A Data-Driven Investigation on Surface Electromyography Based Clinical Assessment in Chronic Stroke

Abstract

Background: Surface electromyography (sEMG) based robot-assisted rehabilitation systems have been adopted for chronic stroke survivors to regain upper limb motor function. However, the evaluation of rehabilitation effects during robot-assisted intervention relies on traditional manual assessments. This study aimed to develop a novel sEMG data-driven model for automated assessment. Method: A data-driven model based on a three-layer backpropagation neural network (BPNN) was constructed to map sEMG data to two widely used clinical scales, i.e., the Fugl-Meyer Assessment (FMA) and the Modified Ashworth Scale (MAS). Twenty-nine stroke participants were recruited in a 20-session sEMG-driven robot-assisted upper limb rehabilitation, which consisted of hand reaching and withdrawing tasks. The sEMG signals from four muscles in the paretic upper limbs, i.e., biceps brachii (BIC), triceps brachii (TRI), flexor digitorum (FD), and extensor digitorum (ED), were recorded before and after the intervention. Meanwhile, the corresponding clinical scales of FMA and MAS were measured manually by a blinded assessor. The sEMG features including Mean Absolute Value (MAV), Zero Crossing (ZC), Slope Sign Change (SSC), Root Mean Square (RMS), and Wavelength (WL) were adopted as the inputs to the data-driven model. The mapped clinical scores from the data-driven model were compared with the manual scores by Pearson correlation. Results: The BPNN, with 15 nodes in the hidden layer and sEMG features, i.e., MAV, ZC, SSC, and RMS, as the inputs to the model, was established to achieve the best mapping performance with significant correlations (r > 0.9, P < 0.001), according to the FMA. Significant correlations were also obtained between the mapped and manual FMA subscores, i.e., FMA-wrist/hand and FMA-shoulder/elbow, before and after the intervention (r > 0.9, P < 0.001). Significant correlations (P < 0.001) between the mapped and manual scores of MASs were achieved, with the correlation coefficients r = 0.91 at the fingers, 0.88 at the wrist, and 0.91 at the elbow after the intervention. Conclusion: An sEMG data-driven BPNN model was successfully developed. It could evaluate upper limb motor functions in chronic stroke and have potential application in automated assessment in post-stroke rehabilitation, once validated with large sample sizes. Clinical Trial Registration: www.ClinicalTrials.gov, identifier: NCT02117089.

Keywords: chronic stroke; clinical assessment; data-driven model; surface electromyography; upper limb.

Conflict of interest statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2021 Ye, Yang, Nam, Xie, Chen and Hu.

Figures

Figure 1

The experimental setup and representative sEMG signals. (A) The sEMG-driven robotic hand system. (B) The setup for data acquisition during the vertical bare hand evaluation task. (C) The representative raw sEMG trials for the four muscles of TRI, BIC, ED, and FD in a vertical bare hand evaluation task. Each movement during the evaluation task was manually marked by the experiment operator.

Figure 2

(A) The procedure of sEMG signal processing. (B) The structure of the three-layer backpropagation neural network (BPNN) data-driven model.

Figure 3

Significant correlations yielded by the four-channel sEMG signals, i.e., ED, FD, BIC, and TRI, before the 20-session intervention. Correlations between the mapped scores and manual (A) Fugl–Meyer Assessment shoulder/elbow (FMA-SE) scores and (B) Fugl–Meyer Assessment wrist/hand (FMA-WH) scores.

Figure 4

Significant correlations yielded by the four-channel sEMG, i.e., ED, FD, BIC, and TRI, after the 20-session intervention. Correlations between the mapped scores and manual (A) Fugl–Meyer Assessment shoulder/elbow (FMA-SE) scores and (B) Fugl–Meyer Assessment wrist/hand (FMA-WH) scores.

Figure 5

Significant correlations yielded by the two-channel sEMG signals after the 20-session intervention. Correlation between the mapped scores and manual (A) Fugl–Meyer Assessment shoulder/elbow (FMA-SE) scores, from muscles pair of BIC and TRI, (B) Fugl–Meyer Assessment wrist/hand (FMA-WH) scores, from muscle pair of ED and FD.

Figure 6

Correlations yielded by the mismatched testing condition with the four-channel sEMG signals, i.e., ED, FD, BIC, and TRI. Correlation between the mapped scores and manual (A) Fugl–Meyer Assessment shoulder/elbow (FMA-SE) scores and (B) Fugl–Meyer Assessment wrist/hand (FMA-WH) scores.

Figure 7

Significant correlations between the mapped scores and MASs scales yielded by the bandpass filtered (10–200 Hz) four-channel sEMG signals, i.e., ED, FD, BIC, and TRI. The correlations between the mapped scores and MASs (A) at elbow for the pre-intervention dataset, (B) at elbow for the post-intervention dataset, (C) at wrist for the pre-intervention dataset, (D) at wrist for the post-intervention dataset, (E) at fingers for the pre-intervention dataset, and (F) at fingers for the post-intervention dataset.

Figure 8

The differences of sEMG parameters between the pre- and post-intervention sessions. The x-axis indicates the target muscles (BIC and FD) and muscle pairs (FD-BIC, FD-TRI, and BIC-TRI). The y-axis indicates the corresponding CI of the muscle pairs and the normalized sEMG activation level at the target muscles. The significant differences are indicated as * for P ≤ 0.05, ** for P ≤ 0.01, and ***for P ≤ 0.001, using the paired t-test.

Figure 9

The differences of clinical scales between the pre- and post-intervention sessions. (A) Sub-FMA scores, (B) MASs. The significant differences are indicated as * for P ≤ 0.05, ** for P ≤ 0.01, and ***for P ≤ 0.001, using the paired t-test for the FMA and the Wilcoxon test for the MAS.