Neural activity modulations and motor recovery following brain-exoskeleton interface mediated stroke rehabilitation

Abstract

Brain-machine interfaces (BMI) based on scalp EEG have the potential to promote cortical plasticity following stroke, which has been shown to improve motor recovery outcomes. However, the efficacy of BMI enabled robotic training for upper-limb recovery is seldom quantified using clinical, EEG-based, and kinematics-based metrics. Further, a movement related neural correlate that can predict the extent of motor recovery still remains elusive, which impedes the clinical translation of BMI-based stroke rehabilitation. To address above knowledge gaps, 10 chronic stroke individuals with stable baseline clinical scores were recruited to participate in 12 therapy sessions involving a BMI enabled powered exoskeleton for elbow training. On average, 132 ± 22 repetitions were performed per participant, per session. BMI accuracy across all sessions and subjects was 79 ± 18% with a false positives rate of 23 ± 20%. Post-training clinical assessments found that FMA for upper extremity and ARAT scores significantly improved over baseline by 3.92 ± 3.73 and 5.35 ± 4.62 points, respectively. Also, 80% participants (7 with moderate-mild impairment, 1 with severe impairment) achieved minimal clinically important difference (MCID: FMA-UE >5.2 or ARAT >5.7) during the course of the study. Kinematic measures indicate that, on average, participants' movements became faster and smoother. Moreover, modulations in movement related cortical potentials, an EEG-based neural correlate measured contralateral to the impaired arm, were significantly correlated with ARAT scores (ρ = 0.72, p < 0.05) and marginally correlated with FMA-UE (ρ = 0.63, p = 0.051). This suggests higher activation of ipsi-lesional hemisphere post-intervention or inhibition of competing contra-lesional hemisphere, which may be evidence of neuroplasticity and cortical reorganization following BMI mediated rehabilitation therapy.

Keywords: Brain-machine interface; Clinical trial; Exoskeletons; Movement related cortical potentials; Stroke rehabilitation.

Conflict of interest statement

N.B. and J.C. have a patent issued (US10,092,205 granted October 9, 2018), which presents methods for detecting motor intentions from EEG signals, including MRCPs. All the remaining authors report no competing interests.

Copyright © 2020 The Author(s). Published by Elsevier Inc. All rights reserved.

Figures

Graphical abstract

Fig. 1

CONSORT flow diagram showcasing patient recruitment, intervention and follow-ups.

Fig. 2

EEG-based BMI control of MAHI exoskeleton for stroke rehabilitation. A) Timeline for the clinical study protocol. B) Schematic representation of the experiment setup, showing a stroke participant’s impaired elbow being trained by the MAHI Exo-II, while EEG and EMG activity are recorded. In this BMI scheme, successful detection of motor intent from EEG is validated against residual EMG activity from impaired arm, before a Go or Wait command is issued to the exoskeleton. A computer screen in front of the participant, cues start and end of trial and provides simultaneous visual feedback of the movement.

Fig. 3

Longitudinal BMI performance. BMI performance in 10 chronic stroke survivors over 12 therapy sessions, averaged by session in sub-plot A and averaged by online testing vs. calibration in sub-plot B. From top to bottom, mean ± s.d. values for BMI’s prediction accuracy, false positives, early detection time, and user approval rating are shown. Results from 2 participants (P9 and P7) with best and worst BMI accuracy are overlaid on the plots. Dotted lines indicate statistically significant trends in accuracy and user rating.

Fig. 4

Motor recovery post-intervention. Clinical outcome metrics assessed post-treatment (post-tt) and at 2-week (2wk f/u) and 2-months (2mon f/u) follow-ups relative to baseline. Shaded regions indicate the 4 – 6 weeks long intervention period. Underneath each data point, the number of scores that were averaged to calculate the mean value are shown.

Fig. 5

Detailed breakdown of motor recovery. Breakdown of FMA-UE and ARAT scores by subscales, shown by averaging across participants (subplot A & B) and individually (subplots C-F) for participants that achieved minimal clinically important difference. Subplots C-F, further group participants based on their FMA-UE and ARAT outcomes at 2 months follow-up. The arrows in subplots A & B indicate the order of administering the test, starting at the first item and then progressing counter-clockwise.

Fig. 6

Improvement in movement quality between start and end of therapy. Movement quality was derived from joint angle velocity using various kinematic metrics. For all metrics except Number of Peaks, an increase in value corresponds to improvement.

Fig. 7

Correlation (ρ) between MRCP amplitude and functional assessment scales. Subfigures A & B compare MRCP amplitudes from central and centro-parietal EEG electrodes with clinical outcomes. In these figures, numbers represent participant I.Ds and the dashed lines represent regression lines between changes in MRCP amplitude versus clinical scores. Subfigures C & D show MRCPs recorded from all the participants at start and end of therapy. Note, MRCPs are aligned with respect to movement onset (t = 0 s).