Search

To develop multi-functionalhuman-machine interfaces that can help disabled people reconstruct lost functions of upper-limbs, machine learning (ML) and deep learning (DL) techniques have been widely implemented to decode human movement intentions from surface electromyography (sEMG) signals. However, due to the high complexity of upper-limb movements and the inherent non-stable characteristics of sEMG, the usability of ML/DL based control schemes is still greatly limited in practical scenarios. To this end, tremendous efforts have been made to improve model robustness, adaptation, and reliability. In this article, we provide a systematic review on recent achievements, mainly from three categories: multi-modal sensing fusion to gain additional information of the user, transfer learning (TL) methods to eliminate domain shift impacts on estimation models, and post-processing approaches to obtain more reliable outcomes. Special attention is given to fusion strategies, deep TL frameworks, and confidence estimation. Research challenges and emerging opportunities, with respect to hardware development, public resources, and decoding strategies, are also analysed to provide perspectives for future developments.

KEY POINTS Amyotrophic lateral sclerosis (ALS) is an incurable neurodegenerative disorder of motor neurons, carrying a short survival. High-density motor unit recordings permit analysis of motor unit size (amplitude) and firing behaviour (afterhyperpolarization duration and muscle fibre conduction velocity). Serial recordings from biceps brachii indicated that motor units fired faster and with greater amplitude as disease progressed. First-recruited motor units in the latter stages of ALS developed characteristics akin to fast-twitch motor units, possibly as a compensatory mechanism for the selective loss of this motor unit subset. This process may become maladaptive, highlighting a novel therapeutic target to reduce motor unit vulnerability. ABSTRACT Amyotrophic lateral sclerosis (ALS) is a progressive neurodegenerative disorder with a median survival of 3 years. We employed serial high-density surface electromyography (HDSEMG) to characterize voluntary and ectopic patterns of motor unit (MU) firing at different stages of disease. By distinguishing MU subtypes with variable vulnerability to disease, we aimed to evaluate compensatory neuronal adaptations that accompany disease progression. Twenty patients with ALS and five patients with benign fasciculation syndrome (BFS) underwent 1-7 assessments each. HDSEMG measurements comprised 30 min of resting muscle and 1 min of light voluntary activity from biceps brachii bilaterally. MU decomposition was performed by the progressive FastICA peel-off technique. Inter-spike interval, firing pattern, MU potential area, afterhyperpolarization duration and muscle fibre conduction velocity were determined. In total, 373 MUs (ALS = 287; BFS = 86) were identified from 182 recordings. Weak ALS muscles demonstrated a lower mean inter-spike interval (82.7 ms) than strong ALS muscles (96.0 ms; P = 0.00919) and BFS muscles (95.3 ms; P = 0.0039). Mean MU potential area (area under the curve: 487.5 vs. 98.7 μV ms; P

The study assessed motor unit loss in muscles paralyzed by spinal cord injury (SCI) using a novel compound muscle action potential (CMAP) scan examination. The CMAP scan of the first dorsal interosseous (FDI) muscle was applied in tetraplegia (n = 13) and neurologically intact (n=13) subjects. MScanFit was used for estimating motor unit numbers in each subject. The D50 value of the CMAP scan was also calculated. We observed a significant decrease in both CMAP amplitude and motor unit number estimation (MUNE) in paralyzed FDI muscles, as compared with neurologically intact muscles. Across all subjects, the CMAP (negative peak) amplitude was 8.01 ± 3.97 mV for the paralyzed muscles and 16.75 ± 3.55 mV for the neurologically intact muscles (p < 0.001). The CMAP scan resulted in a MUNE of 59 ± 37 for the paralyzed muscles, much lower than 108 ± 21 for the neurologically intact muscles (p < 0.001). No significant difference in D50 was observed between the two groups (p= 0.2). For the SCI subjects, there was no significant correlation between MUNE and CMAP amplitude, or any of the clinical assessments including pinch force, grip force, the Graded Redefined Assessment of Strength, Sensibility and Prehension (GRASSP) score, and SCI duration (p > 0.05). The findings provide an evidence of motor unit loss in the FDI muscles of individuals with tetraplegia, which may contribute to weakness and other hand function deterioration. The CMAP scan offers several practical benefits compared with the traditional MUNE techniques because it is noninvasive, automated and can be performed within several minutes.

Objective. This study aims to characterize the motor units (MUs) distribution and recruitment pattern in the spastic and non-spastic bilateral biceps brachii muscles (BBMs) of chronic stroke survivors. Approach. High-density surface electromyography (HD-sEMG) signals were collected from both spastic and non-spastic BBMs of fourteen chronic stroke subjects during isometric elbow flexion at 10%, 30%, 50% and 100% maximal voluntary contractions (MVCs). By combining HD-sEMG decomposition and bioelectrical source imaging, MU innervation zones (MUIZs) of the decomposed MUs were first localized in the 3D space of spastic and non-spastic BBMs. The MU depth defined as the distance between the localized MUIZ and its normal projection on the skin surface was then normalized to the arm radius of each subject and averaged at given contraction level. The averaged MU depth at different contraction levels on a specific arm side (intra-side) and the bilateral depths under a specific contraction level (inter-side) were compared. Main results. The average depth of decomposed MUs increased with the contraction force and significant differences observed between 10% vs 50% (p < 0.0001), 10% vs 100% (p < 0.0001) and 30% vs 100% MVC (p = 0.0017) on the non-spastic side, indicating that larger MUs with higher recruitment threshold locate in deeper muscle regions. In contrast, no force-related difference in MU depth was observed on the spastic side, suggesting a disruption of orderly recruitment of MUs with increase of force level, or the MU denervation and the subsequent collateral reinnervation secondary to upper motor neuron lesions. Inter-side comparison demonstrated significant MU depth difference at 10% (p= 0.0048) and 100% force effort (p= 0.0026). Significance. This study represents the first effort to non-invasively characterize the MU distribution inside spastic and non-spastic bilateral BBM of chronic stroke patients by combining HD-sEMG recording, EMG signal decomposition and bioelectrical source imaging. The findings of this study advances our understanding regarding the neurophysiology of human muscles and the neuromuscular alterations following stroke. It may also offer important MU depth information for botulinum toxin injection in clinical post-stroke spasticity management.

Motor unit number estimation (MUNE) studies of the anconeus muscle are very limited, although the information they provide is useful for neurophysiological investigations. The objective of this study was to estimate the number of motor units in the anconeus muscle.Compound muscle action potential scans of the anconeus muscle were recorded from 11 healthy participants, all of whom were tested on two occasions. MUNE was determined from the MScanFit program.The average MUNE of the anconeus muscle was 55.09 ± 3.27 (mean ± standard error of the mean) for the first test and 54.64 ± 3.70 for the retest, demonstrating excellent measurement reliability, with an intraclass correlation coefficient of 0.90.A relatively low motor unit number is found in the anconeus, a muscle not comprehensively studied in literature.

Background: Electrophysiological measures are being increasingly utilized due to their ability to provide objective measurements with minimal bias and to detect subtle changes with quantitative data on neural function. Heterogeneous reporting of trial outcomes limits effective interstudy comparison and optimization of treatment. Objective: The objective of this systematic review is to describe the reporting of electrophysiological outcome measures in spinal cord injury (SCI) clinical trials in order to inform a subsequent consensus study. Methods: A systematic search of PubMed and EMBASE databases was conducted according to PRISMA guidelines. Adult human SCI clinical trials published in English between January 1, 2008 and September 15, 2018 with at least one electrophysiological outcome measure were eligible. Findings were reviewed by all authors to create a synthesis narrative describing each outcome measure. Results: Sixty-four SCI clinical trials were included in this review. Identified electrophysiological outcomes included electromyography activity (44%), motor evoked potentials (33%), somatosensory evoked potentials (33%), H-reflex (20%), reflex electromyography activity (11%), nerve conduction studies (9%), silent period (3%), contact heat evoked potentials (2%), and sympathetic skin response (2%). Heterogeneity was present in regard to both methods of measurement and reporting of electrophysiological outcome measures. Conclusion: This review demonstrates need for the development of a standardized reporting set for electrophysiological outcome measures. Limitations of this review include exclusion of non-English publications, studies more than 10 years old, and an inability to assess methodological quality of primary studies due to a lack of guidelines on reporting of systematic reviews of outcome measures.

The spatial distributions of muscle innervation zone (IZ) and muscle fiber conduction velocity (CV) were examined in nine healthy young male participants. High-density surface electromyography (EMG) was collected from the biceps brachii muscle when subjects performed isometric elbow flexions at 20% to 80% of the maximal voluntary contraction (MVC). A total of 9498 samples of IZs were identified and CVs were calculated using the Radon transform. The center and width of IZ sample distribution were compared within four different force levels and six medial to lateral electrode column positions using repeated measures ANOVA and multiple comparison tests. Significant shifts of IZ center were observed in the medial columns (Columns 5, 6, and 7) compared with the lateral columns (Columns 3 and 4) (p < 0.05). Similarly, significant differences in the IZ width were found in Column 7 and 8 compared to Column 3 (p < 0.05). In contrast, muscle CV was unaffected by column position. Instead, muscle CV was faster at 40% and 80% MVC compared to 20% MVC (p < 0.05). The findings of this study add further insights into the physiological properties of the biceps brachii muscle.

Surface electromyography (EMG) recorded by a linear or 2-dimensional electrode array can be used to estimate the location of muscle innervation zones (IZ). There are various neurophysiological factors that may influence surface EMG and thus potentially compromise muscle IZ estimation. The objective of this study was to evaluate how surface-EMG-based IZ estimation might be affected by different factors, including varying degrees of motor unit (MU) synchronization in the case of single or double IZs. The study was performed by implementing a model simulating surface EMG activity. Three different MU synchronization conditions were simulated, namely no synchronization, medium level synchronization, and complete synchronization analog to M wave. Surface EMG signals recorded by a 2-dimensional electrode array were simulated from a muscle with single and double IZs, respectively. For each situation, the IZ was estimated from surface EMG and compared with the one used in the model for performance evaluation. For the muscle with only one IZ, the estimated IZ location from surface EMG was consistent with the one used in the model for all the three MU synchronization conditions. For the muscle with double IZs, at least one IZ was appropriately estimated from interference surface EMG when there was no MU synchronization. However, the estimated IZ was different from either of the two IZ locations used in the model for the other two MU synchronization conditions. For muscles with a single IZ, MU synchronization has little effect on IZ estimation from electrode array surface EMG. However, caution is required for multiple IZ muscles since MU synchronization might lead to false IZ estimation.

Muscle networks describe the functional connectivity between muscles quantified through the decomposition of intermuscular coherence (IMC) to identify shared frequencies at which certain muscles are co-modulated by common neural input. Efforts have been devoted to characterizing muscle networks in healthy subjects but stroke-linked alterations to muscle networks remain unexplored. Muscle networks were assessed for eight key upper extremity muscles during isometric force generation in stroke survivors with mild, moderate, and severe impairment and compared against healthy controls to identify stroke-specificalterations in muscle connectivity. Coherence matrices were decomposed using non-negative matrix factorization. The variance accounted for thresholding was then assessed to identify the number of muscle networks. Results showed that the number of muscle networks decreased in stroke survivors compared to age-matched healthy controls (four networks in the healthy control group) as the severity of post-stroke motor impairment increased (three in the mild- and two in the moderate- and severe-strokegroups). Statistically significant reductions of IMC in the synergistic deltoid muscles in the alpha-band in stroke patients versus healthy controls ( ${p}$ < 0.05) were identified. This study represents the first effort, to the best of our knowledge, to assess stroke-linked alterations in functional intermuscular connectivity using muscle network analysis. The findings revealed a pattern of alterations to muscle networks in stroke survivors compared to healthy controls, as a result of the loss of brain function associated with the stroke. These alterations in muscle networks reflected underlying pathophysiology. These findings can help better understand the motor impairment and motor control in stroke and may advance rehabilitation efforts for stroke by identifying the impaired neuromuscular coordination among multiple muscles in the frequency domain.

This study presents a model-based sensitivity analysis of the strength of voluntary muscle contraction with respect to different patterns of motor unit loss. A motor unit pool model was implemented including simulation of a motor neuron pool, muscle force, and surface electromyogram (EMG) signals. Three different patterns of motor unit loss were simulated, including (1) motor unit loss restricted to the largest ones, (2) motor unit loss restricted to the smallest ones, and (3) motor unit loss without size restriction. The model outputs including muscle force amplitude, variability, and the resultant EMG-force relation were quantified under two different motor neuron firing strategies. It was found that motor unit loss restricted to the largest ones had the most dominant impact on muscle strength and significantly changed the EMG-force relation, while loss restricted to the smallest motor units had a pronounced effect on force variability. These findings provide valuable insight toward our understanding of the neurophysiological mechanisms underlying experimental observations of muscle strength, force control, and EMG-force relation in both normal and pathological conditions.

Background High density surface electromyography (EMG) can be used to estimate muscle innervation zones (IZ). The objective of this study was to compare the differences in the distribution of the biceps brachii (BB) IZ derived from voluntary contractions (VC) and electrical stimulation (ES) of the musculocutaneous nerve. Methods Surface EMG signals were recorded from the medial and lateral BB with two 64-channel high density electrode matrices in eight healthy men. The surface EMG was recorded at different percentages of the maximal voluntary contraction (MVC) force (20–100% MVC) and at different percentages of the current needed to elicit a maximal M-wave (20–100% Imax). The IZs of the medial and lateral BB were identified from the EMG signals and expressed as a row number within a given medial-lateral column. Results ES current intensity had no significant effect on the group mean IZ location (p > 0.05). However, The IZ during VC was located more proximally with increasing force (p 0.05), but was more proximal in the latter than the former during 60–100% MVC (p

With the aim of developing a flexible and reliable procedure for superficial muscle innervation zone (IZ) localization, we proposed a method to estimate IZ location using surface electromyogram (EMG) based on robust linear regression. Regression lines were used to model the bidirectional propagation pattern of a single motor unit action potential (MUAP) and visualize the trajectory of the MUAP propagation. IZ localization was performed by identifying the origin of the bidirectional MUAP propagation. Robust linear regression and MUAP peak detection, combined with propagation phase reversal identification, may provide an efficient way to estimate IZ location. Our method offers high resolution in locating IZs based on simulation studies and experimental tests. Furthermore, our method is flexible and may also be applied using a relatively small number of EMG channels. A comparative study of the proposed method with the cross-correlation method for IZ localization was conducted. The results obtained with simulated MUAPs and measured spontaneous MUAPs in the biceps brachii muscle in six subjects (four males and two females, 57 ± 10 years old) with amyotrophic lateral sclerosis (ALS). Our method achieved estimation performance comparable to that obtained by using the cross-correlation method but with higher resolution. This study provides an accurate and practical method to estimate IZ location.

Objective: Myoelectric pattern recognition has been successfully applied as a human-machine interface to control robotic devices such as prostheses and exoskeletons, significantly improving the dexterity of myoelectric control. This study investigates the feasibility of applying myoelectric pattern recognition for controlling a robotic hand in stroke patients. Methods: Myoelectric pattern recognition of six hand motion patterns was performed using forearm electromyogram signals in paretic side of eight stroke subjects. Both the random cross validation (RCV) and the chronological handout validation (CHV) were applied to assess the offline myoelectric pattern recognition performance. Experiments on real-time myoelectric pattern recognition control of an exoskeleton robotic hand were also performed. Results: An average classification accuracy of 84.1% (the mean value from two different classifiers) and individual subject differences were observed in the offline myoelectric pattern recognition analysis using the RCV, while the accuracy decreased to 65.7% when the CHV was used. The stroke subjects achieved an average accuracy of 61.3 ± 20.9% for controlling the robotic hand. However, our study did not reveal a clear correlation between the real-time control accuracy and the offline myoelectric pattern recognition performance, or any specific characteristics of the stroke subjects. Conclusion: The findings suggest that it is feasible to apply myoelectric pattern recognition to control the robotic hand in some but not all of the stroke patients. Each stroke subject should be individually online tested for the feasibility of applying myoelectric pattern recognition control for robot-assisted rehabilitation.

The objective of this study was to develop sample entropy (SampEn) as a novel surface electromyogram (EMG) biomarker to quantitatively examine post-stroke neuromuscular alternations. The SampEn method was performed on surface EMG interference patterns recorded from biceps brachii muscles of nine healthy control subjects, fourteen subjects with subacute stroke, and eleven subjects with chronic stroke, respectively. Measurements were collected during isometric contractions of elbow flexion at different constant force levels. By producing diagnostic decisions for individual muscles, two categories of abnormalities in some paretic muscles were discriminated in terms of abnormally increased and decreased SampEn. The efficiency of the SampEn was demonstrated by its comparable performance with a previously reported clustering index (CI) method. Mixed SampEn (or CI) patterns were observed in paretic muscles of subjects with stroke indicating complex neuromuscular changes at work as a result of a hemispheric brain lesion. Although both categories of SampEn (or CI) abnormalities were observed in both subacute and chronic stages of stroke, the underlying processes contributing to the SampEn abnormalities might vary a lot in stroke stage. The SampEn abnormalities were also found in contralateral muscles of subjects with chronic stroke indicating the necessity of applying interventions to contralateral muscles during stroke rehabilitation. Our work not only presents a novel method for quantitative examination of neuromuscular changes, but also explains the neuropathological mechanisms of motor impairments and offers guidelines for a better design of effective rehabilitation protocols toward improved motor recovery.

This study presents single-fiber electromyography (EMG) analysis for assessment of paretic muscle changes after stroke. Single-fiber action potentials (SFAPs) were recorded from the first dorsal interosseous (FDI) muscle bilaterally in 12 individuals with hemiparetic stroke. The SFAP parameters, including the negative peak duration and the peak-peak amplitude, were measured and further used to estimate muscle fiber diameter through a model based on the quadratic function. The SFAP parameters, fiber density, and muscle fiber diameter derived from the model were compared between the paretic and contralateral muscles. The results show that SFAPs recorded from the paretic muscle had significantly smaller negative peak duration than that from the contralateral muscle. As a result, the derived muscle fiber diameter of the paretic muscle was significantly smaller than that of the contralateral muscle. The fiber density of the paretic muscle was significantly higher than that of the contralateral muscle. These results provide further evidence of remodeled motor units after stroke and suggest that paretic muscle weakness can be due to both complex central and peripheral neuromuscular alterations.

Estimation of muscle contraction force based on the macroscopic feature of surface electromyography (SEMG) has been widely reported, but the use of microscopic neural drive information has not been thoroughly investigated. In this study, a novel method is proposed to process individual motor unit (MU) activities (firing sequences and action potential waveforms) derived from the decomposition of high density SEMG (HD-SEMG), and it is applied to muscle force estimation. In the proposed method, a supervised machine learning approach was conducted to determine the twitch force of each MU according to its action potential waveforms, which enables separate calculation of every MU’s contribution to force. Thus, the muscle force was predicted through a physiologically meaningful muscle force model. In the experiment, HD-SEMG data were recorded from the abductor pollicis brevis muscles of eight healthy subjects during their performance of thumb abduction with the force increasing gradually from zero to four force levels (10%, 20%, 30%, 40% of the maximal voluntary contraction), while the true muscle force was measured simultaneously. When the proposed method was used, the root mean square difference (RMSD) of the error of the estimated force with respect to the measured force was reported to be 8.3% ± 2.8%. The proposed method also significantly outperformed the other four common methods for force estimation (RMSD: from 11.7% to 20%, ${p} ), demonstrating its effectiveness. This study offers a useful tool for exploiting the neural drive information towards muscle force estimation with improved precision. The proposed method has wide applications in precise motor control, sport and rehabilitation medicine.

This study aimed to examine the redistribution of neuromuscular junctions or innervation zones (IZs) after spinal cord injuries (SCI). Fifteen able-bodied subjects and 15 subjects with SCI (American Spinal Injury Association Impairment Scale A to D), participated in the study. Surface electromyography (EMG) signals were collected from the biceps brachii muscle by a customized linear electrode array when subjects generated maximal isometric voluntary contractions. The Radon transform was applied to detect the IZ locations in the multiple channel surface EMG signals which were differentiated between consecutive channels. The distribution of IZs was compared between the SCI and control groups using the student-t test. Statistical analysis disclosed a significantly wider range of IZs in the SCI group compared with the control group (SCI: 3.83 ± 1.32 IED, control: 2.83 ± 0. 87 IED, IED: inter-electrode distance, p 0.05). No remarkable shifts of the center of the distribution were observed between the two groups (SCI: 9.23 ± 2.35 IED, control: 8.53 ± 2.33 IED, p = 0.42). Changes of IZ distribution in the paralyzed muscles could be associated with the complex neuromuscular reorganization after the SCI.

Objective Using the Twente Medical Systems international B.V. (TMSi) electromyography (EMG) system, active signal shielding was applied to clean signals and create data without interference and cable movement artifacts. TMSi, used in high-density surface EMG pattern recognition, controls myoelectric rehabilitation robots, yet few have studied how active signal shielding influences pattern recognition. This study aimed to investigate how active signal shielding used within the TMSi influenced motion pattern recognition. Methods Surface EMG of dominant side forearm and hand muscles was studied in eight healthy participants. The common component’s influence was accessed by the classification performance of wrist and hand functional movements. Results The classification performance of EMG signals with the common component was obviously lower than signals without the common component using one to five electrodes. Conversely, a higher motion classification performance was achieved using signals with the common component using over 12 electrodes. Optimal channel distribution was examined based on the sequential feed-forward selection method, showing that the common component affected the optimal channel location. Conclusions Active signal shielding in the TMSi improved classification accuracy in motion pattern recognition when over 12 electrodes were used. The optimal channel distribution was related to the common component when using the TMSi.

Background Spatial filtering of multi-channel signals is considered to be an effective pre-processing approach for improving signal-to-noise ratio. The use of spatial filtering for preprocessing high-density (HD) surface electromyogram (sEMG) helps to extract critical spatial information, but its application to non-invasive examination of neuromuscular changes have not been well investigated. Methods Aimed at evaluating how spatial filtering can facilitate examination of muscle paralysis, three different spatial filtering methods are presented using principle component analysis (PCA) algorithm, non-negative matrix factorization (NMF) algorithm, and both combination, respectively. Their performance was evaluated in terms of diagnostic power, through HD-sEMG clustering index (CI) analysis of neuromuscular changes in paralyzed muscles following spinal cord injury (SCI). Results The experimental results showed that: (1) The CI analysis of conventional single-channel sEMG can reveal complex neuromuscular changes in paralyzed muscles following SCI, and its diagnostic power has been confirmed to be characterized by the variance of Z scores; (2) the diagnostic power was highly dependent on the location of sEMG recording channel. Directly averaging the CI diagnostic indicators over channels just reached a medium level of the diagnostic power; (3) the use of either PCA-based or NMF-based filtering method yielded a greater diagnostic power, and their combination could even enhance the diagnostic power significantly. Conclusions This study not only presents an essential preprocessing approach for improving diagnostic power of HD-sEMG, but also helps to develop a standard sEMG preprocessing pipeline, thus promoting its widespread application.

Objective. It is of great value to accurately localize innervation zones (IZs) to better diagnose and treat neuromuscular diseases, but it is challenging to do so noninvasively from surface electromyography (sEMG) recordings because of the blurring/distorting effect of the low conductive fat tissues. This study aimed to develop an innovative transcutaneous IZ imaging (TIZI) technique to precisely and efficiently localize the IZ distribution directly over the muscle surface in vivo from high-density sEMG recordings (HD-sEMG). Approach. The TIZI technique was implemented by incorporating HD-sEMG recording, signal decomposition, finite element analysis and inverse calculation. The performance of TIZI was evaluated on the flexor digitorum superficialis (FDS) muscle with simulated sEMG signal and experimental signal recorded from both healthy (n = 3) and stroke participants (n = 4). The accuracy of imaging was validated by both of the Pearson correlation coefficient (PCC) and localization error (LE) between the TIZI results and the ‘true’ IZ distribution. Main results. In the simulation study, results have shown PCCs of 99.85% ± 0.11%, 99.79% ± 0.08%, 99.63% ± 0.22% and 99.31% ± 0.54% at the depth of 10, 15, 20 and 25 mm and SNR of 25 dB. PCCs of 98.74% ± 1.78% and 97.82% ± 1.20% were respectively obtained for experimental signals acquired from the healthy and spastic FDS muscles. The TIZI provided smaller LEs of 1.4 ± 0.92 mm and 2.02 ± 1.3 mm, compared to LEs of 7.42 ± 2.29 mm and 7.8 ± 1.77 mm from skin observations in healthy and spastic FDS, respectively. Significance. Results have demonstrated the high performance of the proposed TIZI technique by transcutaneous imaging of the IZ distribution of the skeletal muscles. The performance improvement can be attributed to the elimination of the blurring/distorting effect caused by the low conductive fat and high conductive skin tissues. TIZI may provide an advanced neurological tool for the clinical treatment of neuromuscular diseases, such as guiding botulinum neurotoxin injections in spasticity management.

With technical advances in both electrode design and signal processing, electromyography (EMG) has broad applicability for assessing properties of single muscle fibers, motor units, gross muscle, or muscle groups. Different EMG techniques have been applied to assess paretic muscle changes in stroke patients, and a number of abnormalities have been identified. In this chapter, we provide a summary of findings from EMG examination of paretic muscles after stroke, with a view toward better understanding motor unit alterations underlying muscle weakness and other associated changes.

Background Neuromuscular electrical stimulation (NMES) is extensively used in stroke motor rehabilitation. How it promotes motor recovery remains only partially understood. NMES could change muscular properties, produce altered sensory inputs, and modulate fluctuations of cortical activities; but the potential contribution from cortico-muscular couplings during NMES synchronized with dynamic movement has rarely been discussed. Method We investigated cortico-muscular interactions during passive, active, and NMES rhythmic pedaling in healthy subjects and chronic stroke survivors. EEG (128 channels), EMG (4 unilateral lower limb muscles) and movement parameters were measured during 3 sessions of constant-speed pedaling. Sensory-level NMES (20 mA) was applied to the muscles, and cyclic stimulation patterns were synchronized with the EMG during pedaling cycles. Adaptive mixture independent component analysis was utilized to determine the movement-related electro-cortical sources and the source dipole clusters. A directed cortico-muscular coupling analysis was conducted between representative source clusters and the EMGs using generalized partial directed coherence (GPDC). The bidirectional GPDC was compared across muscles and pedaling sessions for post-stroke and healthy subjects. Results Directed cortico-muscular coupling of NMES cycling was more similar to that of active pedaling than to that of passive pedaling for the tested muscles. For healthy subjects, sensory-level NMES could modulate GPDC of both ascending and descending pathways. Whereas for stroke survivors, NMES could modulate GPDC of only the ascending pathways. Conclusions By clarifying how NMES influences neuromuscular control during pedaling in healthy and post-stroke subjects, our results indicate the potential limitation of sensory-level NMES in promoting sensorimotor recovery in chronic stroke survivors.

Microvascular decompression (MVD) is the most effective treatment for hemifacial spasm (HFS). During MVD surgery, abnormal muscle response (AMR) is widely used. Z-L response (ZLR) is a new monitoring method for HFS. We compared the effectiveness of AMR plus ZLR and simple AMR.In a retrospective study of 1868 cases of HFS treated using intraoperative monitoring between January 2013 and December 2015, 896 patients underwent simple AMR monitoring and 972 underwent combined intraoperative monitoring of AMR and ZLR.AMR waveforms were observed in 837 (93.42%) patients in the AMR group and in 898 (92.39%) patients in the AMR plus ZLR group (P0.05). During MVD, AMR waveforms disappeared in 772 patients in the AMR group and 854 patients in the AMR plus ZLR group (P0.05). The efficacy rate of MVD in the AMR plus ZLR group was higher compared with the AMR group when AMR was not detected or disappeared during the operation (P0.05). When AMR persisted during the operation, there was no significant difference between the 2 groups in efficacy of the operation (P0.05).Regardless of whether the compression vessels of the facial nerve are simple or complicated, combined intraoperative monitoring of AMR plus ZLR monitoring provides more valuable neurosurgical guidance than simple AMR during MVD for HFS.

This study examines the electromyogram (EMG)-torque relation for chronic stroke survivors using a novel EMG complexity representation. Ten stroke subjects performed a series of submaximal isometric elbow flexion tasks using their affected and contralateral arms, respectively, while a 20-channel linear electrode array was used to record surface EMG from the biceps brachii muscles. The sample entropy (SampEn) of surface EMG signals was calculated with both global and local tolerance schemes. A regression analysis was performed between SampEn of each channel's surface EMG and elbow flexion torque. It was found that a linear regression can be used to well describe the relation between surface EMG SampEn and the torque. Each channel's root mean square (RMS) amplitude of surface EMG signal in the different torque level was computed to determine the channel with the highest EMG amplitude. The slope of the regression (observed from the channel with the highest EMG amplitude) was smaller on the impaired side than on the nonimpaired side in 8 of the 10 subjects, regardless of the tolerance scheme (global or local) and the range of torques (full or matched range) used for comparison. The surface EMG signals from the channels above the estimated muscle innervation zones demonstrated significantly lower levels of complexity compared with other channels between innervation zones and muscle tendons. The study provides a novel point of view of the EMG-torque relation in the complexity domain, and reveals its alterations post stroke, which are associated with complex neural and muscular changes post stroke. The slope difference between channels with regard to innervation zones also confirms the relevance of electrode position in surface EMG analysis.

Motor activities during daily life always involve simultaneous control of multiple degrees of freedom (DOFs), which has not yet been fully explored in myoelectric control due to difficulty in sufficiently decoding the complex neural control information. This study presents a novel framework for simultaneous myoelectric control based on pattern recognition incorporated with a muscle synergy motor control strategy for each DOF. An experiment for discriminating 18 dexterous finger movement tasks was designed to evaluate the performance of the framework for the simultaneous control of 5 DOFs. Task discrimination was assessed with 18 neurologically intact subjects, and the framework exhibited high accuracy (96.79% ±2.46%), outperforming three other methods, including the routine myoelectric pattern-recognition approach for conventional sequential control ( p

Objective: The aim of this study is to compare the performance of three strategies in determining the global innervation zone (IZ) distribution. Methods: High-density surface electromyography was recorded from the biceps brachii muscle of seven healthy subjects under isometric voluntary contractions at 20%, 50%, and 100% of the maximal voluntary contraction and supramaximal musculocutaneous nerve stimulations. IZs were detected: first, by visual identification in a column-specific manner ( IZ-1D ); second, based on decomposed bipolar mapping of motor unit action potentials ( IZ-2D ); and third, by source imaging in the three-dimensional muscle space ( IZ-3D ). Results: All three IZ detection approaches have exhibited excellent trial-to-trial repeatability. Consistent IZ results were found in the axial direction of the arm across all three approaches, yet a difference was observed in the mediolateral direction. Conclusions: Among all three approaches, IZ-3D is capable of providing the most comprehensive information regarding the global IZ distribution, while maintaining high consistency with IZ-1D and IZ-2D results. Significance: IZ-3D approach can be a potential tool for global IZ imaging, which is critical to the clinical diagnosis and treatment of neuromuscular disorders.

Objective The outcome of botulinum toxin (BTX) therapy of post-stroke spasticity relies largely on accuracy of BTX injection to the proximity of innervation zones (IZs). Recently developed three-dimensional IZ imaging (3DIZI) is the only technique currently available to provide 3D distributions of IZs in vivo, yet its performance has not been validated under pathological conditions. Approach The performance of 3DIZI was evaluated in the spastic biceps brachii muscles of four chronic stroke subjects. High-density surface electromyography (sEMG) and intramuscular electromyography (iEMG) were simultaneously recorded. The IZ location in the 3D space of the spastic biceps calculated using the 3DIZI technique from sEMG recordings were compared against the IZ location detected using intramuscular wires. Main results 3DIZI successfully reconstructed the IZs in the 3D space of the spastic biceps of all four stroke subjects, with a localization error of 4.7 ± 2.7 mm, and specifically a depth error of 1.8 ± 0.4 mm. Significance Results have demonstrated the robust performance of 3DIZI under pathological conditions, laying a solid foundation for clinical application of 3D source imaging in leading precise BTX injections for spasticity management.

OBJECTIVE The objective of this study was to investigate the feasibility of applying myoelectric pattern recognition for controlling a robotic hand in individuals with spinal cord injury (SCI). APPROACH Surface electromyogram (sEMG) signals of six hand motion patterns were recorded from 12 subjects with SCI. Online and offline classification performance of two classifiers (Gaussian Naive Bayes classifier, GNB, and support vector machine, SVM) were investigated. An exoskeleton hand was then controlled in real-time using the classification results. The control accuracy and its correlation with function assessments were investigated. MAIN RESULTS Average offline classification accuracy of all tested SCI subjects was (73.6 ± 14.0)% for GNB and (77.6 ± 11.6)% for SVM, respectively. Average online classification accuracy was significantly lower, (64.3 ± 15.0)% for GNB and (70.2 ± 13.2)% for SVM. Average control accuracy of (81.0 ± 16.3)% was achieved in real-time control of the robotic hand using myoelectric pattern recognition. Correlation between control accuracy and grip/pinch force was observed. SIGNIFICANCE The results show that it is feasible to extract hand motion intent from individuals with SCI and control a robotic hand device using myoelectric pattern recognition. The performance of real-time control can be predicted based on functional assessments.

The progressive FastICA peel-off (PFP) is a recently developed blind source separation approach for high-density surface EMG decomposition. This paper explores a novel application of PFP for automatic decomposition of multi-channel intramuscular electromyogram signals. The automatic PFP (APFP) was used to decompose an open access multichannel intramuscular EMG dataset, simultaneously collected from the brachioradialis muscle using 6 to 8 fine wire or needle electrodes. Given usually a limited number of intramuscular electrodes compared with high-density surface EMG recording, a modification was made to the original APFP framework to dramatically increase the decomposition yield. A total of 131 motor units were automatically decomposed by the APFP framework from 10 multichannel intramuscular EMG signals, among which 128 motor units were also manually identified from the expert interactive EMGLAB decomposition. The average matching rate of discharge instants for all the common motor units was (98.71 ± 1.73)%. The outcomes of this study indicate that the APFP framework can also be used to automatically decompose multichannel intramuscular EMG with high accuracies, even though the number of recording channels is relatively small compared with high-density surface EMG.

Reticulospinal (RS) hyperexcitability is observed in stroke survivors with spastic hemiparesis. Habituated startle acoustic stimuli (SAS) can be used to stimulate the RS pathways non-reflexively. However, the role of RS pathways in motor function and its interactions with the corticospinal system after stroke still remain unclear. Therefore, the purpose of this study was to investigate the effects of conditioning SAS on the corticospinal system in healthy subjects and in stroke subjects with spastic hemiparesis. An established conditioning SAS- transcranial magnetic stimulation (TMS) paradigm was used to test the interactions between the RS pathways and the corticospinal system. TMS was delivered to the right hemisphere of eleven healthy subjects and the contralesional hemisphere of eleven stroke subjects during isometric elbow flexor contraction on the non-impaired (or left) side. Conditioning SAS had similar effects on the corticospinal motor system in both healthy and stroke subjects, including similar SAS-induced motor evoked potential (MEP) reduction at rest, but not during voluntary contraction tasks; similar magnitudes of TMS-induced MEP and force increment and shortening of the silent period during voluntary elbow flexor contraction. This study provides evidence that RS excitability on the contralesional side in stroke subjects with spastic hemiparesis is not abnormal, and suggests that RS projections are likely to be primarily unilateral in humans.

This study presents a progressive FastICA peel-off (PFP) framework for high density surface electromyogram (EMG) decomposition. The novel framework is based on a shift-invariant model for describing surface EMG. The decomposition process can be viewed as progressively expanding the set of motor unit spike trains, which is primarily based on FastICA. To overcome the local convergence of FastICA, a “peel off” strategy (i.e. removal of the estimated motor unit action potential (MUAP) trains from the previous step) is used to mitigate the effects of the already identified motor units, so more motor units can be extracted. Moreover, a constrained FastICA is applied to assess the extracted spike trains and correct possible erroneous or missed spikes. These procedures work together to improve the decomposition performance. The proposed framework was validated using simulated surface EMG signals with different motor unit numbers (30, 70, 91) and signal to noise ratios (SNRs) (20, 10, 0 dB). The results demonstrated relatively large numbers of extracted motor units and high accuracies (high F1-scores). The framework was also tested with 111 trials of 64-channel electrode array experimental surface EMG signals during the first dorsal interosseous (FDI) muscle contraction at different intensities. On average 14.1 ± 5.0 motor units were identified from each trial of experimental surface EMG signals.

A compound muscle action potential (CMAP) scan has different settings such as stimulus frequency, duration (or pulse width), and the number of stimuli (or steps). This study aims to evaluate the influence of different stimulation protocols on MScanFit, a recently developed approach to motor unit number estimation (MUNE) from CMAP scan.CMAP scans of the first dorsal interosseous (FDI) muscle were performed using four protocols with different pulse widths (0.1 ms, 0.2 ms) and steps (500, 1000) in twelve neurologically intact subjects. For each CMAP scan, the MUNE was derived using MScanFit.Across all subjects, a significantly higher MUNE was obtained using stimulus pulse width of 0.1 ms (500 steps: 107. 7 ± 21.3; 1000 steps: 94.9 ± 22.07) than that using pulse width of 0.2 ms (500 steps: 81.8 ± 9.9; 1000 steps: 77.8 ± 16.1) (p 0.001). However, no significant difference in MUNE was observed using 500 and 1000 steps (p 0.1). No cross effect of pulse width and steps was found between four different protocols (p 0.1).Given these results, a stimulation protocol of 0.1 ms pulse width and 500 steps is recommended for CMAP scan recording of the FDI muscle toward estimating motor unit number using MScanFit.

INTRODUCTION: This study performed motor unit number estimation (MUNE) of the abductor hallucis (AH) muscle from 16 healthy control subjects based on the compound muscle action potential (CMAP) scan. METHODS: Muscle responses to electrical stimuli ranging from sub-threshold to supramaximal intensity were recorded and MUNE was determined from a model of the responses (MscanFit program). RESULTS: The average CMAP amplitude and MUNE of the AH for the right and left sides combined were 19.6 ± 0.75 mV and 127 ± 5 (mean ± standard error), respectively. DISCUSSION: Findings of the study provide useful information about the motor unit number of the AH.

This paper presents a novel two-source approach for validating the performance of high-density surface electromyogram (EMG) decomposition. The approach was developed taking advantage of surface EMG characteristics of amyotrophic lateral sclerosis (ALS). High-density surface EMG data from ALS patients can be divided to the sparse data set and the interference data set, with the former decomposed by expert visual inspection while the latter independently decomposed by the surface EMG decomposition algorithm. The agreement of the decomposition yields from the two data sets can be quantified for evaluating the surface EMG decomposition performance. The novel validation approach was performed for a recently developed method called automatic progressive FastICA peel-off (APFP) for high-density surface EMG decomposition. The APFP framework was used to automatically decompose high-density surface EMG signals recorded from the first dorsal interosseous muscle of ALS subjects. The common motor units independently decomposed from the interference data set and the sparse data set demonstrated an average matching rate of 99.18% ± 1.11%. The characteristics of the ALS surface EMG also facilitate a step by step illustration of the APFP framework for high-density surface EMG decomposition. The novel approach presented in this paper can supplement conventional two-source validation for accuracy assessment of decomposed motor units from experimental signals, which is essential for development of surface EMG decomposition methods.

This study presents automatic decomposition of high density surface electromyogram (EMG) signals through a progressive FastICA peel-off (PFP) framework. By incorporating FastICA, constrained FastICA and a peel-off strategy, the PFP can progressively expand the set of motor unit spike trains contributing to the EMG signal. A series of signal processing techniques were applied and integrated in this study to automatically implement the two tasks that often require human operator interaction during application of the PFP framework, including extraction of motor unit spike trains from FastICA outputs and reliability judgment of the extracted motor units. Based on these advances, an automatic PFP (APFP) framework was consequently developed. The decomposition performance of APFP was validated using simulated high density surface EMG signals. The APFP was also evaluated with experimental surface EMG signals, and the decomposition results were comparable to those achieved from the PFP with human operator interaction.

This study investigated the effect of handedness on motor unit number index (MUNIX). Maximal hand strength, compound muscle action potential (CMAP), and voluntary surface electromyography (EMG) signals were measured bilaterally for the first dorsal interosseous (FDI) and thenar muscles in 24 right-handed and 2 left handed healthy subjects. Mean (± standard error) grip and pinch forces in the dominant hand were 43.99 ± 2.36 kg and 9.36 ± 0.52 kg respectively, significantly larger than those in the non-dominant hand (grip: 41.37 ± 2.29 kg, p

The control of multifunctional myoelectric prostheses is commonly limited due to the lack of sufficient electromyography (EMG) signals after amputation. With a goal of developing easy-to-use and flexibly controlled multifunctional prostheses, the authors propose two control strategies that fuse EMG and speech signals. In the first, speech signals switch the joint-mode, and EMG signals determine a motion class to actuate the target movement. In the second, speech signals select a motion class directly, and EMG signals actuated the movement. Experimental results showed that the proposed strategies could improve control performance and enhance the operational efficiency significantly, suggesting that signal fusion is a feasible way to effectively strengthen interactions between humans and machines.

BackgroundFasciculation potentials (FPs) are important in supporting the electrodiagnosis of Amyotrophic Lateral Sclerosis (ALS). If classified by shape, FPs can also be very informative for laboratory-based neurophysiological investigations of the motor units. MethodsThis study describes a Matlab program for classification of FPs recorded by multi-channel surface electromyogram (EMG) electrodes. The program applies Principal Component Analysis on a set of features recorded from all channels. Then, it registers unsupervised and supervised classification algorithms to sort the FP samples. Qualitative and quantitative evaluation of the results is provided for the operator to assess the outcome. The algorithm facilitates manual interactive modification of the results. Classification accuracy can be improved progressively until the user is satisfied. The program makes no assumptions regarding the occurrence times of the action potentials, in keeping with the rather sporadic and irregular nature of FP firings. ResultsTen sets of experimental data recorded from subjects with ALS using a 20-channel surface electrode array were tested. A total of 11891 FPs were detected and classified into a total of 235 prototype template waveforms. Evaluation and correction of classification outcome of such a dataset with over 6000 FPs can be achieved within 1-2 days. Facilitated interactive evaluation and modification could expedite the process of gaining accurate final results. ConclusionThe developed Matlab program is an efficient toolbox for classification of FPs. Distribution of 'amplitude' and 'area' can be used to discriminate waveforms recorded by HDsEMG.Interactive modification module increases the accuracy of classification and is efficient time-wise.The spike sorting output, tested by two-source method, is reproducible.In each subject, only 1-2 classes were continuingly firing, albeit numerous classes had been derived.The final result of this program will further be subject to many other research studies.

In this study, we used muscle and motor unit indices, derived from convenient surface electromyography (EMG) measurements, for examination of paretic muscle changes post stroke. For 12 stroke subjects, compound muscle action potential and voluntary surface EMG signals were recorded from paretic and contralateral first dorsal interosseous, abductor pollicis brevis, and abductor digiti minimi muscles. Muscle activation index (AI), motor unit number index (MUNIX), and motor unit size index (MUSIX) were then calculated for each muscle. There was a significant AI reduction for all the three muscles in paretic side compared with contralateral side, providing an evidence of muscle activation deficiency after stroke. The hand MUNIX (defined by summing the values from the three muscles) was significantly reduced in paretic side compared with contralateral side, whereas the hand MUSIX was not significantly different. Furthermore, diverse changes in MUNIX and MUSIX were observed from the three muscles. A major feature of the present examinations is the primary reliance on surface EMG, which offers practical benefits because it is noninvasive, induces minimal discomfort and can be performed quickly.

Electrical impedance myography (EIM) was used to assess the paretic muscle intrinsic electrical properties post stroke. Twenty-seven subjects with chronic hemiparesis participated in this study. Muscle impedance was measured by applying high-frequency, low-intensity alternating current to biceps brachii muscles. Major EIM parameters, resistance ( $R$ ), reactance ( $X$ ), phase angle ( $\theta $ ), and electrical anisotropy ratios (AR) of the three parameters, were examined at 50 kHz. Statistical analysis demonstrated significant reduction of reactance, phase angle, AR of resistance, and AR of reactance in the paretic muscle compared with the contralateral side (Paretic X: $6.16~\pm ~0.55~\Omega $ , contralateral X: $7.62~\pm ~0.46~\Omega $ , and p $\theta $ : $10.97~\pm ~0.84^{\circ }$ , contralateral $\theta $ : 14.5 ± 0.82°, and p < 0.001; Paretic AR of R: 0.969 ± 0.013, contralateral AR of R: 1.008 ± 0.011, and p < 0.02; and Paretic AR of X: 0.981 ± 0.066, contralateral AR of X: 1.114 ± 0.041, and p < 0.02). Correlation analysis, however, did not show any significant relationship between EIM parameters and clinical assessments. Findings of this paper indicated significant changes in the muscular intrinsic electrical properties after stroke, possibly related to structural modifications induced by loss of muscle fibers or fat infiltration as well as changes in the quality of cell membranes post stroke.

The objective of this paperwas to characterize complex neuromuscular changes induced by a hemisphere stroke through a novel clustering index (CI) analysis of surface electromyogram (EMG). The CI analysis was performed using surface EMG signals collected bilaterally from the thenar muscles of 17 subjects with stroke and 12 age-matched healthy controls during their performance of varying levels of isometric muscle contractions. Compared with the neurologically intact or contralateral muscles, mixed CI patterns were observed in the paretic muscles. Two paretic muscles showed significantly increased CI implying dominant neurogenic changes, whereas three paretic muscles had significantly reduced CI indicating dominantmyopathic changes; the other paretic muscles did not demonstrate a significant CI alternation, likely due to a deficit of descending central drive or a combined effect of neuromuscular factors. Such discrimination of paretic muscles was further highlighted using a modified CI method that emphasizes between-side comparison for each individual subject. The CI findings suggest that there appears to be different central and peripheral processes at work in varying degrees after stroke. This paper provides a convenient and quantitative analysis to assess the nature of neuromuscular changes after stroke, without using any special equipment but conventional surface EMG recording. Such assessment is helpful for the development of appropriate rehabilitation strategies for recovery of motor function.

The three dimensional (3D) innervation zone (IZ) imaging approach (3DIZI) has been developed in our group to localize the IZ of a particular motor unit (MU) from its motor unit action potentials decomposed from high-density surface electromyography (EMG) recordings. In this study, the developed 3DIZI approach was combined with electrical stimulation to investigate global distributions of IZs in muscles from electrically elicited M-wave recordings. Electrical stimulations were applied to the musculocutaneous nerve to activate supramaximal muscle response of the biceps brachii in one healthy subject, and high-density (128 channels) surface EMG signals of the biceps brachii muscles were recorded. The 3DIZI approach was then employed to image the IZ distribution of IZs in the 3D space of the biceps brachii. The performance of the M-wave based 3DIZI approach was evaluated with different stimulation intensities. Results show that the reconstructed IZs under supramaximal stimulation are spatially distributed in the center region of muscle belly which is consistent with previous studies. With sub-maximal stimulation intensity, the imaged IZ centers became more proximally and deeply located. The proposed M-wave based 3DIZI approach demonstrated its capability of imaging global distribution of IZs in muscles, which provide valuable information for clinical applications such as guiding botulinum toxin injection in treating muscle spasticity.

A hand exoskeleton driven by myoelectric pattern recognition was designed for stroke rehabilitation. It detects and recognizes the user’s motion intent based on electromyography (EMG) signals, and then helps the user to accomplish hand motions in real time. The hand exoskeleton can perform six kinds of motions, including the whole hand closing/opening, tripod pinch/opening, and the “gun” sign/opening. A 52-year-old woman, 8 months after stroke, made 20×2-hour visits over 10 weeks to participate in robot-assisted hand training. Though she was unable to move her fingers on her right hand before the training, EMG activities could be detected on her right forearm. In each visit, she took 4×10-minute robot-assisted training sessions, in which she repeated the aforementioned six motion patterns assisted by our intent-driven hand exoskeleton. After the training, her grip force increased from 1.5 kg to 2.7 kg, her pinch force increased from 1.5 kg to 2.5 kg, her score of Box & Block test increased from 3 to 7, her score of Fugl-Meyer (Part C) increased from 0 to 7, her hand function increased from Stage 1 to Stage 2 in Chedoke-McMaster assessment. The results demonstrate the feasibility of robot-assisted training driven by myoelectric pattern recognition after stroke.