This application requests support for continuing studies using novel EEG signal recording and analysis techniques as a control tool for communication or operation of prosthetic devices. Over the past five years, the applicants have demonstrated that, with appropriate subject training, coupled with a variety of methods for signal processing of the resulting EEG signals, that normal and impaired subjects can use brain electrical signals to control a computer interface to enhance communication. The applicants propose to continue this line of research, relying on their three initially enunciated hypotheses which were: Increasing online adaptability of the interface will improve its performance. Time-domain EEG components can augment control now provided by frequency-domain analyses. The interface can support cursor-based menu selection, and operate a neuroprosthesis. To test the first hypothesis, the online algorithm will be expanded to incorporate automatic selection of optimal EEG components, electrode locations and frequencies in these components, as well as optimal spatial filters and gain functions. The utility of these changes in improving the performance of the interface will be assessed. To test the second hypothesis, time-domain EEG components, such as slow cortical potentials and error-rated potentials will be added to the online algorithm and their capacity to supplement the control provided by frequency-domain components alone will be assessed. To test the third hypothesis, the interface will be applied to cursor- based letter or icon selection and to the operation of a neuroprosthesis designed and tested by the FES group at Case Western Reserve University. If successful, this work is likely to improve performance and versatility of EEG-based communication and interface systems.
Showing the most recent 10 out of 83 publications
