Surface Electromyography Signals Research Articles

ResNet1D-Based Personal Identification with Multi-Session Surface Electromyography for Electronic Health Record Integration

Personal identification is an important aspect of managing electronic health records (EHRs), ensuring secure access to patient information, and maintaining patient privacy. Traditionally, biometric, signature, username/password, photo identity, etc., are employed for user authentication. However, these methods can be prone to security breaches, identity theft, and user inconvenience. The security of personal information is of paramount importance, particularly in the context of EHR. To address this, our study leverages ResNet1D, a deep learning architecture, to analyze surface electromyography (sEMG) signals for robust identification purposes. The proposed ResNet1D-based personal identification approach using the sEMG signal can offer an alternative and potentially more secure method for personal identification in EHR systems. We collected a multi-session sEMG signal database from individuals, focusing on hand gestures. The ResNet1D model was trained using this database to learn discriminative features for both gesture and personal identification tasks. For personal identification, the model validated an individual’s identity by comparing captured features with their own stored templates in the healthcare EHR system, allowing secure access to sensitive medical information. Data were obtained in two channels when each of the 200 subjects performed 12 motions. There were three sessions, and each motion was repeated 10 times with time intervals of a day or longer between each session. Experiments were conducted on a dataset of 20 randomly sampled subjects out of 200 subjects in the database, achieving exceptional identification accuracy. The experiment was conducted separately for 5, 10, 15, and 20 subjects using the ResNet1D model of a deep neural network, achieving accuracy rates of 97%, 96%, 87%, and 82%, respectively. The proposed model can be integrated with healthcare EHR systems to enable secure and reliable personal identification and the safeguarding of patient information.

Passive activity enhances residual control ability in patients with complete spinal cord injury

Abstract Patients with complete spinal cord injury retain the potential for volitional muscle activity in muscles located below the spinal injury level. However, because of prolonged inactivity, initial attempts to activate these muscles may not effectively engage any of the remaining neurons in the descending pathway. A previous study unexpectedly found that a brief clinical round of passive activity significantly increased volitional muscle activation, as measured by surface electromyography. In this study, we further explored the effect of passive activity on surface electromyographic signals during volitional control tasks among individuals with complete spinal cord injury. Eleven patients with chronic complete thoracic spinal cord injury were recruited. Surface electromyography data from eight major leg muscles were acquired and compared before and after the passive activity protocol. The results indicated that the passive activity led to an increased number of activated volitional muscles and an increased frequency of activation. Although the cumulative root mean square of surface electromyography amplitude for volitional control of movement showed a slight increase after passive activity, the difference was not statistically significant. These findings suggest that brief passive activity may enhance the ability to initiate volitional muscle activity during surface electromyography tasks and underscore the potential of passive activity for improving residual motor control among patients with motor complete spinal cord injury.

Lower Limb Motion Recognition with Improved SVM Based on Surface Electromyography

During robot-assisted rehabilitation, failure to recognize lower limb movement may efficiently limit the development of exoskeleton robots, especially for individuals with knee pathology. A major challenge encountered with surface electromyography (sEMG) signals generated by lower limb movements is variability between subjects, such as motion patterns and muscle structure. To this end, this paper proposes an sEMG-based lower limb motion recognition using an improved support vector machine (SVM). Firstly, non-negative matrix factorization (NMF) is leveraged to analyze muscle synergy for multi-channel sEMG signals. Secondly, the multi-nonlinear sEMG features are extracted, which reflect the complexity of muscle status change during various lower limb movements. The Fisher discriminant function method is utilized to perform feature selection and reduce feature dimension. Then, a hybrid genetic algorithm-particle swarm optimization (GA-PSO) method is leveraged to determine the best parameters for SVM. Finally, the experiments are carried out to distinguish 11 healthy and 11 knee pathological subjects by performing three different lower limb movements. Results demonstrate the effectiveness and feasibility of the proposed approach in three different lower limb movements with an average accuracy of 96.03% in healthy subjects and 93.65% in knee pathological subjects, respectively.

Evaluating the Ability of Brachial Plexus-Injured Patients to Control an Externally Powered (Myoelectric) Hand Prosthesis.

Restoration of hand function after traumatic brachial plexus injury (BPI) remains a formidable challenge. Traditional methods such as nerve or free muscle transfers yield suboptimal results. Advancements in myoelectric prostheses, characterized by novel signal acquisition and improved material technology, show promise in restoring functional grasp. This study evaluated the ability of adults with a BPI injury to control an externally powered prosthetic hand using nonintuitive signals, simulating the restoration of grasp with a myoelectric prosthesis. It also assessed the effectiveness of a comprehensive multidisciplinary evaluation in guiding treatment decisions. A multidisciplinary brachial plexus team assessed adults with compromised hand function due to BPI. The feasibility of amputation coupled with fitting of a myoelectric prosthesis for grasp reconstruction was evaluated. Participants' ability to control a virtual or model prosthetic hand using surface electromyography (EMG) as well as with contralateral shoulder motion-activated linear transducer signals was tested. The patient's input and injury type, along with the information from the prosthetic evaluation, were used to determine the reconstructive plan. The study also reviewed the number of participants opting for amputation and a myoelectric prosthetic hand for grasp restoration, and a follow-up survey was conducted to assess the impact of the initial evaluation on decision-making. Of 58 subjects evaluated, 47 (81%) had pan-plexus BPI and 42 (72%) received their initial assessment within 1 year post-injury. Forty-seven patients (81%) could control the virtual or model prosthetic hand using nonintuitive surface EMG signals, and all 58 could control it with contralateral uniscapular motion via a linear transducer and harness. Thirty patients (52%) chose and pursued amputation, and 20 (34%) actively used a myoelectric prosthesis for grasp. The initial evaluation was informative and beneficial for the majority of the patients, especially in demonstrating the functionality of the myoelectric prosthesis. The study indicates that adults with traumatic BPI can effectively operate a virtual or model myoelectric prosthesis using nonintuitive control signals. The simulation and multidisciplinary evaluation influenced informed treatment choices, with a high percentage of patients continuing to use the myoelectric prostheses post-amputation, highlighting its long-term acceptance and viability. Therapeutic Level IV. See Instructions for Authors for a complete description of levels of evidence.

Muscle fatigue detection based on sEMG signal using autocorrelation function and neural networks

Abstract Feature extraction from an recorded surface electromyography (sEMG) signal plays an important role in identifying and quantifying the characteristics of muscle activities. These features can be used for various applications like muscle function assessment, muscle fatigue detection, etc. Common features extracted from sEMG signal are time-domain or frequency-domain features. However, features which are sensitive to uncertainties in the signal like noise, movement artifacts, and outliers should be avoided. Autocorrelation function (ACF), which is a measure of similarity between a signal and its time delayed version, is considered in this work as a feature to overcome the impact of noise, artifacts, and outliers. An artificial neural network (ANN) is developed to differentiate between fatigue and non-fatigue conditions using the calculated ACF from sEMG segments. The performance of an ANN model that can be adapted by means of various regularization methods was investigated. The proposed ANN model achieved an accuracy of about 97.62 %, a precision of about 95.50 % and a sensitivity of about 100 % in the classification of fatigue and non-fatigue sEMG segments, outperforming k-means and linear support vector machine approaches that served as references.

Intelligent Human–Computer Interaction: Combined Wrist and Forearm Myoelectric Signals for Handwriting Recognition

Intelligent Human–Computer Interaction: Combined Wrist and Forearm Myoelectric Signals for Handwriting Recognition

How to use one surface electromyography sensor to recognize six hand movements for a mechanical hand in real time: a method based on Morse code.

Surface electromyography (sEMG) signal is a kind of physiological signal reflecting muscle activity and muscle force. At present, the existing methods of recognizing human motion intention need more than two sensors to recognize more than two kinds of movements, the sensor pasting positions are special, and the hardware conditions for execution are high. In this work, a real-time motion intention recognition method based on Morse code is proposed and applied to the mechanical hand. The short-time and long-term muscle contraction signals collected by a single sEMG sensor were extracted and encoded with the Morse code method, and then the developed mapping method from Morse code to six hand movements were used to recognize hand movements. The average recognition accuracy of hand movements was 94.8704 ± 2.3556%, the average adjusting time was 34.89s for all subjects, and the execution time of a single movement was 381ms. The corresponding experiment video can be found in the attachment to show the experiment. The method proposed in this work is a method with one sensor to recognize six movements, low hardware conditions, high recognition accuracy, and insensitive to the difference of sensor pasting position.

Accuracy of EMG linear envelope in identifying the peak of muscular activity during walking

BackgroundThe linear-envelope peak (LEP) of surface EMG signal is widely used in gait analysis to characterize muscular activity, especially in clinics. Research questionThis study is designed to evaluate LEP accuracy in identifying muscular activation and assessing activation timing during walking. MethodsSurface EMG signals from gastrocnemius lateralis (GL) and tibialis anterior (TA) were analyzed in 100 strides per subject (31 healthy subjects) during ground walking. Signals were full-wave rectified and low-pass filtered (cut-off frequency=5 Hz) to extract the linear envelope. LEP accuracy in identifying muscle activations and the associated error in peak detection were assessed by direct comparison with a reference method based on wavelet transform. LEP accuracy in identifying the timing of higher signalenergy levels was also assessed, increasing the reference-algorithm selectivity. ResultsThe detection error (percentage number of times when LEP falls outside the correspondent reference activation interval) is close to zero. Detection error increases up to 70% for intervals including only signal energy higher than 90% of energy peak. Mean absolute error (MAE, the absolute value of the distance between LEP timing and the correspondent actual timing of the sEMG-signal peak computed by reference algorithm) is 54.1±20.0 ms. Detection error and MAE are significantly higher (p<0.05) in TA data compared to GL signals. Differences among MAE values detected adopting different values for LE cut-off frequency are not statistically significant. SignificanceLEP was found to be accurate in identifying the number of muscle activations during walking. However, the use of LEP to assess the timing of highest sEMG-signal energy (signal peak) should be considered carefully. Indeed, it could introduce a relevant inaccuracy in muscle-activation identification and peak-timing quantification. The type of muscle to analyze could also influence LEP performances, while the cut-off frequency chosen for envelope extraction appears to have a limited impact.

Fusion inception and transformer network for continuous estimation of finger kinematics from surface electromyography.

Decoding surface electromyography (sEMG) to recognize human movement intentions enables us to achieve stable, natural and consistent control in the field of human computer interaction (HCI). In this paper, we present a novel deep learning (DL) model, named fusion inception and transformer network (FIT), which effectively models both local and global information on sequence data by fully leveraging the capabilities of Inception and Transformer networks. In the publicly available Ninapro dataset, we selected surface EMG signals from six typical hand grasping maneuvers in 10 subjects for predicting the values of the 10 most important joint angles in the hand. Our model's performance, assessed through Pearson's correlation coefficient (PCC), root mean square error (RMSE), and R-squared (R2) metrics, was compared with temporal convolutional network (TCN), long short-term memory network (LSTM), and bidirectional encoder representation from transformers model (BERT). Additionally, we also calculate the training time and the inference time of the models. The results show that FIT is the most performant, with excellent estimation accuracy and low computational cost. Our model contributes to the development of HCI technology and has significant practical value.

Corticomuscular and intermuscular coherence as a function of age and walking balance difficulty

We determined beta-band intermuscular (IMC) and corticomuscular coherence (CMC) as a function of age and walking balance difficulty. Younger (n=14, 23y) and older individuals (n=19, 71y) walked 13 m overground, on a 6-cm-wide ribbon overground, and on a 6-cm-wide (5-cm-high) beam. Walking distance as a proxy for walking balance and speed were computed. CMC was estimated between electroencephalographic signal at Cz electrode and surface electromyographic signals of seven leg muscles, while IMC was calculated in four pairs of leg muscles, during stance and swing gait phases. With increasing difficulty, walking balance decreased in old individuals and speed decreased gradually independent of age. Beam walking increased IMC, while age increased IMC in proximal muscle pairs, and decreased IMC in distal muscle pairs. Age and difficulty increased CMC independent of gait phases. Concluding, CMC and IMC increased with walking balance difficulty and age, except for distal muscle pairs, which had lower IMC with age. These findings suggest an age-related increase in corticospinal involvement in the neural control of walking balance. Data availabilityThe datasets used in this study are available from the corresponding author upon reasonable request.

Spectral characterization of human leg EMG signals from an open access dataset for the development of computational models.

Large-scale neuromusculoskeletal models have been used for predicting mechanisms underlying neuromuscular functions in humans. Simulations of such models provide several types of signals of practical interest, such as surface electromyographic signals (EMG), which are compared with experimental data for interpretations of neurophysiological phenomena under study. Specifically, realistic characterization of spectral properties of simulated EMG signals is important for achieving powerful inferences, whereas considerations should be taken for myoelectric signals of different muscles. In this study, we characterized spectral properties of surface interference pattern EMG signals and motor unit action potentials (MUAP) acquired from three plantar flexor muscles: Soleus (SO), Medial Gastrocnemius (MG), and Lateral Gastrocnemius (LG); and one dorsiflexor muscle: Tibialis Anterior (TA). Surface EMG signals were acquired from 20 participants using the same convention for electrode placement. Specifically, interference pattern EMG signals were obtained during isometric constant force contractions at 5%, 10% and 20% of maximum voluntary contraction (MVC), whereas surface MUAPs were decomposed from surface EMG signals obtained at low contraction forces. We compared the spectrum median frequency (MDF) estimated from interference pattern EMG signals across muscles and contraction intensities. Additionally, we compared MDF and durations of MUAPs between muscles. Our results showed that MDF of interference pattern EMG signals acquired from TA were higher compared to SO, MG, and LG for all contraction intensities i.e., 5%, 10%, and 20% MVC. Consistently, MUAPs acquired from TA also had higher MDF values and shorter durations compared to the other leg muscles. We provide herein a dataset with the surface MUAPs waveforms and interference pattern EMG signals obtained for this study, which should be useful for implementing and validating the simulation of myoelectrical signals of leg muscles. Importantly, these results indicate that spectral properties of myoelectrical signals should be considered for improving EMG modeling in large-scale neuromusculoskeletal models.

Mapping Method of Human Arm Motion Based on Surface Electromyography Signals.

This paper investigates a method for precise mapping of human arm movements using sEMG signals. A multi-channel approach captures the sEMG signals, which, combined with the accurately calculated joint angles from an Inertial Measurement Unit, allows for action recognition and mapping through deep learning algorithms. Firstly, signal acquisition and processing were carried out, which involved acquiring data from various movements (hand gestures, single-degree-of-freedom joint movements, and continuous joint actions) and sensor placement. Then, interference signals were filtered out through filters, and the signals were preprocessed using normalization and moving averages to obtain sEMG signals with obvious features. Additionally, this paper constructs a hybrid network model, combining Convolutional Neural Networks and Artificial Neural Networks, and employs a multi-feature fusion algorithm to enhance the accuracy of gesture recognition. Furthermore, a nonlinear fitting between sEMG signals and joint angles was established based on a backpropagation neural network, incorporating momentum term and adaptive learning rate adjustments. Finally, based on the gesture recognition and joint angle prediction model, prosthetic arm control experiments were conducted, achieving highly accurate arm movement prediction and execution. This paper not only validates the potential application of sEMG signals in the precise control of robotic arms but also lays a solid foundation for the development of more intuitive and responsive prostheses and assistive devices.

Nonlinear dynamics of postural control system under visual-vestibular habituation balance practice: evidence from EEG, EMG and center of pressure signals.

Human postural control system is inherently complex with nonlinear interaction among multiple subsystems. Accordingly, such postural control system has the flexibility in adaptation to complex environments. Previous studies applied complexity-based methods to analyze center of pressure (COP) to explore nonlinear dynamics of postural sway under changing environments, but direct evidence from central nervous system or muscular system is limited in the existing literature. Therefore, we assessed the fractal dimension of COP, surface electromyographic (sEMG) and electroencephalogram (EEG) signals under visual-vestibular habituation balance practice. We combined a rotating platform and a virtual reality headset to present visual-vestibular congruent or incongruent conditions. We asked participants to undergo repeated exposure to either congruent (n = 14) or incongruent condition (n = 13) five times while maintaining balance. We found repeated practice under both congruent and incongruent conditions increased the complexity of high-frequency (0.5-20 Hz) component of COP data and the complexity of sEMG data from tibialis anterior muscle. In contrast, repeated practice under conflicts decreased the complexity of low-frequency (<0.5 Hz) component of COP data and the complexity of EEG data of parietal and occipital lobes, while repeated practice under congruent environment decreased the complexity of EEG data of parietal and temporal lobes. These results suggested nonlinear dynamics of cortical activity differed after balance practice under congruent and incongruent environments. Also, we found a positive correlation (1) between the complexity of high-frequency component of COP and the complexity of sEMG signals from calf muscles, and (2) between the complexity of low-frequency component of COP and the complexity of EEG signals. These results suggested the low- or high-component of COP might be related to central or muscular adjustment of postural control, respectively.

Analysis of Muscle Fatigue Progression Using Geometric Features of Surface Electromyography Signals and Explainable XGBoost Classifier

Analysis of Muscle Fatigue Progression Using Geometric Features of Surface Electromyography Signals and Explainable XGBoost Classifier

Study on Gesture Recognition Method with Two-Stream Residual Network Fusing sEMG Signals and Acceleration Signals.

Currently, surface EMG signals have a wide range of applications in human-computer interaction systems. However, selecting features for gesture recognition models based on traditional machine learning can be challenging and may not yield satisfactory results. Considering the strong nonlinear generalization ability of neural networks, this paper proposes a two-stream residual network model with an attention mechanism for gesture recognition. One branch processes surface EMG signals, while the other processes hand acceleration signals. Segmented networks are utilized to fully extract the physiological and kinematic features of the hand. To enhance the model's capacity to learn crucial information, we introduce an attention mechanism after global average pooling. This mechanism strengthens relevant features and weakens irrelevant ones. Finally, the deep features obtained from the two branches of learning are fused to further improve the accuracy of multi-gesture recognition. The experiments conducted on the NinaPro DB2 public dataset resulted in a recognition accuracy of 88.25% for 49 gestures. This demonstrates that our network model can effectively capture gesture features, enhancing accuracy and robustness across various gestures. This approach to multi-source information fusion is expected to provide more accurate and real-time commands for exoskeleton robots and myoelectric prosthetic control systems, thereby enhancing the user experience and the naturalness of robot operation.

Wearable Sensor‐Based Multi‐modal Fusion Network for Automated Gait Dysfunction Assessment in Children with Cerebral Palsy

Gait, fundamental to human movement, becomes compromised in cerebral palsy (CP), a childhood‐onset central nervous system motor disorder. Precise assessment of patients’ gait is crucial for tailored rehabilitation interventions. Currently, clinical scales assessing CP gait dysfunction mostly, while valuable, rely on subjective clinician observations. To enhance objectivity and efficiency in CP diagnosis and rehabilitation, there is a need for more objective assessment procedures. This study introduces a multi‐modal and multi‐scale feature fusion (MMFF) framework, a new framework for automating gait dysfunction assessment in children with CP. By utilizing surface electromyography and acceleration signals recorded during children's walking, MMFF generates a feature vector enriched with adaptively refined feature maps, cross‐mode correlations, and both local and global information. Validation of MMFF's effectiveness is evident through an accomplished classification accuracy of 99.13%. The mean values for precision, recall, and F1‐score in Gross Motor Function Classification System (GMFCS)‐1, GMFCS‐2, and GMFCS‐3, reaching 99.00%, 99.00%, and 98.33%, respectively, further reflect the accuracy of functional assessments at each level. This study underscores MMFF's potential as an objective, streamlined tool for clinicians, promising improved gait assessment and well‐informed rehabilitation strategies for children with CP.

FFCSLT: A Deep Learning Model for Traffic Police Hand Gesture Recognition Using Surface Electromyographic Signals

FFCSLT: A Deep Learning Model for Traffic Police Hand Gesture Recognition Using Surface Electromyographic Signals

Motor unit discharge rate modulation during isometric contractions to failure is intensity- and modality-dependent.

The physiological mechanisms determining the progressive decline in the maximal muscle torque production capacity during isometric contractions to task failure are known to depend on task demands. Task-specificity of the associated adjustments in motor unit discharge rate (MUDR), however, remains unclear. This study examined MUDR adjustments during different submaximal isometric knee extension tasks to failure. Participants performed a sustained and an intermittent task at 20% and 50% of maximal voluntary torque (MVT), respectively (Experiment 1). High-density surface EMG signals were recorded from vastus lateralis (VL) and medialis (VM) and decomposed into individual MU discharge timings, with the identified MUs tracked from recruitment to task failure. MUDR was quantified and normalised to intervals of 10% of contraction time (CT). MUDR of both muscles exhibited distinct modulation patterns in each task. During the 20% MVT sustained task, MUDR decreased until ∼50% CT, after which it gradually returned to baseline. Conversely, during the 50% MVT intermittent task, MUDR remained stable until ∼40-50% CT, after which it started to continually increase until task failure. To explore the effect of contraction intensity on the observed patterns, VL and VM MUDR was quantified during sustained contractions at 30% and 50% MVT (Experiment 2). During the 30% MVT sustained task, MUDR remained stable until ∼80-90% CT in both muscles, after which it continually increased until task failure. During the 50% MVT sustained task the increase in MUDR occurred earlier, after ∼70-80% CT. Our results suggest that adjustments in MUDR during submaximal isometric contractions to failure are contraction modality- and intensity-dependent. KEY POINTS: During prolonged muscle contractions a constant motor output can be maintained by recruitment of additional motor units and adjustments in their discharge rate. Whilst contraction-induced decrements in neuromuscular function are known to depend on task demands, task-specificity of motor unit discharge behaviour adjustments is still unclear. In this study, we tracked and compared discharge activity of several concurrently active motor units in the vastii muscles during different submaximal isometric knee extension tasks to failure, including intermittent vs. sustained contraction modalities performed in the same intensity domain (Experiment 1), and two sustained contractions performed at different intensities (Experiment 2). During each task, motor units modulated their discharge rate in a distinct, biphasic manner, with the modulation pattern depending on contraction intensity and modality. These results provide insight into motoneuronal adjustments during contraction tasks posing different demands on the neuromuscular system.

Automated detection and removal of artifacts from sEMG signals based on fuzzy inference system and signal decomposition methods

Surface electromyography (sEMG) signal quality decreases when it is contaminated by different types of artifacts. Detection and removal of the contaminants from sEMG signals were a major and interesting task for several research works in recent years. In this paper, a novel automatic approach was designed for the detection and removal of artifacts from sEMG signals. The proposed method consists of a combined model that based on a fuzzy inference system (FIS), a signal decomposition method, and a denoising technique. Based on three spectral statistical features, which are calculated in frequency domain using power spectral density: composite multiscale entropy (CMSE), kurtosis (K), and skewness (S), FIS detects the artifactual epochs of the sEMG signal. Thereafter, those epochs are decomposed into their components by a signal decomposition. In this work, we have tested the performance of two decomposition methods namely stationary wavelet transform (SWT) and variational mode decomposition (VMD). A second FIS is used to select the contaminated components. Depending on the type of artifact, two procedures are considered, the first aims to denoise motion artifact (MOA), electrocardiography interference (ECG) and power line interference (PLI), whereas the second procedure aims to eliminate the additive white gaussian noise (AWGN). Finally, the denoised signal is reconstructed. The evaluation performance of the proposed method is compared with the performance of some existing methods using the following metrics: SNR improvement (ΔSNR), the reduction in artifact (λ(%)), improvement in signal to noise and distortion ratio (ΔSNDR) and improvement in the root mean square error (ΔRMSE). The results emphazed the superiority of our method over others in terms of all the used metrics and computational time.

InRes-ACNet: Gesture Recognition Model of Multi-Scale Attention Mechanisms Based on Surface Electromyography Signals

Surface electromyography (sEMG) signals are the sum of action potentials emitted by many motor units; they contain the information of muscle contraction patterns and intensity, so they can be used as a simple and reliable source for grasping mode recognition. This paper introduces the InRes-ACNet (inception–attention–ACmix-ResNet50) model, a novel deep-learning approach based on ResNet50, incorporating multi-scale modules and self-attention mechanisms. The proposed model aims to improve gesture recognition performance by enhancing its ability to extract channel feature information within sparse sEMG signals. The InRes-ACNet model is evaluated on the NinaPro DB1 and NinaPro DB5 datasets; the recognition accuracy for these datasets can reach 87.94% and 87.04%, respectively, and recognition accuracy can reach 88.37% in the grasping mode prediction of an electromyography manipulator. The results show that the fusion of multi-scale modules and self-attention mechanisms endows a strong ability for the task of gesture recognition based on sparse sEMG signals.