Flexible analytic wavelet transform in a EEG based brain computer Interface Paradigm: a study in end users with mo-tor disabilities

Motor imagery electroencephalogram based brain computer interface systems can help people with disabilities to communicate with an external device and to realize rehabilitation therapies. The paper proposes flexible analytic wavelet transform (FAWT) as feature extraction method. The method was tested on a dataset that contains EEG signals acquired from subjects with motor disabilities. Classifiers linear discriminant analysis (LDA), quadratic discriminant analysis (QDA), k nearest neighbors(kNN), Mahalanobis distance (MD) and support vector machine (SVM) were utilized to classsify the extracted features of right hand motor imagery and feet motor imagery (FEET). The best performance was given by QDA classifier with a classification rate of 97 %, sensitivity 99.65%, specificity 98.47%, kappa coefficient 0.97 and F1 score 0.98. The proposed method shows through the obtained results that can be used and easy to implement for assisting rehabitation on real time BCI systems.