Feedback Optimization for Restorative Brain-Computer Interfaces

Abstract

A brain-computer interface (BCI) provides an alternative communication channel for the human brain to directly interact with computers or machines. This technology has enabled patients with locked-in-syndrome to communicate with the outside world that otherwise would be impossible. It also promises recovery to stroke patients by supplying a platform to practice motor imagery of their impaired motor functions and receive feedback. The latter application is called motor imagery based BCI (MI-BCI) and has already provided promising results for stroke rehabilitation. However, its widespread application necessitates optimization. This thesis investigates enhancement ofMI-BCIs for stroke rehabilitation through feedback optimization, exploring the feedback modality (proprioceptive and visual) effect on BCI performance. It suggests that proprioceptive feedback is the superior choice for therapeutic BCIs. Next, it compares the effect of a short and a long proprioceptive feedback update interval (FUI) on BCI performance. It concludes that people with short reaction time benefit more from a short FUI whereas their slower counterparts show improved performance with motor imagery practice using a long FUI. In another study, which was run as a proof-of-principle study, we find a significant improvement in one stroke patient hand movement, after attending MI-BCI training sessions optimised through our findings on FUI length and proprioceptive feedback. Overall, the research outcomes in this thesis highlight the effects of feedback modality and feedback update interval on MI-BCI performance. Furthermore, the single case study on a stroke patient provides primary evidence and motif for larger studies on the efficacy of the proposed strategies to enhance MI-BCI performance in stroke rehabilitation.