Upper-limb functional assessment after stroke using mirror contraction: A pilot study

Highlights

• The sEMG features bias between bilateral arms are changed after stroke.

• Mirror contraction of bilateral arm contributes to EMG-based stroke recognition and severity assessment.

• The bilateral-arms differences of stroke patients are linear correlated with Fugl–Meyer scores.

• Intelligent stroke evaluation based on sEMG analytics can be automated with less involvement of therapists.

Abstract

The clinical assessment after stroke depends on the rating scale, usually lack of quantitative feedback such as biomedical signal captured from stroke patients. This study attempts to develop a unified assessment framework for persons after stroke via surface electromyography (sEMG) bias from bilateral limbs, based on four types of selected movements, namely forward lift arm, lateral lift arm, forearm internal/external rotation, forearm pronation/supination. Eleven healthy subjects and six stroke patients are recruited to participate in the experiment to perform the bilateral-mirrored paradigm with six channels of sEMG signals recorded from each of their arms. The linear discriminant analysis (LDA), random forest algorithm (RF) and support vector machine (SVM) are adopted, trained and used for stroke patients qualitative recognition. The bilateral bias diagnosis algorithm (BBDA) is developed to evaluate the stroke severity quantitatively based on the similarity index (SI) of the sEMG. The results reveal that: (1) the sEMG feature bias of bilateral arms for stroke patients is different from that of healthy people; (2) the RF and SVM demonstrate a better performance with an average recognition accuracy of 0.92 ± 0.12 and 0.93 ± 0.12 than LDA (0.84 ± 0.20) in distinguishing stroke patients from healthy subjects; (3) there is a strong positive correlation between SI and the Fugl-Meyer score (r = 0.93). These research findings indicate that the dominant qualitative assessment after stroke could be complementary by its counterpart quantitative solutions, and stroke rehabilitation could be automated with less involvement of professional therapists.

Introduction

Stroke is a common global health-care problem that has had serious negative impact on the life quality of the patients [1]. Accurate and prompt evaluation of stroke will increase eligibility of patients to receive proper therapy. According to the basis that whether the output depends on the therapists clinical experience as well as their subjective judgment of the patients’ limb movements, the assessment methods of upper limb and hand motor function are categorized as subjective and objective evaluations. The subjective evaluations in clinical settings mainly include different scales, such as Fugl–Meyer Assessment [2], Brunnstrom Assessment [3], modified Ashworth Scale [4], and Barthel Index [5], which are manually performed by experienced therapist focusing on muscle, motor pattern, changes of upper limb or hand function evaluation. Though the human-administered clinical scales are the accepted standard for quantifying motor performance after stroke, these measurement methods depend on the clinical experience of therapist causing limitation of interrater and intrarater reliability. They are also limited in terms of sensitivity for reflecting muscle changes and being time-consuming to apply clinically. Therefore, the objective method is required to improve the efficiency, sensitivity and reliability for the upper-limb function evaluation in stroke rehabilitation.

The objective evaluations for stroke are based on the state-of-the-art human–machine systems, which assess stroke through human biological signal analysis and the interaction between stroke patients and machine. For example, computed tomography (CT) as well as magnetic resonance imaging (MRI) protocols provide excellent tools for the evaluation of acute ischemic stroke [6], Golestani et al. reported that the functional magnetic resonance imaging (fMRI) demonstrated the impact of stroke on functional connections throughout the brain, which could be used for stroke evaluation even during the resting state of a patient [7]. Robotics have been proved to be a powerful measurement tool to assess the rehabilitation process of stroke [8], Semrau et al. used robotics to characterize the magnitude and timing of proprioceptive as well as motor recovery, and found robotic measures were correlated with clinical measures [9]. Besides, ultrasound is another kind of alternative biomedical signals applied in human–machine interface for neural rehabilitation and shows promising results in detecting muscle deformation [10], [11]. However, the aforementioned methods are along with expensive, complicated and bulky devices. Some other stroke assessment methods are more portable and easier to apply, such as the inertial measurement unit [12], surface electromyography (sEMG) and electroencephalogram (EEG) [13]. And compared with EEG, sEMG is with higher signal-to-noise ratio and more convenient to acquire.

The sEMG signals, containing a wealth of physiological information, have been widely applied in human–machine interfaces [14], [15], [16]. Many research efforts have been focused on applying EMG to stroke assessment and rehabilitation. Zhang et al. proposed the EMG-based closed-loop torque control in functional electrical stimulation (FES) showing a potential for limb function reconstruction after stroke [17]. Li et al. applied muscle and motor unit (MU) indices extracted from sEMG to evaluate muscle changes post stroke [18]. Kallenberg et al. applied high-density sEMG to investigate MU characteristics of the biceps brachii in poststroke patients and found that ratio of root-mean-square (RMS) value of MU action potentials on the affected side divided by that on the unaffected side correlated significantly with the Fugl–Meyer score [19]; Hu et al. made use of EMG representing the affected hand muscles activation levels and their co-contraction indexes for a quantitative evaluation of motor functional recovery process in chronic stroke patients during wrist training [20], and they also monitored the variations in the muscular coordination patterns by EMG parameters across the whole EMG-driven robot hand training sessions to understand the progress of the training quantitatively [21]. Besides, motor impairment after stroke typically affects one side of the body [1], thus the healthy side is involved in the rehabilitation process and the bimanual training is an effective rehabilitation strategy based on natural inter-limb coordination [22]. Leonardis et al. applied the sEMG of unaffected side to drive robotic hand exoskeleton for bilateral rehabilitation [23]. And the sEMG bias of the two arms was applied to drive FES for stroke rehabilitation in [24]. These studies indicate that EMG signals can be used to represent neuromuscular pathological state and motion intention in stroke rehabilitation. However, currently, sEMG technologies are mainly applied to be the trigger signal for rehabilitation devices or as a kind of metrics for monitoring the intra-patients rehabilitation process in clinical settings and researches [19]. It is challenging for inter-patients sEMG analysis, because sEMG is a kind of time-varying and non-stationary signal with huge variation across different people. Thus, there are still lack of effective and universal stoke auxiliary diagnosis methods based on sEMG.

Based on the aforementioned understandings, the upper-limb function evaluation method based on bilateral-arms sEMG was proposed and tested in this study, in order to drive the stroke auxiliary diagnosis research frontier boundary forward. As shown in Fig. 1, the aim of this pilot study is to investigate the feasibility of stroke patients recognition based on pattern recognition techniques and to assess the severity of stroke quantitatively, based on the sEMG features and their bias between bilateral arms.