BCI Research at the ISRC, University of Ulster N. Ireland, UK

1. BCI Research at the ISRC, University of UlsterN. Ireland, UK By Dr. Girijesh Prasad Intelligent Systems Research Centre, School of Computing and Intelligent Systems, Faculty of Engineering, University of Ulster, Magee Campus, N. Ireland, UK e-mail: g.prasad@ulster.ac.uk ISRC URL: http://isrc.infm.ulst.ac.uk/

2. RTD facilities BCI research expertise Ongoing research work Project Proposal Presentation outline

3. We are part of a recently launched £20M centre of excellence in Intelligent Systems with core activities in: Software and hardware implementations of neural networks, fuzzy systems, Gas Bio-inspired Intelligent Systems – Brain Modeling Brain computer Interfacing (BCI) Intelligent Robotics Embedded systems Re-configurable computing/FPGA systems Multiple valued logic systems Machine vision Intelligent process control BCI laboratory facilities: A state-of-art BCI experimental setup with 56 EEG channels, 8 EMG channels An 8 EEG/EMG channel mobile unit A 4 EEG channel wireless mobile unit EMF screened room Multiple electrode systems RTD Facilities

4. Main BCI research objectives: To devise a robust BCI system that can self-organise and adapt to each individual’s EEG autonomously and to the inherent day-to-day changes To effectively account for uncertainties inherently present in EEG due to non-stationary brain dynamics and varying noise characteristics of the measurement environment RTD work Undertaken and ongoing : Time-series prediction approach for pre-processing and feature extraction Comprehensive and conclusive analysis of best spectral approaches to feature extraction Type-2 fuzzy logic (T2FL) approach for classifier design Design of neurofeedback A self-organising non-parametric BCI Real-time implementation and extensive experimental evaluation on more than 14 subjects over several months Expertise

5. Expertise …cont. Type-2 Fuzzy Logic-based Methodology EEG trial Feature Extraction C3 C3 feature vector C4 feature vector C4 T2 FL-based Classifier Class label THEN IF LEFT or RIGHT hand imagery • Promising results: Significantly higher robustness and classification accuracy

6. Ongoing research work Advanced modelling approaches for improving feature separability Enhancing mental practice through BCI-based neurofeedback for post stroke rehabilitation Self-adaptive Asynchronous Brain-Computer Interface Participation in large-scale national and EU projects e.g. SenseMaker - a multi-sensory, task-specific adaptable perception system, (IST-2001-34712)Funded under EUFP5 on Life-like Perception, (0.5M Euros), 2002-05. Expertise …cont.

7. An EEG-based brain-computer interface for neurofeedback. Development of an appropriate neurofeedback to motivate subjects for focussed mental practice of rehabilitation exercises over long-term. Industrial involvement for prototype product development. Incorporation of a robotic system that facilitates physical exercises in case of extreme disability. Involvement of health specialists such as neuro-psychologists. Provision for modifications of therapeutic strategy dependent on intelligent sensory data (e.g. EMG and EEG) analysis, and human responses. An Intelligent System for Post-stroke Rehabilitation: A project proposal

8. Partners are sought with relevant expertise. We are also interested in joining other consortia for: BCI Research and Development Advanced Intelligent Systems for assistive technology Computational Intelligence and cognitive Systems Project consortia

9. Robustness, uncertainty handling, and adaptability are key challenges. Our TSP approach as an EEG data preprocessing procedure significantly enhances effectiveness of feature extraction procedures. T2FLS based classification approach shows potential in more effective handling of signal variability and uncertainty. Further algorithmic enhancement is needed for developing a practical BCI. Planned ongoing work is addressing this. Project partners are sought. Conclusions

10. Questions Thank you