ABSTRACT
Neurofeedback training consisting of movement control and sensory feedback based on brain-computer interface (BCI) is a promising rehabilitation therapy developed for post-stroke patients with motor dysfunctions. By allowing patients proactively conduct closed-loop rehabilitation training, the therapy guarantees direct brain participation and promotes brain functional reorganization, thus breaking through the bottleneck and improving the effect of motor rehabilitation training. This study conducts an in-depth investigation on the motor neurofeedback training system incorporating hand function rehabilitation robots and reveals the current development of the neurofeedback training system by introducing paradigms inducing motor neuron activities, and the neural decoding methodologies. We also propose a three-layer neural network model based on the characteristics of the motor-sensory rhythms and fine-tuned for personalized protocols with quantitative comparisons. Finally, we discuss future directions to investigate BCI-based motor rehabilitation concerning the efficacy of the training system in both paradigms designing and neural decoding.
