Enhancing motor rehabilitation using a hybrid BCI paradigm: integrating high-frequency SSVEP with action observation and motor imagery

Both action observation (AO) and motor imagery (MI) can elicit activity in mirror neurons and the motor cortex. Therefore, brain–computer interface (BCI) systems based on AO and MI hold broad application prospects in the field of motor rehabilitation. An increasing number of studies have incorporated steady-state visual evoked potentials (SSVEP) into AO or MI paradigms to construct hybrid BCI systems and enhance robustness of the systems. In this study, AO, MI, and AO combined with MI experiments were designed, based on a SSVEP paradigm. In the AO experiment, subjects were required to observe alternating and flickering fisted- and extended-hand pictures. In the MI task, subjects were required to perform MI tasks while focusing on flickering extended-hand pictures. In the AO combined with MI experiment, subjects were required to execute the same task for the AO experiment while simultaneously perform the MI task. Task-discriminant component analysis and Tikhonov regularizing common spatio-spectral pattern algorithms were used to separately process inputs from the two modalities, with the system ultimately outputting a fusion result. The results found that AO combined with MI and MI alone more effectively activated the motor cortex compared with AO alone. The fusion classification accuracy of the proposed hybrid paradigm reached 86.42% ​± ​8.42%, 88.54% ​± ​10.31%, and 88.91% ​± ​9.61% for AO, MI, and AO combined with MI, respectively. This study provided an alternative for the construction of a more robust and comfortable rehabilitation system.