Brain Computer Interface

Flinders University and Medical Centre have long been leaders in the area of Brain Computer Interface with Prof. Richard Clark demonstrating a computer distinguishing yes from no back in the 90s. The Flinders approach to Brain Computer Interface has been unique in its focus on processing time, location and relationship of sensory-motor, cognitive and affective events in the cortex, whilst other approaches have tended to be based on detecting broad frequency characteristics and often, implicitly or explicitly, depend on biofeedback to achieve good performance.

Brain Computer Interface can take two forms: 1. monitoring naturally occurring brain activity to develop 'neuromarkers' corresponding to particular cognitive events, or 2. wiring the brain to particular devices which can then be controlled by thinking in a specific way, both of which depend on 3. detecting and eliminating muscular artefact and noise.

1. Neuromarkers

Much of our work on Neuromarkers has focussed on learning. Part of this depends on being able to use electroencephalography (EEG) reliably even whilst the subject is performing normal, even physical, activities. Our work with the Australian Defense Science and Technology Organization (DSTO) has had this focus on physical skill acquisition.

2. Device Control

Our work on device control has been focussed on driving a wheelchair by thought control, and we have FCRGS funding support from Flinders University in partnership with FMDAT partner Novitatech to take our proof of concept work (as part of the successful PhD research of Sean Fitzgibbon with David Powers and Richard Clark), and develop a real-time wheelchair control model. An important aspect of device control is to separate background non-directive thoughts from thought commands directed at the wheelchair. As with the Neuromarker work, it is important to be able to use the BCI in everyday conditions which means dealing with muscular artefact and noise.

3. Artefact and Noise

A major focus of the group has been how to identify and deal with muscular contamination (artefact) and noise. Conventionally, in cognitive neuroscience laboratory experiments, one discards trials that are contaminated by an eye blink or similar. One aspect of the group's work has shown that there is much more contamination than has traditionally been acknowledged, and that higher gamma range frequencies are particular susceptible. Conversely, the group has developed techniques that allow removal of muscular contamination, and we are currently studying how much of this invisible contamination we can eliminate.

Chief Investigator(s)

David Powers, Artificial Intelligence Lab, School of Computer Science, Engineering & Mathematics
Richard Clark, Cognitive Neuroscience Lab, Psychology
Kenneth Pope, Artificial Intelligence Lab, School of Computer Science, Engineering & Mathematics
John Willoughby, Human EEG Lab, Medicine
Sean Fitzgibbon, Human EEG Lab, Medicine
Trent Lewis, Artificial Intelligence Lab, School of Computer Science, Engineering & Mathematics