Automatic Identification of Alzheimer's Disease and Epilepsy Based on MRI

Alzheimer's disease (AD) and epilepsy are both common chronic diseases in neurology. A certain proportion of AD patients have been found to have epilepsy complication. Neuroimaging such as structural magnetic resonance imaging (MRI) has been proved to be useful in assessing the pathology of AD and epilepsy. Computer-aided diagnosis (CAD) on automatical MRI identification can be applied to assist physicians in diagnosing both diseases. In this paper, it is investigated that the performance of identification on AD, AD complicated with Epilepsy, and Epilepsy based on different MRI brain tissues and feature extraction methods. 17 AD patients, 17 AD patients complicated with epilepsy, 15 epilepsy patients, and 10 healthy control subjects from West China Hospital, Sichuan University were studied. Several preprocessing steps were performed for each MRI to obtain gray matter (GM) and white matter (WM) tissue voxels. Principal component analysis (PCA) and partial least squares (PLS) were adopted to extract features. Three classes of patients and healthy controls were distinguished separately by support vector machine (SVM). The performance is evaluated by k-fold cross-validation strategy. The approach on combination of GM and WM tissues with PCA archieved the optimal performance, with the accuracy of 87.41%, 83.7%, and 75.2% for AD, AD complicated with epilepsy, and epilepsy identification respectively. Our proposed approach appears to have significant potential as a clinical decision support tool in assisting physicians in their clinical routine.