Andrea Navares Juanco

MSc Semester Project


Classification of Motor imaginary mental tasks for brain computer interfacing


Andrea Navares Juanco


Prof. Dr. Touradj Ebrahimi

Assistant: Ashkan Yazdani


September 29, 2010




A brain computer interface (BCI) can be realized by using visual evoked potentials (VEP) or movement related potentials (MRP). This project concentrates on the usage of MRPs for the EEG-based BCI. Clear representations of MRP can be observed in the EEG’s mu-rhythm (8-12Hz) and/or Beta rhythm when a person performs a motor activity or imagines a motor activity. Such an activity can be easily captured from EEG channels C3, and C4. Current designs of BCI usually consist of four main stages, which are, raw signal acquisition, signal pre-processing or conditioning, feature extraction, and finally classification of features into intended actions. Among these stages, feature extraction and classification methods plays an important role because a successful BCI depends on its ability to extract EEG features according to different tasks and to efficiently classify them in a real time environment.
The following tasks should be performed in this project:

  • Reviewing the literature about EEG signal processing, BCI modalities, and motor imaginary signal extraction.
  • Studying an available dataset used for BCI competition and analyzing it.
  • Studying state of the art for preprocessing the EEG signals, feature extraction, and classification of EEG signals and selecting an appropriate methodology.
  • Implementing the selected methodology and testing it on the dataset.
  • Implementing the selected preprocessing, feature extraction, and classification methods.
  • Assessing and comparing the results of different features extracted from the signal.
  • Comparing the results obtained in this study, with results of other participants in the BCI competition.