Binary Classification of Alzheimer Disease using sMRI Imaging modality and Deep Learning
share
Alzheimer Disease (AD) is the most common form of dementia affecting the elderly population worldwide. Many neuroimaging modalities have been used to check the detection and progression of AD of which structural Magnetic Resonance Imaging (sMRI) is an important one. The recent rise in the popularity of deep learning methods with applications in computer vision, reinforcement learning and artificial intelligence has created a resurgence in the application of these methods to the classification of AD through different imaging modalities. In this study, by utilizing the concept of transfer learning in deep learning, we propose a classification framework to differentiate subjects with Clinical Dementia Rating (CDR) of zero from subjects with CDR greater than zero by using deep learning architectures such as Xception and Inception version 3 in the Keras deep learning library. The attained validation set accuracies are as high as 99.12 version 3 network and 97.97 suggest that meaningful predictors composed of sMRI and network measures may offer the possibility for early detection of subjects in the early stages of AD.
READ FULL TEXT