Two-Source Validation of Progressive FastICA Peel-Off for Automatic Surface EMG Decomposition in Human First Dorsal Interosseous Muscle

2CFastICA: A Novel Method for High Density Surface EMG Decomposition Based on Kernel Constrained FastICA and Correlation Constrained FastICA

This study presents a novel high density surface electromyography (EMG) decomposition method, named as 2CFastICA, because it incorporates two key algorithms: kernel constrained FastICA and correlation constrained FastICA. The former focuses on overcoming the local convergence of FastICA without requiring the peel-off strategy used in the progressive FastICA peel-off (PFP) framework. The latter further refines the output of kernel constrained FastICA by correcting possible erroneous or missed spikes. The two constrained FastICA algorithms supplement each other to warrant the decomposition performance. The 2CFastICA method was validated using simulated surface EMG signals with different motor unit numbers and signal to noise ratios (SNRs). Two source validation was also performed by simultaneous high density surface EMG and intramuscular EMG recordings, showing a matching rate (MR) of (97.2 ± 3.5)% for 170 common motor units. In addition, a different form of two source validation was also conducted taking advantages of the high density surface EMG characteristics of patients with amyotrophic lateral sclerosis, showing a MR of (99.4 ± 0.9)% for 34 common motor units from interference and sparse datasets. Both simulation and experimental results indicate that 2CFastICA can achieve similar decomposition performance to PFP. However, the efficiency of decomposition can be greatly improved by 2CFastICA since the complex signal processing procedures associated with the peel-off strategy are not required any more. Along with this paper, we also provide the MATLAB open source code of 2CFastICA for high density surface EMG decomposition.

Unsupervised neural decoding for concurrent and continuous multi-finger force prediction

Reliable prediction of multi-finger forces is crucial for neural-machine interfaces. Various neural decoding methods have progressed substantially for accurate motor output predictions. However, most neural decoding methods are performed in a supervised manner, i.e., the finger forces are needed for model training, which may not be suitable in certain contexts, especially in scenarios involving individuals with an arm amputation. To address this issue, we developed an unsupervised neural decoding approach to predict multi-finger forces using spinal motoneuron firing information. We acquired high-density surface electromyogram (sEMG) signals of the finger extensor muscle when subjects performed single-finger and multi-finger tasks of isometric extensions. We first extracted motor units (MUs) from sEMG signals of the single-finger tasks. Because of inevitable finger muscle co-activation, MUs controlling the non-targeted fingers can also be recruited. To ensure an accurate finger force prediction, these MUs need to be teased out. To this end, we clustered the decomposed MUs based on inter-MU distances measured by the dynamic time warping technique, and we then labeled the MUs using the mean firing rate or the firing rate phase amplitude. We merged the clustered MUs related to the same target finger and assigned weights based on the consistency of the MUs being retained. As a result, compared with the supervised neural decoding approach and the conventional sEMG amplitude approach, our new approach can achieve a higher R2 (0.77 ± 0.036 vs. 0.71 ± 0.11 vs. 0.61 ± 0.09) and a lower root mean square error (5.16 ± 0.58 %MVC vs. 5.88 ± 1.34 %MVC vs. 7.56 ± 1.60 %MVC). Our findings can pave the way for the development of accurate and robust neural-machine interfaces, which can significantly enhance the experience during human-robotic hand interactions in diverse contexts.

Adaptive HD-sEMG decomposition: towards robust real-time decoding of neural drive

Objective. Neural interfacing via decomposition of high-density surface electromyography (HD-sEMG) should be robust to signal non-stationarities incurred by changes in joint pose and contraction intensity.Approach. We present an adaptive real-time motor unit decoding algorithm and test it on HD-sEMG collected from the extensor carpi radialis brevis during isometric contractions over a range of wrist angles and contraction intensities. The performance of the algorithm was verified using high-confidence benchmark decompositions derived from concurrently recorded intramuscular electromyography.Main results. In trials where contraction conditions between the initialization and testing data differed, the adaptive decoding algorithm maintained significantly higher decoding accuracies when compared to static decoding methods.Significance. Using "gold standard" verification techniques, we demonstrate the limitations of filter re-use decoding methods and show the necessity of parameter adaptation to achieve robust neural decoding.

sEMG-Based Inter-Session Hand Gesture Recognition via Domain Adaptation with Locality Preserving and Maximum Margin

Surface electromyography (sEMG)-based gesture recognition can achieve high intra-session performance. However, the inter-session performance of gesture recognition decreases sharply due to the shift in data distribution. Therefore, developing a robust model to minimize the data distribution difference is crucial to improving the user experience. In this work, based on the inter-session gesture recognition task, we propose a novel algorithm called locality preserving and maximum margin criterion (LPMM). The LPMM algorithm integrates three main modules, including domain alignment, pseudo-label selection, and iteration result selection. Domain alignment is designed to preserve the neighborhood structure of the feature and minimize the overlap of different classes. The pseudo-label selection and iteration result selection can avoid the decrease in accuracy caused by mislabeled samples. The proposed algorithm was evaluated on two of the most widely used EMG databases. It achieves a mean accuracy of 98.46% and 71.64%, respectively, which is superior to state-of-the-art domain adaptation methods.

Alterations of Motor Unit Characteristics Associated With Muscle Fatigue

This study aims to characterize motor unit (MU) features associated with muscle fatigue, using high-density surface electromyography (HD-sEMG). The same MUs recruited before/after, and during muscle fatigue were identified for analysis. The surface location of the innervation zones (IZs) of the MUs was identified from the HD-sEMG bipolar motor unit action potential (MUAP) map. The depth of the MU was also identified from the decay pattern of the MUAP along the muscle fiber transverse direction. Both the surface IZ location and the MU depth information were utilized to ensure the same MU was examined during the contraction before/after muscle fatigue. The MUAP similarity, defined as the correlation coefficient between MUAP morphology, was adopted to reveal the alterations in MU characteristics under the condition of fatigue. The biomarkers of the same MUs were compared before/after fatigue (task 1) at 5%, 10%, and 15% maximal voluntary contraction (MVC) and in the process of continuous fatigue (task 2) at 20% MVC. Our results indicate that the MUAP morphology similarity of the same MUs was 0.91 ± 0.06 (task 1) and 0.93 ± 0.04 (task 2). The results showed that MUAP morphology maintained good stability before/after, and during muscle fatigue. The findings of this study may advance our understanding of the mechanism of MU neuromuscular fatigue.

MUNIX repeatability evaluation method based on FastICA demixing

To enhance the reproducibility of motor unit number index (MUNIX) for evaluating neurological disease progression, this paper proposes a negative entropy-based fast independent component analysis (FastICA) demixing method to assess MUNIX reproducibility in the presence of inter-channel mixing of electromyography (EMG) signals acquired by high-density electrodes. First, composite surface EMG (sEMG) signals were obtained using high-density surface electrodes. Second, the FastICA algorithm based on negative entropy was employed to determine the orthogonal projection matrix that minimizes the negative entropy of the projected signal and effectively separates mixed sEMG signals. Finally, the proposed experimental approach was validated by introducing an interrelationship criterion to quantify independence between adjacent channel EMG signals, measuring MUNIX repeatability using coefficient of variation (CV), and determining motor unit number and size through MUNIX. Results analysis shows that the inclusion of the full (128) channel sEMG information leads to a reduction in CV value by $1.5 \pm 0.1$ and a linear decline in CV value with an increase in the number of channels. The correlation between adjacent channels in participants decreases by $0.12 \pm 0.05$ as the number of channels gradually increases. The results demonstrate a significant reduction in the number of interrelationships between sEMG signals following negative entropy-based FastICA processing, compared to the mixed sEMG signals. Moreover, this decrease in interrelationships becomes more pronounced with an increasing number of channels. Additionally, the CV of MUNIX gradually decreases with an increase in the number of channels, thereby optimizing the issue of abnormal MUNIX repeatability patterns and further enhancing the reproducibility of MUNIX based on high-density surface EMG signals.

Bio-robotics research for non-invasive myoelectric neural interfaces for upper-limb prosthetic control: a 10-year perspective review

A decade ago, a group of researchers from academia and industry identified a dichotomy between the industrial and academic state-of-the-art in upper-limb prosthesis control, a widely used bio-robotics application. They proposed that four key technical challenges, if addressed, could bridge this gap and translate academic research into clinically and commercially viable products. These challenges are unintuitive control schemes, lack of sensory feedback, poor robustness and single sensor modality. Here, we provide a perspective review on the research effort that occurred in the last decade, aiming at addressing these challenges. In addition, we discuss three research areas essential to the recent development in upper-limb prosthetic control research but were not envisioned in the review 10 years ago: deep learning methods, surface electromyogram decomposition and open-source databases. To conclude the review, we provide an outlook into the near future of the research and development in upper-limb prosthetic control and beyond.

Toward a generalizable deep CNN for neural drive estimation across muscles and participants

Objective.High-density electromyography (HD-EMG) decomposition algorithms are used to identify individual motor unit (MU) spike trains, which collectively constitute the neural code of movements, to predict motor intent. This approach has advanced from offline to online decomposition, from isometric to dynamic contractions, leading to a wide range of neural-machine interface applications. However, current online methods need offline retraining when applied to the same muscle on a different day or to a different person, which limits their applications in a real-time neural-machine interface. We proposed a deep convolutional neural network (CNN) framework for neural drive estimation, which takes in frames of HD-EMG signals as input, extracts general spatiotemporal properties of MU action potentials, and outputs the number of spikes in each frame. The deep CNN can generalize its application without retraining to HD-EMG data recorded in separate sessions, muscles, or participants.Approach.We recorded HD-EMG signals from the vastus medialis and vastus lateralis muscles from five participants while they performed isometric contractions during two sessions separated by ∼20 months. We identified MU spike trains from HD-EMG signals using a convolutive blind source separation (BSS) method, and then used the cumulative spike train (CST) of these MUs and the HD-EMG signals to train and validate the deep CNN.Main results.On average, the correlation coefficients between CST from the BSS and that from deep CNN were0.983±0.006for leave-one-out across-sessions-and-muscles validation and0.989±0.002for leave-one-out across-participants validation. When trained with more than four datasets, the performance of deep CNN saturated at0.984±0.001for cross validations across muscles, sessions, and participants.Significance.We can conclude that the deep CNN is generalizable across the aforementioned conditions without retraining. We could potentially generate a robust deep CNN to estimate neural drive to muscles for neural-machine interfaces.

Compressed spike-triggered averaging in iterative decomposition of surface EMG

BACKGROUND AND OBJECTIVE

Analysis of motor unit activity is important for assessing and treating diseases or injuries affecting natural movement. State-of-the-art decomposition translates high-density surface electromyography (HDsEMG) into motor unit activity. However, current decomposition methods offer far from complete separation of all motor units.

METHODS

This paper proposes a peel-off approach to automatic decomposition of HDsEMG into motor unit action potential (MUAP) trains, based on the Fast Independent Component Analysis algorithm (FastICA). The novel steps include utilizing compression by means of Principal Component Analysis and spike-triggered averaging, to estimate surface MUAP distributions with less noise, which are iteratively subtracted from the HDsEMG dataset. Furthermore, motor unit spike trains are estimated by high-dimensional density-based clustering of peaks in the FastICA source output. And finally, a new reliability measure is used to discard poor motor unit estimates by comparing the variance of the FastICA source output before and after the peel-off step. The method was validated using reconstructed synthetic data at three different signal-to-noise levels and was compared to an established deflationary FastICA approach.

RESULTS

Both algorithms had very high recall and precision, over 90%, for spikes from matching motor units, referred to as matched performance. However, the peel-off algorithm correctly identified more motor units for all noise levels. When accounting for unidentified motor units, total recall was up to 33 percentage points higher; and when accounting for duplicate estimates, total precision was up to 24 percentage points higher, compared to the state-of-the-art reference. In addition, a comparison was done using experimental data where the proposed algorithm had a matched recall of 97% and precision of 85% with respect to the reference algorithm.

CONCLUSION

These results show a substantial performance increase for decomposition of simulated HDsEMG data and serve to validate the proposed approach. This performance increase is an important step towards complete decomposition and extraction of information of motor unit activity.

Non-invasive analysis of motor unit activation during simultaneous and continuous wrist movements

Surface electromyography (EMG) signals have shown promising applications in human-machine interfacing (HMI) systems such as orthotics, prosthetics, and exoskeletons. Nevertheless, existing myoelectric control methods, generally based on time-domain or frequency-domain features, could not directly interpret neural commands. EMG decomposition techniques have become a prevailing solution to decode the motor neuron discharges from the spinal cord, whereas only single degree-of-freedom (DoF) movements are primarily involved in the current neural-based interfaces, resulting in limited intuitiveness and functionality. Here, we propose a non-invasive framework to analyze motor unit activities and estimate wrist torques during simultaneous contractions of multiple DoFs. Motor unit discharges were decoded from surface EMG signals and pooled into groups during sequential wrist movements. Then three neural features were extracted and linearly projected to the torques of multi-DoF tasks. On average, there were 4413 motor units identified for each motion with a PNR value of 25.82.9 dB. The neural features outperformed the classic EMG feature on the estimation accuracy with higher correlation coefficients and smoothness. These results demonstrate the feasibility and superiority of the proposed framework in kinetics estimation of simultaneous movements, extending the potential applications of surface EMG decomposition in human-machine interfaces.

A hybrid method for real-time stimulation artefact removal during functional electrical stimulation with time-variant parameters

Objective. In this study, a hybrid method combining hardware and software architecture is proposed to remove stimulation artefacts (SAs) and extract the volitional surface electromyography (sEMG) in real time during functional electrical stimulations (FES) with time-variant parameters.Approach. First, an sEMG detection front-end (DFE) combining fast recovery, detector and stimulator isolation and blanking is developed and is capable of preventing DFE saturation with a blanking time of 7.6 ms. The fragment between the present stimulus and previous stimulus is set as an SA fragment. Second, an SA database is established to provide six high-similarity templates with the current SA fragment. The SA fragment will be de-artefacted by a 6th-order Gram-Schmidt (GS) algorithm, a template-subtracting method, using the provided templates, and this database-based GS algorithm is called DBGS. The provided templates are previously collected SA fragments with the same or a similar evoking FES intensity to that of the current SA fragment, and the lengths of the templates are longer than that of the current SA fragment. After denoising, the sEMG will be extracted, and the current SA fragment will be added to the SA database. The prototype system based on DBGS was tested on eight able-bodied volunteers and three individuals with stroke to verify its capacity for stimulation removal and sEMG extraction.Results.The average stimulus artefact attenuation factor, SA index and correlation coefficient between clean sEMG and extracted sEMG for 6th-order DBGS were 12.77 ± 0.85 dB, 1.82 ± 0.37 dB and 0.84 ± 0.33 dB, respectively, which were significantly higher than those for empirical mode decomposition combined with notch filters, pulse-triggered GS algorithm, 1st-order and 3rd-order DBGS. The sEMG-torque correlation coefficients were 0.78 ± 0.05 and 0.48 ± 0.11 for able-bodied volunteers and individuals with stroke, respectively.Significance.The proposed hybrid method can extract sEMG during dynamic FES in real time.

Concurrent Prediction of Finger Forces Based on Source Separation and Classification of Neuron Discharge Information

A reliable neural-machine interface is essential for humans to intuitively interact with advanced robotic hands in an unconstrained environment. Existing neural decoding approaches utilize either discrete hand gesture-based pattern recognition or continuous force decoding with one finger at a time. We developed a neural decoding technique that allowed continuous and concurrent prediction of forces of different fingers based on spinal motoneuron firing information. High-density skin-surface electromyogram (HD-EMG) signals of finger extensor muscle were recorded, while human participants produced isometric flexion forces in a dexterous manner (i.e. produced varying forces using either a single finger or multiple fingers concurrently). Motoneuron firing information was extracted from the EMG signals using a blind source separation technique, and each identified neuron was further classified to be associated with a given finger. The forces of individual fingers were then predicted concurrently by utilizing the corresponding motoneuron pool firing frequency of individual fingers. Compared with conventional approaches, our technique led to better prediction performances, i.e. a higher correlation ([Formula: see text] versus [Formula: see text]), a lower prediction error ([Formula: see text]% MVC versus [Formula: see text]% MVC), and a higher accuracy in finger state (rest/active) prediction ([Formula: see text]% versus [Formula: see text]%). Our decoding method demonstrated the possibility of classifying motoneurons for different fingers, which significantly alleviated the cross-talk issue of EMG recordings from neighboring hand muscles, and allowed the decoding of finger forces individually and concurrently. The outcomes offered a robust neural-machine interface that could allow users to intuitively control robotic hands in a dexterous manner.

Muscle Force Estimation Based on Neural Drive Information From Individual Motor Units

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%, ), 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.

Impaired Firing Behavior of Individually Tracked Paretic Motor Units During Fatiguing Contractions of the Dorsiflexors and Functional Implications Post Stroke

Introduction: This study quantified stroke-related changes in the following: (1) the averaged discharge rate of motor units (individually tracked and untracked) identified from high-density electromyography (HD-EMG) recordings, (2) global muscle EMG properties of the dorsiflexors during a fatiguing contraction, and the relationship between task endurance and measures of leg function.

Methods: Ten individuals with chronic stroke performed a sustained sub-maximal, isometric, fatiguing dorsiflexion contraction in paretic and non-paretic legs. Motor-unit firing behavior, task duration, maximal voluntary contraction strength (MVC), and clinical measures of leg function were obtained.

Results: Compared to the non-paretic leg, the paretic leg task duration was shorter, and there was a larger exercise-related reduction in motor unit global rates, individually tracked discharge rates, and overall magnitude of EMG. Task duration of the paretic leg was more predictive of walking speed and lower extremity Fugl-Meyer scores compared to the non-paretic leg.

Discussion: Paretic leg muscle fatigability is increased post stroke. It is characterized by impaired rate coding and recruitment and relates to measures of motor function.

A Method for Suppressing Electrical Stimulation Artifacts from Electromyography

When surface electromyography (EMG) signal is used in a real-time functional electrical stimulation (FES) system for feedback control, the artifact from electrical stimulation is a key challenge for EMG signal processing. To address this challenge, this study proposes a novel method to suppress stimulation artifacts in the EMG-driven closed-loop FES system. The proposed method is inspired by an experimental study that compares artifacts generated by electrical stimulations with different current intensities. It is found that (1) spikes of stimulation artifacts are susceptible to the current intensity and (2) tailing components are similar under different current intensities. Based on these observations, the proposed method combines the blanking and template subtracting strategies for suppressing stimulation artifact. The length of blanking window for suppressing the stimulation spike is adaptively determined by a spike detection algorithm and the first-order derivative analysis of signal. An autoregressive model is used to estimate the tailing part of stimulation artifact, which is an adaptive template for subtracting the artifact. The proposed method is evaluated on both semi-synthetic and experimental datasets. Verified on the semi-synthetic dataset, the proposed method achieves better performance than the classic blanking method. Validated on the experimental dataset, the proposed method substantially decreases the power of stimulation artifact in the EMG. These results indicate that the proposed method can effectively suppress the stimulation artifact while retains the useful EMG signal for an EMG-driven FES system.

Global Innervation Zone Identification With High-Density Surface Electromyography

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.

The Relationship Between Blood Flow and Motor Unit Firing Rates in Response to Fatiguing Exercise Post-stroke

We quantified the relationship between the change in post-contraction blood flow with motor unit firing rates and metrics of fatigue during intermittent, sub-maximal fatiguing contractions of the knee extensor muscles after stroke. Ten chronic stroke survivors (>1-year post-stroke) and nine controls participated. Throughout fatiguing contractions, the discharge timings of individual motor units were identified by decomposition of high-density surface EMG signals. After five consecutive contractions, a blood flow measurement through the femoral artery was obtained using an ultrasound machine and probe designed for vascular measurements. There was a greater increase of motor unit firing rates from the beginning of the fatigue protocol to the end of the fatigue protocol for the control group compared to the stroke group (14.97 ± 3.78% vs. 1.99 ± 11.90%, p = 0.023). While blood flow increased with fatigue for both groups (p = 0.003), the magnitude of post-contraction blood flow was significantly greater for the control group compared to the stroke group (p = 0.004). We found that despite the lower magnitude of muscle perfusion through the femoral artery in the stroke group, blood flow has a greater impact on peripheral fatigue for the control group; however, we observed a significant correlation between change in blood flow and motor unit firing rate modulation (r² = 0.654, p = 0.004) during fatigue in the stroke group and not the control group (r² = 0.024, p < 0.768). Taken together, this data showed a disruption between motor unit firing rates and post-contraction blood flow in the stroke group, suggesting that there may be a disruption to common inputs to both the reticular system and the corticospinal tract. This study provides novel insights in the relationship between the hyperemic response to exercise and motor unit firing behavior for post-stroke force production and may provide new approaches for recovery by improving both blood flow and muscle activation simultaneously.

Automatic Multichannel Intramuscular Electromyogram Decomposition: Progressive FastICA Peel-Off and Performance Validation

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.