A reliable and reproducible real-time access to sensorimotor rhythm with a small number of optically pumped...

Objective.Recent advances in biomagnetic sensing have led to the development of compact, wearable devices capable of detecting weak magnetic fields generated by biological activity. Optically pumped magnetometers (OPMs) have shown significant promise in functional neuroimaging. Brain rhythms play a crucial role in diagnostics, cognitive research, and neurointerfaces. Here we demonstrate that a small number of OPMs can reliably capture sensorimotor rhythms (SMRs).Approach.We conducted movement execution and motor imagery (MI) experiments with nine participants in two distinct magnetically shielded rooms (MSRs), each equipped with different ambient field suppression systems. We used only 4 OPMs located above the sensorimotor region and standard common-spatial-patterns (CSPs) based processing to decode the real and imaginary movement intentions of our participants. We evaluated reproducibility of the CSP components' spectral profiles and assessed the decoding accuracy deterioration with reduction of OPM's count. We also assessed the influence of the magnetic field orientation on the decoding accuracy and implemented a real-time MI brain-computer interface (BCI) solution.Main results.Under optimal conditions, OPM sensors deliver informative signals suitable for practical MI BCI applications. Those subjects who participated in the experiments in both MSRs exhibit highly reproducible SMR spectral patterns across two different magnetically shielded environments. The magnetic field components with radial orientation yield higher decoding accuracy than their tangential counterparts. In some subjects we observed more than 80% of binary decoding accuracy using a single OPM sensor. Finally we demonstrate real-time performance of our system along with clearly pronounced and behaviorally relevant fluctuations of the SMR power.Significance.For the first time, we demonstrated reliable and reproducible tracking of SMR components using a small number of contactless OPM sensors during movement execution and MI. Our findings pave the way for more efficient post-stroke neurorehabilitation by enabling MI-based BCI solutions to accelerate functional recovery.