A number of people with physical disabilities have difficulty performing any physical movement and would benetit trom a direct brain interface, an interface that accepts commands directly from the brain. The University of Michigan Direct Brain Interface (UMDBI) research partnership is a collaboration which includes the Departments of Biomedical Engineering, Electrical Engineering and Computer Science, Physical Medicine and Rehabilitation, Neurology, Surgery and Radiology from the University of Michigan; the Departments of Neurology from the Henry Ford Hospital, and the Institute of Biomedical Engineering from the Technical University Graz. These partners propose to address the functional evaluation of a direct brain interface and the optimization of detection methods used in the direct brain interface. The (time-domain based) template matching detection method developed by the UM-DBI has demonstrated sufficient accuracy in off-line experiments to warrant real-time, on-line implementation and testing with subjects at the University of Michigan and Henry Ford Hospitals who have implanted electrodes for purposes related to epilepsy surgery. (While these subjects are not members of the target user population, the presence of implanted cortical electrodes in these subjects provides a unique opportunity for direct brain interface development). The proposed functional evaluation includes: 1) Development of an on-line, real-time testing system for direct brain interface methods; 2) Examination of the ability of subjects to learn to voluntarily improve event-related potential (ERP) quality and detection performance given appropriate feedback; 3) Determination of the accuracy and speed with which a direct brain interface can be used to perform functional tasks; and 4) Identification of the relationship between the location of electrocorticogram (ECoG) recorded brain events and the activated portion of the brain as observed through functional magnetic resonance imaging. Improvements in the accuracy by which brain events can be detected will be approached through development and optimization of time-domain based detection methods (performed primarily at UM) and evaluation of the performance of frequency-domain based detection methods on ECoG (performed primarily at Graz). In addition, off-line analysis will be used to I) Investigate the ability of current detection methods to differentiate between brain activity related to different actions and 2) Determine the increased accuracy of event detection achievable using ECoG versus EEG. The proposed research is intended to conclusively demonstrate that a direct brain interface based on the detection of human ERPs recorded intracranially can be used for control of functional tasks. While a simple direct brain interface would be a valuable tool for people with severe disabilities, it is intended that an initial interface would also form the foundation for future generations of direct brain interfaces of ever increasing complexity which would rely on advanced signal processing methods (such as those explored here). Beyond the scope of the proposed work, the results of these studies will form the foundation for clinical testing of the direct brain interface with individuals from the target user populations using subdural electrodes.
| Brunner, Clemens; Graimann, Bernhard; Huggins, Jane E et al. (2005) Phase relationships between different subdural electrode recordings in man. Neurosci Lett 375:69-74 |
