A Brain–Computer Interface System Based on Magnetoencephalography With Optically Pumped Magnetometers

This study presents a brain-computer interface (BCI) approach that uses optically pumped magnetometers (OPMs)-based magnetoencephalography (MEG) associated with motor imagery (MI) for the synchronous operation of a brain-controlled virtual drone in 3-D space. This BCI system combines signal acquisition, command processing and communication, interface, and the virtual drone control strategy. Individuals are asked to control a virtual drone to avoid specified obstacles by progressively completing MI tasks. For that, four self-developed OPMs are used to measure the MEG. We focus on the feature quantification of $\boldsymbol {\mu }$ / $\boldsymbol {\beta }$ rhythms-related desynchronization (ERD) and event-related synchronization (ERS), and use machine learning as a classifier to determine the motor intentions of subjects, which we then translate into machine language to operate external equipment. Experiments were used to confirm the viability and efficiency of the suggested strategy. The accuracy rates of classification were up to 86.7%, while perfect classification was gained in the train set. The preliminary results indicate that the proposed BCI approach might accomplish the intended objective and has potential to be utilized for practical applications.