Feature extraction and selection methods for motor imagery EEG signals: A review

Extraction and selection of electroencephalography (EEG) features is a pivotal task. The brain-computer interface (BCI) for motor imagery (MI) task is analysed with respect to the classification accuracies in following described papers. The paper gives a brief discussion on various feature extraction and selection techniques that has been studied for different motor imagery functions. The comparison table is made with respect to the features extraction methods, selection methods, EEG data used for analysis, number of electrodes for data acquisition, computation time and method implemented. Different techniques such as JayaNFCSSCGLH, LPSVD, sparse weighted CSP, IMF, CBN, SBCSP are discussed. Flowcharts for every method is discussed. The techniques determines the defining characteristic in the procedure that helps in producing better signal for analysing and differentiating brain signal at it utmost probability. Lastly the discussion is made as to which technique outperformed when motor imagery task is taken into consideration for the (BCI) brain-computer interfacing mechanism. To clarify better the classification accuracies of studied methods are compared.