Brain computer interface training with motor imagery and functional electrical stimulation for patients with severe upper limb paresis after stroke: a randomized controlled pilot trial

Abstract

Background: Restorative Brain-Computer Interfaces (BCI) that combine motor imagery with visual feedback and functional electrical stimulation (FES) may offer much-needed treatment alternatives for patients with severely impaired upper limb (UL) function after a stroke.

Objectives: This study aimed to examine if BCI-based training, combining motor imagery with FES targeting finger/wrist extensors, is more effective in improving severely impaired UL motor function than conventional therapy in the subacute phase after stroke, and if patients with preserved cortical-spinal tract (CST) integrity benefit more from BCI training.

Methods: Forty patients with severe UL paresis (< 13 on Action Research Arm Test (ARAT) were randomized to either a 12-session BCI training as part of their rehabilitation or conventional UL rehabilitation. BCI sessions were conducted 3-4 times weekly for 3-4 weeks. At baseline, Transcranial Magnetic Stimulation (TMS) was performed to examine CST integrity. The main endpoint was the ARAT at 3 months post-stroke. A binominal logistic regression was conducted to examine the effect of treatment group and CST integrity on achieving meaningful improvement. In the BCI group, electroencephalographic (EEG) data were analyzed to investigate changes in event-related desynchronization (ERD) during the course of therapy.

Results: Data from 35 patients (15 in the BCI group and 20 in the control group) were analyzed at 3-month follow-up. Few patients (10/35) improved above the minimally clinically important difference of 6 points on ARAT, 5/15 in the BCI group, 5/20 in control. An independent-samples Mann-Whitney U test revealed no differences between the two groups, p = 0.382. In the logistic regression only CST integrity was a significant predictor for improving UL motor function, p = 0.007. The EEG analysis showed significant changes in ERD of the affected hemisphere and its lateralization only during unaffected UL motor imagery at the end of the therapy.

Conclusion: This is the first RCT examining BCI training in the subacute phase where only patients with severe UL paresis were included. Though more patients in the BCI group improved relative to the group size, the difference between the groups was not significant. In the present study, preserved CTS integrity was much more vital for UL improvement than which type of intervention the patients received. Larger studies including only patients with some preserved CST integrity should be attempted.

Fig. 1

The RecoveriX BCI system

Fig. 2

The flow of patients through the study

Fig. 3

Distribution of scores on Action Research Arm Test (ARAT) at baseline (blue) and 3 months post-stroke for the BCI group and the control group. The green bars denote patients with improvement of at least 6 points on ARAT, the red bars with a change of less than 6 on ARAT

Fig. 4

Boxplots show the EEG performance indices for each patient are depicted, namely: the left plot shows broadband (8–30 Hz) ERD for the affected hemisphere during less affected hand imagery for the values reflecting the start and end of therapy, and the right plot broadband LI during less affected hand imagery. Gray lines connect the values of the respective index computed at the start and the end of the BCI treatment for each patient. The asterisk (*) symbol is used to indicate statistically significant differences (p < 0.05) between paired values, representing the start and end of the therapy