Dingle's model-based EEG peak detection using a rule-based classifier

Abstract

The employment of peak detection algorithm is prominent in several clinical applications such as diagnosis and treatment of epilepsy patients, assisting to determine patient syndrome, and guiding paralyzed patients to manage some devices. In this study, the performances of four different peak models of time domain approach which are Dumpala's, Acir's, Liu's, and Dingle's peak models are evaluated for electroencephalogram (EEG) signal peak detection algorithm. The algorithm is developed into three stages: peak candidate detection, feature extraction, and classification. Rule-based classifier with an estimation technique based on particle swarm optimization (PSO) is employed in the classification stage. The evaluation result shows that the best peak model is Dingle's peak model with the highest test performance is 88.78%.