An efficient classification scheme for ADHD problem based on Binary Coded Genetic Algorithm and McFIS

In this paper, we propose an efficient classification scheme for Attention Deficit Hyperactivity Disorder (ADHD) problem. Presented classifier is a combination of well known Meta - Cognitive Neuro-Fuzzy Interface System (McFIS) approach and proposed feature selection mechanism based on Binary Coded Genetic Algorithm coupled with Extreme Learning Machine (BCGA-ELM). In this research we choose magnetic resonance images (MRI) of hippocampus area taken from ADHD-200 data base. Each MRI scan is processed by computing population averages and applying region of interest mask. The set of hippocampus features is then processed by proposed BCGA-ELM approach. In BCGA-ELM each hippocampus feature represents as one bit value. “1” means that the feature is chosen for learning, “0” means the feature is skipped. BCGA-ELM performs a search for set of features with the most promising performances. The chosen features is then used as input for Meta - Cognitive Neuro-Fuzzy Inference System (McFIS) learning approach. Experiment results show that the presented BCGA-ELM-McFIS perform well in chosen hippocampus area.