Automated FES for Upper Limb Rehabilitation Following Stroke and Spinal Cord Injury

Abstract

Neurorehabilitation aims to induce beneficial neural plasticity in order to restore function following injury to the nervous system. There is an increasing evidence that appropriately timed functional electrical stimulation (FES) can promote associative plasticity, but the dosage is critical for lasting functional benefits. Here, we present a novel approach to closed-loop control of muscle stimulation for the rehabilitation of reach-to-grasp movements following stroke and spinal cord injury (SCI). We developed a simple, low-cost device to deliver assistive stimulation contingent on users' self-initiated movements. The device allows repeated practice with minimal input by a therapist, and is potentially suitable for home use. Pilot data demonstrate usability by people with upper limb weakness following SCI and stroke, and participant feedback was positive. Moreover, repeated training with the device over 1-2 weeks led to functional benefits on a general object manipulation assessment. Thus, automated FES delivered by this novel device may provide a promising and readily translatable therapy for upper limb rehabilitation for people with stroke and SCI.

Fig. 1.

A – The automated FES device. Participants reached for the cube, grasped it and pulled it toward themselves along a rail of length 300 mm. When released, the cube automatically returned to the start position. Assistive stimulation was delivered by an Odstock Medical OS2CHS stimulator, modified to be controlled by a microcontroller (Arduino Micro) which received input from digital proximity sensors (Sharp GP2Y0D810Z0F) at either end of the rail. B – To stimulate wrist and finger extension the active electrode was positioned over extensor digitorum communis (EDC), and the indifferent electrode over extensor pollicis longus (EPL) and abductor pollicis longus (AbPL). To stimulate extension of the arm, the active electrode was placed over the anterior deltoid and the indifferent electrode over the triceps.

Fig. 2.

System schematic: The triggers for the two channel stimulator were controlled by a microcontroller. This received inputs from two sets of proximity sensors and used these signals to provide stimulation to open the hand and extend the arm at appropriate times during the reaching and grasping cycle.

Fig. 3.

A diagram showing the closed-loop created by the device, stimulator, controller and participant in this study.

Fig. 4.

The intervention protocol: 1.The participant was given an auditory (double beep) and visual cue (LED on) to reach and grasp the 5cm cube, and FES was given to open the hand and, in most cases, extend the arm. 2. When proximity sensors (10cm range) detected that the open hand was over the block (marked by a single beep, LED off), the FES was turned off allowing the block to be gripped. 3. The participant pulled the block to the finish position with no FES assistance. 4. A proximity sensor detected the return was complete (single beep) and the microcontroller initiated a 1.5s delay. 5. Cues (single beep, LED on) indicated that the block should be released and FES was applied to open the hand. 6. When proximity sensors detected that the release was complete (the block was in the start position), FES was turned off (single beep, LED off). The participant then rested for 5 seconds before returning to step 1. Timings shown were calculated using data from participants with SCI (n = 7) for a block of 25 trials on day 3 of the intervention. Timings (mean (± SE)) are: Reach 1.4s (±0.2), Grasp and Pull 1.0s (±0.15), Hold 1.5s, Release 0.9s (±0.07), and Rest 5s. Similar timings were observed for participants with stroke.

Fig. 5.

ARAT scores for stroke survivors completing the pilot study as assessed by the blinded, independent assessor. Assessments were completed before the intervention period, immediately after, and 1 week and 1 month after the completion of the intervention period. For reference, the original assessor’s scores for the before condition were: 10, 14, 29 and 3. Participant 4, who is not shown due to an incomplete dataset, had an original assessor score of 31. ^* indicates visit 1 and ^** indicates visit 2 for Participant 1, which were separated by 6 months.

Fig. 6.

Panel A - The ARAT scores for the trained side for participants with SCI before and after the intervention. Panel B - The ARAT scores for the untrained side before and after the intervention. ARAT scores are as assessed by the blinded, independent assessor. For reference, the original assessor’s scores for the before condition for participants 1 to 7 were (trained / untrained): 8 / 7, 35 / 5, 16 / 55, 27 / 57, 41 / 56, 30 / 34 and 35 / 39 respectively.

Fig. 7.

The mean change in ARAT score for the trained and untrained sides for participants with SCI. values show the statistical significance measured using the paired two-sided Wilcoxon signed-rank test for between the before and after conditions on the train and untrained sides (n = 7, and 5.5 respectively), and between the two sides (n = 7, ). Error bars show standard error.

Fig. 8.

A selection of the qualitative data collected using a Likert scale. The number of respondents was 7, 4 and 9 for the SCI, stroke and physiotherapist groups respectively. ^* Participants stated that they would require assistance with initial set-up and placing of electrodes, but could otherwise use the device independently. ^** Participants often added the caveat that they would require training. Note that percentages may not add to 100% due to rounding error.