Motor Imagery Classification of EEG for Elbow Movement Using SVM and PNN as Signal Classification

A study on classification of motor imagery for right hand elbow movement has been worked in this research. Motivated by the need to help post-stroke patient regain their motoric function, this paper proposed to use motor imagery activities for extracting movement intention of the patient. This research aims to discover combination of signal processing to achieve best classification accuracy of elbow movement based on electroencephalography signals. The classifications are based according to phenomena of event-related synchronization (ERS) and event-related desynchronization (ERD). The approach has been designed by doing a following task and that is signal acquisition using the Neurostyle Electroencephalograph System, selecting channel based on previous research, bandpass filtering using 5^th-order Butterworth filter and spectrum analysis utilizing Fast Fourier Transform (FFT), feature extraction of maximum band power of mu and beta frequency range, and frequency of mu and beta at maximum band power, and then classification using Support Vector Machine (SVM) and Probabilistic Neural Network (PNN) classifier. The classification result of using SVM with utilized features provide experimental accuracy at 41.44%, while PNN achieved accuracy at 54.33%.