Effects and neural mechanisms of a brain-computer interface-controlled soft robotic glove on upper limb function in patients with subacute stroke: a randomized controlled fNIRS study

Abstract

Background and purpose: The brain-computer interface-based soft robotic glove (BCI-SRG) holds promise for upper limb rehabilitation in subacute stroke patients, yet its efficacy and neural mechanisms are unclear. This study aimed to investigate the therapeutic effects and neural mechanisms of BCI-SRGs by functional near-infrared spectroscopy (fNIRS).

Methods: Forty subacute stroke patients with left-sided hemiparesis were randomized into the BCI-SRG (n = 20) and soft robotic glove (SRG) (n = 20) groups. Both groups received 20 sessions of intervention over 4 weeks in addition to conventional rehabilitation. The BCI-SRG group was trained using a soft robotic glove controlled by a brain‒computer interface (BCI), whereas the SRG group used the same soft robotic glove without BCI control. The clinical outcomes included the Action Research Arm Test (ARAT), the Fugl-Meyer Assessment Upper Limb (FMA-UL), and Modified Barthel Index (MBI) scores. In addition, fNIRS was used to explore potential clinical brain mechanisms. All assessments were performed before treatment and after 4 weeks of treatment.

Results: A total of 39 participants completed the intervention and clinical assessments (BCI-SRG: n = 20; SRG: n = 19). Compared with the SRG group, the BCI-SRG group showed greater improvements in the ARAT (Z = - 2.139, P = 0.032) and FMA-UL (Z = - 2.588, P = 0.010), with no notable difference in the MBI (Z = - 1.843, P = 0.065). fNIRS data were available for 35 participants (BCI-SRG: n = 17; SRG: n = 18). Within-group comparisons revealed significant postintervention increases in cortical activation in the bilateral sensorimotor cortex (SMC) and medial prefrontal cortex (MPFC) in the BCI-SRG group, whereas no significant changes were observed in the SRG group. Between-group comparisons further revealed significantly greater changes in HbO concentrations in the BCI-SRG group than in the SRG group across the same cortical regions. Moreover, changes in prefrontal activation (post-pre) were positively correlated with improvements in ARAT scores, with significant correlations observed in the left dorsal lateral prefrontal cortex (LDLPFC) (Ch9, r = 0.592, P = 0.012; Ch25, r = 0.488, P = 0.047) and right dorsal lateral prefrontal cortex (RDLPFC) (Ch19, r = 0.671, P = 0.003).

Conclusions: BCI-SRG training significantly enhances upper limb function and facilitates bilateral motor and sensory cortical reorganization. PFC activation is correlated with functional improvements, suggesting a potential mechanism underlying the benefits of rehabilitation in stroke patients.

Trial registration: This trial was registered under the Chinese Clinical Trial Registry (ChiCTR2400082786) and was retrospectively registered on April 8, 2024.

Keywords: Brain‒computer interface; Functional near-infrared spectroscopy; Soft robotic glove; Stroke; Upper limb.

Fig. 1

CONSORT flow diagram

Fig. 2

System design and protocol for BCI-SRG training. a A participant engaged in the BCI-SRG training session and wore the soft robotic glove while EEG signals were recorded. b EEG electrode placement according to the 10–20 system. c Soft robotic glove worn by the participant and equipped with sensors for feedback. d Configuration of fNIRS channels. e Task fNIRS measurement protocol, detailing a 5-cycle procedure of preparation, grasping task, rest, and end. L: left hemisphere; R: right hemisphere

Fig. 3

Flowchart of the BCI-SRG training protocol, outlining the sequence of tasks: preparation, motor imagery, task execution, and feedback

Fig. 4

Comparison of clinical scale scores between the BCI-SRG and SRG groups at different time points. a ARAT, action research arm test; b FMA-UL, Fugl-Meyer assessment for upper limb; c MBI, modified Barthel index. ***P < 0.001, *P < 0.05

Fig. 5

Cortical activation maps of HbO levels. a Changes in brain activation in the BCI-SRG group before and after the intervention; brilliant colours (red to bright yellow) indicate increased cortical activation. b Group comparison of mean HbO activation differences during the grasping task, highlighting significantly higher activation levels in the BCI-SRG group than in the SRG group. The colour bar represents t values. c The channels with significant differences within the BCI-SRG group are highlighted in red. d Channels with significant differences between groups (BCI-SRG vs. SRG) are highlighted in red. L: left hemisphere; R: right hemisphere

Fig. 6

Correlations between changes in HbO levels and ARAT score differences in the BCI-SRG group. Scatter plots show significant positive correlations between changes in HbO levels (Post–Pre) during the grasping task and ARAT score improvements (∆ARAT) in channel 9 (LDLPFC, r = 0.592, p = 0.012), channel 19 (RDLPFC, r = 0.671, p = 0.003), and channel 25 (LDLPFC, r = 0.488, p = 0.047). The solid lines represent the regression fit, and the dashed lines indicate the 95% confidence intervals. javascript:void(0);