EEG-based BCI evaluation was a non-invasive way to assess the brain function of stroke patients. With higher temporal resolution, it can be used to see the brain activation change when patients were asked to perform a specific task like motor imagery (MI). For patients with hand dysfunction after stroke, specific task-based EEG evaluation can not only be used to assess brain function but also be a reference in BCI rehabilitation system improvement suitable for stroke patients.

The research recruited 31 stroke patients with hand and upper extremity motor dysfunction. Patients were asked to perform motor imagery task according to the tips on the screen. EEG signals of stroke patients were recorded. BCI performances of this system were evaluated. We chose 12 and 14 channels to respectively calculate the BCI performances. They total 12 are C3, FC1, FC5, CP1, CP5, P3 in the left hemisphere and C4, FC2, FC6, CP2, CP6, P4 in the right hemisphere. In addition, we add F3 and F4 to a total 14 channels.

We found that in the lower alpha-α, the BCI performance of ischemia was significant higher than that of hemorrhage (P<0.05) both in the 12 and 14 channels. We also found that the higher the BCI accuracy, the higher FMA score the patient will get (P<0.05). And there was no statistical significance in other EEG frequencies related to motor imagery task.

The brain activation and BCI accuracy can be used as monitor and assessment for stroke patients. The difference in BCI accuracy between ischemia and hemorrhage may be an improvement reference in BCI system. For hemorrhagic patients, the BCI feedback threshold may be lower in other to get a better interaction. And the correlative result between FMA scores and BCI accuracies suggests that the person with higher hand function, the higher BCI accuracy he may get.