Individuals with severe impairments may use brain-computer interface (BCI) technologies in order to interact with their external environment. One non-invasive brain-monitoring technology which may be suitable for this purpose is transcranial Doppler ultrasound (TCD). Previous research has shown that TCD is useful in detecting changes in cerebral blood flow velocities after the performance of cognitive tasks which are often lateralized towards a specific hemisphere of the brain. However, to date, TCD has not been used in a BCI system. This thesis first explores TCD in an offline study, showing that on average, accuracies of 80.0% are attainable with user-specific training data and 74.6% with user-independent training data. Furthermore, consecutive sequential lateralizations do not decrease classification accuracies. In a subsequent online experiment, a TCD-BCI system yielded an average accuracy of 61.4%, but revealed key findings about the effects of user motivation and error streaks in an online system.
| Identifer | oai:union.ndltd.org:TORONTO/oai:tspace.library.utoronto.ca:1807/33315 |
| Date | 20 November 2012 |
| Creators | Aleem, Idris Syed |
| Contributors | Chau, Tom |
| Source Sets | University of Toronto |
| Language | en_ca |
| Detected Language | English |
| Type | Thesis |
Page generated in 0.0018 seconds