Main Investigator: Prof. Kenneth P. Camilleri, Centre for Biomedical Cybernetics
Main Investigator: Prof. Tracey Camilleri, Department of Systems and Control Engineering
Research Support Officer: Dr Natasha Padfield, Centre for Biomedical Cybernetics
Externally funded: Xjenza Malta through the SINO-MALTA Fund 2023 Call (Science and Technology Cooperation) EUR 181,808.
Speech imagery (SI) is a brain-computer interface (BCI) paradigm which can enable subjects to intuitively control external devices such as graphical user interfaces or robots in a hands-free manner, by using just their thoughts. However, the widespread use of the SI paradigm has been impeded by the relatively low decoding accuracies obtained when using electroencephalogram (EEG) data, which is a leading non-invasive method of reading brain signals. These low accuracies lead to poor and unrobust BCI performance. This project aims to investigate novel and innovative signal processing, machine learning, and deep learning techniques to improve the accuracy of SI decoding. It also proposes an investigation into fundamental aspects of SI, including the impact of background noise (such as music) on data quality, and how the detection rate of words varies for a vast lexicon. Finally, the findings of this project will be further explored and validated through the implementation of online SI BCIs that the user can interact with. From a technical perspective, this project aims to investigate the scalp regions and frequency bands that are most important for SI decoding. It also aims to investigate the efficacy of various classifiers for SI decoding as well as novel knowledge-based, and collaborative learning techniques to improve the decoding of SI. To conclude, this project aims to make significant contributions towards the improved decoding of SI data from EEG data, which is fundamental for the widespread adoption of practical SI-based BCIs.
