to lessen the loss of cognitive, other memory since this disease cannot be reversed, and no cure is available until now. This study has put forward a 3-Dimensional Convolutional neural network (3D-CNN) framework for binary classification of Alzheimer disease as Healthy Control (HC) and Alzheimer Disease Control (AD) using the pre-processed volumetric T1 weighted Magnetic Resonance Images obtained from the MIRIAD dataset. The pre-processing pipeline applied on the MRI Images obtained from the MIRIAD dataset is bias correction, skull stripping, and registration. This research also highlights the broad areas for future research on multimodal and multiclass Alzheimer detection.