Electroneurophysiological Signals for Movement-Based Brain-Computer Interface Applications
- Author(s): McCrimmon, Colin Matthew
- Advisor(s): Nenadic, Zoran
- et al.
Brain-computer interfaces (BCIs) translate neural signals into machine commands for applications such as non-verbal communication, brain-controlled prostheses, motor rehabilitation, and even entertainment. However, there are knowledge gaps and practical limitations of non-invasive (e.g. electroencephalographic) and invasive (e.g. electrocorticographic (ECoG)) BCIs that need to be addressed before these devices can be used by patients, clinicians, and the community. For example, it was unclear whether BCI use is safe and effective for post-stroke movement physiotherapy, so we conducted a phase I clinical trial in stroke survivors demonstrating that non-invasive BCI physiotherapy does not worsen gait function, and may even be beneficial. However, for BCIs to be widely adopted by patients and clinicians for at-home/outpatient rehabilitation, they need to be available as small, portable, inexpensive systems. Therefore, we developed one such system and showed that these design constraints do not compromise decoding performance compared to large, expensive conventional BCIs. For individuals with paraplegia or tetraplegia due to spinal cord injury (SCI), invasive BCIs are ideal for prosthetic use (e.g. exoskeleton control of limb movements/walking) due to their superior spatiotemporal resolution (and therefore decoding accuracy) and their capacity to restore proprioceptive movement sensation. For practical reasons, these BCIs should be developed as implantable systems. However, current digital signal processors that are amenable to implantation have limited computational resources for decoding. Therefore, studies that enhance our understanding of the electroneurophysiological changes during movements are needed in order to design efficient and effective hardware/software for implantable BCIs. In one preliminary ECoG study, we demonstrated that the amplitude of gamma-band (40-200 Hz) signals from the motor cortex (M1) was associated with force output during upper-extremity movements. In another study, we observed that the amplitude and envelope frequency of leg M1 gamma-band signals were related to the duration and stepping rate of human walking. Findings such as these can be incorporated into the design of future, fully-implantable BCI prostheses for restoring movement in SCI survivors.