Optimized Feature Selection Technique for Automatic Classification of MRI Images for Alzheimer's Disease

摘要

Alzheimer's disease (AD) is becoming more common among the elders globally. Physical and mental assessment were the norm to identify AD, however Magnetic Resonance Imaging (MRI) is being utilized for automatic medical analysis/interpretation. Recently, many higher dimensional classification techniques were suggested for discriminating between patients with Alzheimer's, "Mild Cognitive Impairment (MCI)" or normal brain automatically sing brain images obtained from MRI. For effective classification features are extracted from CerebroSpinal Fluid (CSF), White Matter (WM) and Grey Matter (GM) as well as the classifier trained. Feature extraction faces the "curse of dimensionality" leading to poor classifier performance. In the current study, Fish Swarm Metaheuristic is studied for effective features selection. Experiments show the effectiveness of the proposed technique.