Logistic-Based OVA-CNN Model for Alzheimer’s Disease Detection and Prediction Using MR Images

Alzheimer’s disease (AD) is a form of dementia that diminishes a patient’s cognition and memory. By predicting AD progression early, physicians can begin medication that may reduce the disease’s progression. AD detection and prediction have been extensively performed using convolutional neural networks (CNNs). Despite this, CNN’s use for AD detection is limited by the number of parameters and the extensive training data required. Consequently, this study proposes a logistically based OVA-CNN model for AD detection and prediction using Magnetic Resonance Images (MRI). OVA-CNN consists of the four binary CNN which classified four classes of AD and the logistic regressor predicts the time conversion of pMCI patient who converts AD. The experimental results on the Alzheimer’s Disease Neuro Imaging Initiative (ADNI) demonstrate that the proposed model achieves a high level of accuracy of 93%.