الفهرس | Only 14 pages are availabe for public view |
Abstract The ability to bridge the communication gap between man and machines through manmachine communication interfaces has led to the innovative use of human-computer interaction systems. Moreover, BCI (Brain Computer Interface), a widely accepted humancomputer interaction system, has gained high popularity among the neuroscientific community. Advancement in biomedical signal processing techniques has directed the Electroencephalography (EEG) signals not only as a diagnostic tool for brain disease, but also as a controller in BCI field. BCI systems translate raw acquired brain signals into commands to control an external device. BCI gives new control and communication that does not depend on the brain’s normal output channels of peripheral nerves and muscles. In this thesis, the reactivity of EEG rhythms in association with imagery of hand movements was studied, as a step towards building BCI system to assist people with disabilities. EEG signal was preprocessed by removing EOG artifact then band pass filtered from (8-30) Hz. First qualitative analysis was made to investigate EEG signal from figures and plots. Secondly discriminative features for EEG signal were extracted by four variant approaches (autoregressive parameters, DWT coefficients, PSD and bispectrum). Combinations between features from those approaches were made and form five sets of features. Each set was classified by four variants of classifiers (LDA, SVM, logistic regression and ensemble discriminant). Study investigation showed that set 2 features with logistic regression was the best among others in terms of accuracy and kappa coefficient. the results of the proposed technique (depending on feature set 2 and logistic regression) showed that right-hand imagery decreases the activity of hand area in the brain left side and left-hand imagery decreases the activity of hand area in the brain right side. The results successfully show that motor imagery EEG phenomena can be utilized in a BCI based motor restoration. Time domain, frequency domain and combined time-frequency features are extracted with emphasis on recognizing discriminative features representing EEG trials recorded during imagination of two hand movements. Then, classification of the imagined action into two hand movements is carried out. Abstract III The proposed method is evaluated using EEG signals from nine subjects during motor imagery tasks. The performance obtained by the proposed method is evaluated using different metrics. The Kappa value has been calculated to rate and compare the proposed technique with previous studies applied to the same dataset (2b from competition IV). The comparison shows that the proposed method achieved a high position among other competitors. Moreover, the proposed method utilized a resized small input that depends only on two electrodes, which reduces calculation complexity, cost and resulting in a relatively faster training. Moreover, using only two electrodes in our proposal open the window to a portable BCI |