Diagnosis of Neuromuscular Disorders Using TQWT and Random Subspace Ensemble Classifier

Automated EMG Signal Classification for Diagnosis of Neuromuscular Disorders Using DWT and Bagging

Medical decision support system for diagnosis of neuromuscular disorders using DWT and fuzzy support vector machines

Human muscles are composed of motor units and each motor (MU) unit is composed of a specific α-motor neuron and the muscle fibres it innervates. A motor neuron innervates the muscle fibres of a MU via the neuromuscular junction (NMJ) formed at the terminal end of each branch of its axon. Voluntary muscle contractions are initiated when the central nervous system recruits MUs by activating their motor neurons, which in turn, via their NMJs, activate their muscle fibres. At each NMJ, a region of transmembrane current is produced across the sarcolemma membrane of its corresponding fibre when the motor neuron is activated (i.e. discharges an action potential). This transmembrane current creates a change in transmem‐ brane potential (or action potential) which propagates along the fibre and initiates/co-ordinates its contraction [1]. The currents creating the action potentials of the activated fibres of recruited MUs summate to create dynamic electric fields in the volume conductor in and around muscles. Electrodes placed in these electric fields detect time changing voltage signals which are the electromyographic (EMG) signals discussed in this chapter. When a muscle is affected by a neuromuscular disorder, characteristics of its action potentials, and as a result of the EMG signals they create, change depending on whether the muscle is affected by a myopathic or neurogenic disorder and the extent to which the muscle is affected. Therefore, quantitative EMG signal analysis can be used to support the diagnosis of neuromuscular disorders. Clinical quantitative electromyography (QEMG) attempts to use the information contained in an EMG signal to characterize the muscle from which it was detected to support clinical decisions related to the diagnosis, treatment or management of neuromuscular disorders.

Machine learning methods can be used to diagnose neuromuscular illnesses using electromyographic (EMG) signals. This research examines the tunable-Q factor wavelet transform (TQWT) for feature extraction and analyses various learning methods for classifying EMG signals in order to detect neuromuscular diseases. TQWT decomposes each type of EMG signal into sub-bands first. From each sub-band, statistical parameters such as mean absolute values (MAV), inter quartile range (IQR), kurtosis, mode, standard deviation, skewness, and ratio are calculated. Finally, the extracted features are fed into classifiers to differentiate between ALS, myopathy, and normal EMG data. The random forest classifier with TQWT achieved higher classification results in neuromuscular disorders diagnosis than the other classifiers tested in this study, according to experimental results. The accuracy of the random forest approach using TQWT was 98.64%, with an F-measure of 0.986 and a kappa value of 0.979.

Electromyographic (EMG) signal decomposition is the process of resolving an EMG signal into its constituent motor unit potential trains (MUPTs). The purpose of EMG signal decomposition is to provide an estimate of the firing pattern and motor unit potential (MUP) template of each active motor unit (MU) that contributed significant MUPs to the EMG signal. The extracted MU firing patterns, MUP templates, and their estimated feature values can assist with the diagnosis of neuromuscular disorders (Stalberg & Falck, 1997; Troger & Dengler, 2000; Fuglsang-Frederiksen, 2006; Pino et al., 2008; Farkas et al., 2010), the understanding of motor control ( De Luca et al. 1982a, 1982b; Contessa et al.,2009), and the characterization of MU architecture (Lateva & McGill, 2001), but only if they are valid trains. Depending on the complexity of the signal being decomposed, the variability of MUP shapes and MU firing patterns, and the criteria and parameters used by the decomposition algorithm to merge or split the obtained MUPTs, several invalid MUPTs may be created.

This paper presents the use of a Wavelet Neural Network (WNN) as an efficient classifier of Electromyographic (EMG) signals. Generally, an EMG signal requires advanced methods for detection, decomposition, processing and classification. In this paper a WNN model will relate the firing frequency of motor unit action potentials (MUAPs) and three different muscle force levels, in order to improve the classification process showed by other common processing techniques. Adequate EMG classification provides an important source of information in fields such as the diagnosis of neuromuscular disorders, management rehabilitation and prosthesis control were identify and classify MUAPs is a priority task. Accurate and computational efficient EMG classifier was obtained employing a WNN model; the success classification rate was greater than 90% for original registers and 83.33% in adding 50% of noise. WNN allow the feature extraction of EMG signals while creating a classification model, all in a single step, becoming an innovative data processing tool.

The shapes and sounds of isolated motor unit action potentials (MUAPs) in an electromyographic (EMG) signal provide a significant source of information for diagnosis, treatment and management of neuromuscular disorders. These parameters can be analyzed qualitatively by an expert or quantitatively by using pattern recognition techniques. Due to the advantages of quantitative EMG method, developing robust automated MUAP classifiers have been explored and several systems have been developed for this purpose by now, but the accuracy of the existing methods is not high enough to be used in clinical environments. In this paper, a novel classification strategy based on ensemble of support vector machines (SVMs) classifiers in hybrid serial/parallel architecture is proposed to determine the class label (myopathic, neuropathic, or normal) for a given MUAP. The developed system employs both time domain and time-frequency domain features of the MUAPs extracted from an EMG signal using an EMG signal decomposition system. Different classification strategies including single classifier and multiple classifiers with several subsets of features were investigated. Experimental results using a set of real EMG signals showed robust performance of multi-classifier methods proposed here. Of the methods studied, the multi-classifier that uses multiple features sets and a combination of both trainable and nontrainable fusion techniques to aggregate base classifiers showed the best performance with average accuracy of 97% which is significantly higher than the average accuracy of single SVM-based classifier system (i.e., 88%).

Electromyographic (EMG) signal provide a significant source of information for diagnosis, treatment and management of neuromuscular disorders. This paper is aim at introducing an effective multi-classifier approach to enhance classification accuracy .The proposed system employs both time domain and time-frequency domain features of motor unit action potentials (MUAPs) extracted from an EMG signal. Different classification strategies including single classifier and multiple classifiers with time domain and time frequency domain features were investigated. Support Vector Machine (SVM) and K-nearest neighborhood (KNN) classifier used predict class label (Myopathic, Neuropathic, or Normal) for a given MUAP. Extensive analysis is performed on clinical EMG database for the classification of neuromuscular diseases and it is found that the proposed methods provide a very satisfactory performance in terms overall classification accuracy.

Intramuscular Electromyography (EMG) signal provides a significant source of information that plays an inevitable role in the diagnosis of neuromuscular disorders. The ensemble method represents a supervised machine learning algorithm that constructs a combination of classifiers to achieve accurate classification decision. In this respect, the aim of this study is to propose classification method for diagnosis of neuromuscular disorders including Amyotrophic lateral sclerosis (ALS) and myopathy diseases, through using publicly available EMG data, recorded by intramuscular single-channel EMG electrode. The ensemble of bagged tree algorithm generates and averages multiple versions of decision tree predicatore that are formed by producing a bootstrap replicate of the EMG learning data set. A set of ten time-domain features extracted from the intramuscular EMG recordings are classified by the ensemble of bagged tree algorithm. A classification accuracy rate of 92.8% was obtained with the proposed bagged tree classifier, which reports high classification accuracy. The outcome of this study can assist the clinicians to decide the correct diagnosis of neuromuscular disorders.

Abstract: Electromyography (EMG) is a widely used technique for analyzing muscle activity and has significant applications in prosthetics, rehabilitation, and neuromuscular disorder diagnosis. The classification of EMG signals remains a challenging task due to their complex, non-stationary nature and susceptibility to noise. To improve classification accuracy, this study employs a hybrid approach that integrates multiple feature extraction techniques with soft computing methods. The proposed methodology involves data acquisition from multiple subjects performing different hand and forearm movements. The raw EMG signals are preprocessed using noise filtering, segmentation, and normalization techniques to ensure high-quality input data. A comprehensive feature extraction process is then applied, combining time-domain features such as Mean Absolute Value, Root Mean Square, Waveform Length, and Zero Crossing, along with frequency-domain features including Mean Frequency, Median Frequency, Power Spectral Density, and Wavelet Coefficients. This combined feature set provides a more detailed representation of the EMG signals, capturing both temporal and spectral characteristics essential for effective classification. The study highlights the importance of integrating diverse feature extraction techniques to enhance EMG signal interpretation. The findings contribute to the development of more accurate EMG-based control systems for assistive technologies, including prosthetic devices and rehabilitation tools. By leveraging a comprehensive approach to EMG signal processing, this research aims to improve muscle activity classification and enable more precise control of biomedical applications. Soft computing techniques, including ANN, SVM Systems, have shown great potential in handling the inherent variability of EMG signals. These methods leverage learning-based approaches to model complex patterns and improve classification accuracy. By integrating time-domain and frequency-domain features, a more comprehensive representation of the EMG signal can be achieved, enabling soft computing models to perform better in distinguishing different muscle activities. Future research will focus on optimizing deep learning techniques for real-time EMG classification and extending the approach to broader applications such as brain-computer interfaces and human-computer interaction. The proposed methodology offers a promising step toward the development of intelligent, high-performance EMG-based systems. Keywords: Electromyography, EMG Signal Classification, Feature Extraction, Soft Computing, Artificial Neural Networks, Support Vector Machines.

Electromyography (EMG) is a technique for evaluating and recording the electrical activity produced by skeletal muscles to see muscle condition. Nervous system always controls the muscle activities such as relaxation or contraction. An efficient analysis of electromyography (EMG) signals plays an inevitable role in the diagnosis of neuromuscular disorders, prosthesis, and several related applications. Our aim in this study is to differentiate neuromuscular disorder patients from healthy people based on EMG signals. The EMG signals used in this research were recorded from biceps. Artificial Neural Network (ANN) was used for the classification. Eleven features were extracted from the EMG signals for the classification purpose. A comparative analysis was done based on the results. The outcome of this study encourages possible extension of this approach to improve stronger, more resilient and effective implementations.

A 55-year-old woman was admitted to the hospital with unstable gait followed by four-limb weakness with lower limb predominance. Several days before the onset of neurological symptoms, she developed mild respiratory symptoms, generalized myalgias without fever, and an episode of diarrhea. On initial examination, she demonstrated moderate weakness mainly affecting the lower limbs (she could stand but was unable to walk), decreased proprioception, distal hypoesthesia, and generalized areflexia. A polymerase chain reaction nasal swab test was positive for severe acute respiratory syndrome–coronavirus 2; respiratory support was not required. Blood tests, cerebrospinal fluid analysis, chest X ray, and cranial computed tomography revealed no abnormalities. Due to suspected Guillain-Barré syndrome (GBS), electrodiagnostic studies (EDx) were performed on day 9 after admission, and the patient was treated with intravenous immunoglobulin 0.4 g/kg for 5 days. An initial nerve conduction study (NCS) revealed increased temporal dispersion (158%) and reduced proximal/distal compound muscle action potential (CMAP) size (0.5) in the right median and ulnar nerves with elbow stimulation (normal distal CMAP), reduced tibial conduction velocities bilaterally (29 m/s), and absent tibial and median F waves. Thus, the patient met three criteria for demyelinating polyneuropathy.1 Electromyographic signals were recorded, using a 38 × 0.45-mm Neuroline concentric needle (Ambu, Ballerup, Denmark), from the right deltoid, extensor digitorum communis, first dorsal interosseous, tensor fascia latae, and vastus lateralis, and bilaterally from the tibialis anterior and gastrocnemius, and then bandpass filtered at 20 Hz to 10 kHz and stored using a KeyPoint.Net 3.22 device (Alpine Biomed, Fountain Valley, California). Spontaneous activity was qualitatively analyzed and motor unit potentials (MUPs) from electromyographic (EMG) signals with 200 ± 50 turns/s were quantitatively analyzed offline using decomposition-based quantitative EMG2 and near-fiber EMG (NFEMG),3, 4 which together aid in the diagnosis of neuromuscular disorders by quantifying intrinsic motor unit (MU) morphological and electrophysiological properties. A near-fiber MUP (NFM) is created by low-pass double-differentiation filtering a MUP, which, like SFEMG bandpass filtering, emphasizes contributions from fibers close to the needle detection surface (near fibers [NFs]). MUP area represents MU size. NFM duration (the time between the NFM onset and end positions) and NFM dispersion (the time between the first and last detected NF contribution) do not reflect MU size; rather, they reflect MU electrophysiological temporal dispersion (ie, differences in MU axonal branch conduction, neuromuscular junction [NMJ] transmission, and muscle fiber action potential [MFAP] conduction times). NFM segment jitter reflects NFM temporal instabilities, caused by variability in MU axonal branch conduction, NMJ transmission, and MFAP conduction times. Initially, low-amplitude and -frequency positive sharp waves and fibrillation potentials were recorded bilaterally from the tibialis anterior and gastrocnemius. Recruitment was reduced in all muscles sampled, with a predominance of large-area and irregularly shaped MUPs recorded from early-recruited MUs (more pronounced in distal lower limb muscles). Initial NFEMG measures showed increased dispersion and segment jitter in nearly all lower limb muscles sampled (Figure 1). Subsequent EDx, performed 5 weeks later, showed improved NCS results. Although the criteria for demyelinating polyneuropathy were still met, significant clinical improvement was seen, as the patient was able to walk with minor assistance. In addition, MUP area, NFM dispersion, and segment jitter were all reduced Figure 1. Figure 2 shows examples of initial and subsequently recorded MUPs and NFMs. In GBS, early recruitment of MUs with large MUPs is a consequence of conduction block affecting small-diameter axons.5 However, the electrophysiology of MUPs in GBS (ie, quantification of MUP size, temporal dispersion, and stability) is not usually included in the diagnostic protocols for GBS.6 Our results could be explained by both a transient impairment of smaller diameter myelinated motor axons (possibly a consequence of conduction block affecting proximal nerve segments, manifesting as early recruitment of large MUs and reduced numbers of small area MUPs) and possible transient electrophysiological impairment of either MU distal axonal branches or their NMJs (manifesting as transient increased NFM dispersion and segment jitter). Also, axonal degeneration and rapid regeneration of short myelinated segments of intramuscular terminal axonal branches has been associated with immune-mediated subtypes of GBS,7 which could explain the early active denervation and rapid motor recovery observed. These findings support the combined use of EMG and NFEMG in suspected polyneuropathy. Because MUP area reflects MU size while NFM duration, dispersion, and stability reflect MU electrophysiological dispersion and stability, respectively, their combined use can provide valuable information for early diagnosis and management of treatable disorders. The authors declare no potential conflicts of interest. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. Data available on request due to privacy/ethical restrictions

The neuromuscular disorders are diagnosed using electromyographic (EMG) signals. Machine learning algorithms are employed as a decision support system to diagnose neuromuscular disorders. This paper compares bagging and boosting ensemble learning methods to classify EMG signals automatically. Even though ensemble classifiers' efficacy in relation to real-life issues has been presented in numerous studies, there are almost no studies which focus on the feasibility of bagging and boosting ensemble classifiers to diagnose the neuromuscular disorders. Therefore, the purpose of this paper is to assess the feasibility of bagging and boosting ensemble classifiers to diagnose neuromuscular disorders through the use of EMG signals. It should be understood that there are three steps to this method, where the step number one is to calculate the wavelet packed coefficients (WPC) for every type of EMG signal. After this, it is necessary to calculate statistical values of WPC so that the distribution of wavelet coefficients could be demonstrated. In the last step, an ensemble classifier used the extracted features as an input of the classifier to diagnose the neuromuscular disorders. Experimental results showed the ensemble classifiers achieved better performance for diagnosis of neuromuscular disorders. Results are promising and showed that the AdaBoost with random forest ensemble method achieved an accuracy of 99.08%, F-measure 0.99, AUC 1, and kappa statistic 0.99.

The neuromuscular disorders are identified by using electromyographic (EMG) signals. Machine learning techniques are used as a decision support system to diagnose the neuromuscular disorders. The time-frequency methods are widely employed to extract features from the EMG signals for the diagnosis of the neuromuscular disorders. In this chapter, wavelet-based time-frequency techniques are compared for the automatic classification of EMG signals. The experiments are carried out to categorize EMG signals into ALS (amyotrophic lateral sclerosis), control, or myopathic. The proposed framework is composed of three main modules. In the first module, EMG signals are denoised by using MSPCA (multiscale principal component analysis) denoising technique. In the second module, the coefficients of wavelet-based time-frequency methods are calculated for each category of EMG signal, and then statistical values of each sub-band are computed. In the last module, the extracted features are employed as an input to a classifier to diagnose different neuromuscular disorders. The obtained results obviously show that features extracted by using DT-CWT (dual-tree complex wavelet transform) are highly discriminative for the MUAP (motor unit action potential) classification as compared to other wavelet-based time-frequency methods. Using DT-CWT features along with the SVM (support vector machine) classifier accomplished a classification accuracy of 99.6%. Hence, the proposed technique can be employed for the diagnosis of neuromuscular disorders as a decision support system.

Amputation is the removal of limb by trauma, medical illness, or surgery. A transplant or prosthesis is the only option for recovering the loss. This issue can be solved by, measuring muscle activation via electric potential, referred to as electromyography (EMG), has traditionally been used for medical research and diagnosis of neuromuscular disorders. Our focus is on reproducing moving operations using non-invasive electromyogram signals. In this article the authors present a novel approach for developing a robotic arm for amputees based on raw EMG signal acquired from amputees. The main objective is to develop a Prosthesis that functions using the EMG signals generated in the own body of Amputees. The development of prosthetic arm includes Arduino microcontroller for two motions such as upper and down movement. Also the paper is extended in a microprocessor platform involving my-RIO microprocessor of national instruments for accurate EMG signal acquisition and robotic arm movements.