New kernel method for MRI classification

The classification of brain Magnetic Resonance Images (MRI) into healthy or pathological subjects can be considered as the key for the preclinical state of a patient. If we consider classifying MRIs manually, this procedure can be time-consuming. In this work, we aim to present an automatic Computer Aided Design (CAD) system for brain MRI classification based on the new Downsized Kernel Principal Component Analysis (DKPCA) and Artificial Neural Network (ANN) entitled ANN-DKPCA. The proposed study contains three main steps; Data acquisition and preprocessing stage, feature extraction and feature reduction stage and finally the classification stage. The Alzheimer's Disease Neuroimaging Initiative (ADNI) database was used to validate the proposed ANN-DKPCA method. The results show that the proposed algorithm is effective and robust compared with other recent works.