Key Screw and Cylindrical Grasp Motion Classfication from Same Hand

摘要

Brain Computer Interface(BCI) has enormous potential to improve the life style of disabled person. BCI is a system which creates a parallel path of communication between human brain and prosthesis devices. Preliminary requirements of development of a efficient BCI system are suitable feature extraction and classification techniques. EEG signal measures brain electrical activity placing electrodes over scalp. Analysis of Motor imagery or Motor Executionmovementis very popular method of developing BCI system. This paper investigates the possibility of discriminating between the EEG associated with cylindrical and key screw grasp movements. The EEG was recorded from four subjects as they executed and imagined two essential hand movements from same hand. Important features in frequency (fast Fourier transforms) as well as time frequency domain (wavelet) have been extracted to get useful information from the of fourteen channel EEG data of four healthy individuals. Motor execution as well as Motor imagery grasping tasks has been classified using liner discriminant analysis (LDA) and Na?ve Bias algorithms. Highest classification accuracy of 92.37% has been achieved using LDA and RMS of frequency domain spectra for motor imagery and 87.41% has been achieved using LDA and Statistical parameter of wavelet for motor execution. This shows that EEG discriminant between two hand grasping movements is possible. The research introduces a new combination of motor tasks to BCI based devices.