Background: After stroke, wrist extension dyscoordination precludes functional arm/hand. We developed a more spatially precise brain signal for use in brain computer interface (BCI's) for stroke survivors.
New method: Combination BCI protocol of real-time functional magnetic resonance imaging (rt-fMRI) sequentially followed by functional near infrared spectroscopy (rt-fNIRS) neurofeedback, interleaved with motor learning sessions without neural feedback. Custom Matlab and Python code was developed to provide rt-fNIRS-based feedback to the chronic stroke survivor, system user.
Results: The user achieved a maximum of 71 % brain signal accuracy during rt-fNIRS neural training; progressive focus of brain activation across rt-fMRI neural training; increasing trend of brain signal amplitude during wrist extension across rt-fNIRS training; and clinically significant recovery of arm coordination and active wrist extension.
Comparison with existing methods: Neurorehabilitation, peripherally directed, shows limited efficacy, as do EEG-based BCIs, for motor recovery of moderate/severely impaired stroke survivors. EEG-based BCIs are based on electrophysiological signal; whereas, rt-fMRI and rt-fNIRS are based on neurovascular signal.
Conclusion: The system functioned well during user testing. Methods are detailed for others' use. The system user successfully engaged rt-fMRI and rt-fNIRS neurofeedback systems, modulated brain signal during rt-fMRI and rt-fNIRS training, according to volume of brain activation and intensity of signal, respectively, and clinically significantly improved limb coordination and active wrist extension. fNIRS use in this case demonstrates a feasible/practical BCI system for further study with regard to use in chronic stroke rehab, and fMRI worked in concept, but cost and some patient-use issues make it less feasible for clinical practice.
Keywords: BCI; BMI; Biofeedback; Coordination; MRI; NIRS; Neural feedback; Stroke; motor control.
Published by Elsevier B.V.