Combined Action Observation, Motor Imagery and Steady State Motion Visual Evoked Potential Based Brain Computer Interface System

dc.contributor.authorRavi, Aravind
dc.date.accessioned2024-11-12T20:50:09Z
dc.date.available2024-11-12T20:50:09Z
dc.date.issued2024-11-12
dc.date.submitted2024-10-03
dc.description.abstractStroke is one of the leading causes of long-term acquired disability in adults worldwide. Gait recovery is a major objective in post-stroke rehabilitation programs. Conventional gait therapy encourages patient involvement, but the results can be slow and/or limited, leading to sub-optimal recovery. Active patient involvement, collaboration, and motivation are key factors that promote efficient motor learning. Therefore, there is a need to develop novel rehabilitation strategies to enhance user engagement by utilizing their movement intent. Brain-computer interfaces (BCIs) based on electroencephalography (EEG) offer an attractive approach for rehabilitation as they enable an alternative method for active participation in therapy. Current visual BCIs provide high decoding accuracy but typically do not activate sensorimotor areas critical for motor recovery. Conversely, BCIs based on motor imagery (MI) activate motor areas but suffer from high inter-subject variability and long user training, resulting in poorer movement intent detection accuracy and potentially leading to high cognitive demand. This thesis proposed a novel BCI paradigm called CAMS—Combined Action Observation (AO), Motor Imagery (MI), and Steady-State Motion Visual Evoked Potentials (SSMVEP). The CAMS paradigm aimed to induce acute changes in movement-related areas of the cortex through the observation and imagery of gait movements, activating both motor and visual cortices to elicit SSMVEP-like responses. Furthermore, the responses elicited by the CAMS paradigm were investigated in two distinct applications to detect user movement intent with the aim of actively engaging the participant. The research conducted across three studies investigates the efficacy of CAMS in enhancing cortical excitability, decoding gait phases, and improving asynchronous visual BCI performance. Twenty-five healthy volunteers participated in this study wherein they observed and imagined lower limb movements of gait as part of the CAMS intervention, which was compared with an SSMVEP control condition. Study I aimed to investigate the acute changes in cortical excitability induced by the CAMS intervention. The results demonstrated significant increases in movement-related cortical potential (MRCP) components, indicating enhanced cortical excitability. For instance, the magnitude of BP1 at channel C1 increased from -1.41 ± 0.54 µV pre-intervention to -3.23 ± 0.5 µV post-intervention (p = 0.009), highlighting the potential of CAMS to engage motor- related brain areas and promote neuroplasticity. Study II focused on decoding the phases of gait (swing and stance) from EEG responses elicited by the CAMS paradigm. Using Linear Discriminant Analysis (LDA) and Support Vector Machine (SVM), the study achieved a classification accuracy of 75% and 78%, respectively, in decoding the swing and stance phases of gait. Study III introduced a novel detection algorithm based on Complex Convolutional Neural Networks (C-CNN) for asynchronous offline CAMS BCI. The C-CNN method achieved high F1-scores for asynchronous operation. Median F1-scores for C-CNN were 0.88 (W=1s), 0.92 (W=2s), and 0.96 (W=3s), with corresponding False activation rates (FARs) of 0.34, 0.30, and 0.27. Additionally, larger stimulus frequency differences resulted in stronger visual BCI classification performance, with combinations (7.5 Hz, 12 Hz) and (8.57 Hz, 12 Hz) yielding the highest accuracies of 87% and 78%, respectively. These findings underscore the potential of the CAMS BCI paradigm in enhancing cortical excitability, eliciting responses for decoding gait phases, and improving asynchronous visual BCI performance while simultaneously engaging the movement related areas of the cortex. By providing a comprehensive investigation of the CAMS paradigm, this work contributes to existing knowledge and helps guide future clinical applications in neurorehabilitation.
dc.identifier.urihttps://hdl.handle.net/10012/21180
dc.language.isoen
dc.pendingfalse
dc.publisherUniversity of Waterlooen
dc.subjectbrain computer interface
dc.subjecteeg
dc.subjectelectroencephalography
dc.subjectmachine learning
dc.subjectcams
dc.subjecthybrid bci
dc.subjectssvep
dc.subjectmotor imagery
dc.subjectstroke rehab
dc.subjectrehabilitation
dc.subjectneural engineering
dc.subjectneurorehabilitation
dc.subjectstroke
dc.subjectbrain damage
dc.subjectgait
dc.subjectgait phases
dc.titleCombined Action Observation, Motor Imagery and Steady State Motion Visual Evoked Potential Based Brain Computer Interface System
dc.typeDoctoral Thesis
uws-etd.degreeDoctor of Philosophy
uws-etd.degree.departmentSystems Design Engineering
uws-etd.degree.disciplineSystem Design Engineering
uws-etd.degree.grantorUniversity of Waterlooen
uws-etd.embargo.terms0
uws.contributor.advisorTung, James
uws.contributor.advisorJiang, Ning
uws.contributor.affiliation1Faculty of Engineering
uws.peerReviewStatusUnrevieweden
uws.published.cityWaterlooen
uws.published.countryCanadaen
uws.published.provinceOntarioen
uws.scholarLevelGraduateen
uws.typeOfResourceTexten

Files

Original bundle

Now showing 1 - 1 of 1
Loading...
Thumbnail Image
Name:
Ravi_Aravind.pdf
Size:
2.31 MB
Format:
Adobe Portable Document Format

License bundle

Now showing 1 - 1 of 1
No Thumbnail Available
Name:
license.txt
Size:
6.4 KB
Format:
Item-specific license agreed upon to submission
Description: