A Modified Convolutional Neural Networks For MRI-based Images For Detection and Stage Classification Of Alzheimer Disease

Abstract

Artificial Intelligence (AI) combined with efficient image processing tools help doctors better predict disease progression. Alzheimer's Disease (AD) early diagnosis is one of the most difficult challenges in medical imaging systems involving AD classification beyond AD detection. In this paper, a Convolutional Neural Network (CNN) is implemented for AD detection and stage classification for Magnetic Resonance Imaging (MRI). The implementation methodology begins with basic preprocessing techniques, such as image resizing and pixel normalization, and then extracted features are reconstructed into a 1-Dimensional vector to be fed to the CNN with accompanying labels. Four different labels are used according to the four different AD stages considered, which are (Non-Demented, Mild Demented, Moderate Demented, and Very Mild Demented). The prediction model's evaluation shows an efficient result in accuracy and model loss for only ten epochs; the accuracy of the model recorded was 97%. The model was designed using Python Programming language with typical Machine Learning (ML) environment consisting of TensorFlow and OpenCV libraries. Graphical User Interface (GUI) is considered to enable further system testing and the visual insertion of images on the computer desktop.