نازنين عليزاده يوالاري

Brain-computer interfaces (BCI) have begun to revolutionize many aspects of human life ranging from health to smart living and communication devices. The primary goal of a BCI system is to provide a pathway for controlling an external device via electrical activity of the brain. Therefore, BCI technologies require accurate recognition and classification systems of the brain responses to a variety of motor imagery (MI) movements through electroencephalogram (EEG) signals. The following methods are suggested and eva‎luated in this work with the intention of enhancing BCI system performance in the categorization of various MI-EEG tasks: 1. In the first proposed framework, to reach an improved performance in MI-EEG signal classification, we proposed a new strategy based on continuous wavelet transform (CWT)-based time-frequency maps to generate two-dimensional (2D) EEG images of adaptively reconstructed signals from the extracted multivariate mode decomposition (MVMD) modes. In the next step, the extracted signals are projected in the new space by means of the proposed multi-class common spatial pattern (MCCSP) filtering. The resulting images are then fed to the convolutional neural network (CNN) architectures (AlexNet and LeNet). The proposed framework has the benefit of achieving high classification accuracy to be attained even with a significant amount of input data. LeNet and AlexNet reach the best average accuracy rates of 95.33% and 93.66% on dataset 1 from BCI competition IV and the results on dataset 2a from competition IV are more promising than the current state of the arts. Our results depict that MI-EEG task recognition using image classification approaches, along with CNNs, is as comparable or even superior to other existing traditional approaches and provides high potential for upcoming research. The second proposed method presented a novel approach for classifying MI-EEG mental tasks using empirical mode decomposition (EMD)-based signal processing and MCCSP. With the EMD algorithm, intrinsic mode functions (IMFs) are selected by comparing correlation coefficients between the original bandpass-filtered EEG signals and each decomposed IMF. The main advantages of the suggested framework include reducing computational burden by choosing the most effective IMF, improving feature consistency, generalizing spatial patterns in multi-class modes, reducing feature dimension through MCCSP, and achieving high accuracy classification with a small number of feature vectors. The experiments were eva‎luated using dataset 2a of the BCI competition IV (22 channels) containing four MI tasks involving the right hand, left hand, foot, and tongue. Finally, the performance of the proposed feature extraction method is investigated using several classifiers. Based on the results from BCI-C IV-2a, the K-nearest neighbor (K-NN) achieved the highest mean classification accuracy rate of 91.28%. Our findings recommend that a more elevated accuracy performance of 96.71% can be reached by raising the feature dimension. Also compared to state-of-the-art algorithms, the performance of the proposed method is highly convincing and motivating for future studies.

رابط مغز رايانه، الكتروانسفالوگرافي، تصورات حركتي، الگوهاي فضايي مشترك چندكلاسه، شبكه عصبي كانولوشني، تجزيه حالت تجربي.

Brain-computer interface, electroencephalogram, motor imagery, multi-class common spatial pattern, convolutional neural network, empirical mode decomposition

nazanin alizadeh yuvalari

Dr. M. R. Mosavi