Hybrid CNN-SVM for Alzheimer's Disease Classification from Structural MRI and the Alzheimer’s Disease Neuroimaging Initiative (ADNI)
Download as PDF
Lan Lin, Baiwen Zhang, Shuicai Wu
Alzheimer's disease (AD) is a progressive neurological disorder among the elders, which results in memory-related issues in subjects. An accurate classification of patients with AD and mild cognitive impairment (MCI) from healthy control subjects (HC) based on structural magnetic resonance imaging (MRI) is of critical clinical importance. In this paper, good intermediate representations of MRI are obtained from a pre-trained convolutional neural network (CNN). Principal component analysis (PCA) and sequential feature selection (SFS) are applied for feature selection, while a support vector machine (SVM) is adopted to evaluate the classification accuracy. 422 Alzheimer's Disease Neuroimaging Initiative (ADNI) baseline MRI were used for development and validation of our proposed method. As a result, this paper achieved a classification accuracy of 90% for binary classification of AD and HC, 81% for AD and MCI and 72% for MCI and HC.
Alzheimer'S Disease, Classification, Transfer Learning