Comparison of EEG Pattern Recognition of Motor Imagery for Finger Movement Classification

The detection of a hand movement beforehand can be a beneficent tool to control a prosthetic hand for upper extremity rehabilitation. To be able to achieve smooth control, the intention detection is acquired from the human body, especially from brain signal or electroencephalogram (EEG) signal. However, many constraints hamper the development of this brain-computer interface (BCI, especially for finger movement detection). Most of the researchers have focused on the detection of the left and right-hand movement. This article presents the comparison of various pattern recognition method for recognizing five individual finger movements, i.e., the thumb, index, middle, ring, and pinky finger movements. The EEG pattern recognition utilized common spatial pattern (CSP) for feature extraction. As for the classifier, four classifiers, i.e., random forest (RF), support vector machine (SVM), k-nearest neighborhood (kNN), and linear discriminant analysis (LDA) were tested and compared to each other. The experimental results indicated that the EEG pattern recognition with RF achieved the best accuracy of about 54%. Other published publication reported that the classification of the individual finger movement is still challenging and need more efforts to make the best performance