Electromyographic (EMG) signals contain both neural and muscle information. Consequently, EMG signals can be modelled as the composition of two component signals, one of these being a low frequency neural input, the other a relatively high frequency, constant spectrally shaped, stationary, unitary muscle response. Utilizing this model and homomorphic processing estimates of the two component signals can be obtained. These estimates contain neural and muscle information respectively. This thesis establishes the basis for the use of this multiplicative model. It also outlines the application of multiplicative homomorphic processing to EMG signals. The results of this processing are shown to be valid and to contain useful information. The thesis concludes that the model is both appropriate and useful. It also points out that the use of this model and homomorphic processing allows the simultaneous extraction of both neural and muscle information from the EMG signal,a result which is not possible with other currently used processing techniques. / Thesis / Master of Engineering (ME)

The main goal of this thesis was to discover the relationships between MU characteristics and MUP features. To reach this goal, several features explaining the anatomical structure of the muscle were introduced. Additionally, features representing specific properties of the EMG signal detected from that muscle, were defined. Since information regarding the underlying anatomy was not available from real data, a physiologically based muscle model was used to extract the required features. This muscle model stands out from others, by providing similar acquisition schemes as the ones utilized by physicians in real clinical settings and by modelling the interactions among different volume conductor factors and the collection of MUs in the muscle in a realistic way. Having the features ready, several relationship discovery techniques were used, to reveal relationships between MU features and MUP features. To interpret the results obtained from the correlation analysis and pattern discovery techniques properly, several algorithms and new statistics were defined. The results obtained from correlation analysis and pattern discovery technique were similar to each other, and suggested that to maximize the inter-relationships between MUP features and MU features, MUPs could be filtered based on their slope values, specifically MUPs with slopes lower than 0. 6 v/s could be excluded. Additionally PDT results showed that high slope MUPs were not as informative about the underlying MU and could be excluded to maximize the relationships between MUP features and MU characteristics. Certain MUP features were determined to be highly related to certain MU characteristics. MUP <em>area</em> and <em>duration</em> were shown to be the best representative feature for the MU size and <em>average fiber density</em>, respectively. For the distribution of fiber diameter in the MU, <em>duration</em> and <em>number of turns</em> were determined to reflect <em>mean fiber diameter</em> and <em>stdv of fiber diameter</em> the best, correspondingly.

The main goal of this thesis was to discover the relationships between MU characteristics and MUP features. To reach this goal, several features explaining the anatomical structure of the muscle were introduced. Additionally, features representing specific properties of the EMG signal detected from that muscle, were defined. Since information regarding the underlying anatomy was not available from real data, a physiologically based muscle model was used to extract the required features. This muscle model stands out from others, by providing similar acquisition schemes as the ones utilized by physicians in real clinical settings and by modelling the interactions among different volume conductor factors and the collection of MUs in the muscle in a realistic way. Having the features ready, several relationship discovery techniques were used, to reveal relationships between MU features and MUP features. To interpret the results obtained from the correlation analysis and pattern discovery techniques properly, several algorithms and new statistics were defined. The results obtained from correlation analysis and pattern discovery technique were similar to each other, and suggested that to maximize the inter-relationships between MUP features and MU features, MUPs could be filtered based on their slope values, specifically MUPs with slopes lower than 0. 6 v/s could be excluded. Additionally PDT results showed that high slope MUPs were not as informative about the underlying MU and could be excluded to maximize the relationships between MUP features and MU characteristics. Certain MUP features were determined to be highly related to certain MU characteristics. MUP <em>area</em> and <em>duration</em> were shown to be the best representative feature for the MU size and <em>average fiber density</em>, respectively. For the distribution of fiber diameter in the MU, <em>duration</em> and <em>number of turns</em> were determined to reflect <em>mean fiber diameter</em> and <em>stdv of fiber diameter</em> the best, correspondingly.

abstract: The use of electromyography (EMG) signals to characterize muscle fatigue has been widely accepted. Initial work on characterizing muscle fatigue during isometric contractions demonstrated that its frequency decreases while its amplitude increases with the onset of fatigue. More recent work concentrated on developing techniques to characterize dynamic contractions for use in clinical and training applications. Studies demonstrated that as fatigue progresses, the EMG signal undergoes a shift in frequency, and different physiological mechanisms on the possible cause of the shift were considered. Time-frequency processing, using the Wigner distribution or spectrogram, is one of the techniques used to estimate the instantaneous mean frequency and instantaneous median frequency of the EMG signal using a variety of techniques. However, these time-frequency methods suffer either from cross-term interference when processing signals with multiple components or time-frequency resolution due to the use of windowing. This study proposes the use of the matching pursuit decomposition (MPD) with a Gaussian dictionary to process EMG signals produced during both isometric and dynamic contractions. In particular, the MPD obtains unique time-frequency features that represent the EMG signal time-frequency dependence without suffering from cross-terms or loss in time-frequency resolution. As the MPD does not depend on an analysis window like the spectrogram, it is more robust in applying the timefrequency features to identify the spectral time-variation of the EGM signal. / Dissertation/Thesis / M.S. Electrical Engineering 2012

No description available.

The electrical activity of skeletal muscle¡ªthe electromyogram (EMG)¡ªis of value to many different application areas, including ergonomics, clinical biomechanics and prosthesis control. For many applications the EMG is related to muscular tension, joint torque and/or applied forces. In these cases, a goal is for an EMG-torque model to emulate the natural relationship between the central nervous system and peripheral joints and muscles. This thesis mainly describes an experimental study which relates the simultaneous biceps/triceps surface EMG of 12 subjects to elbow torque at seven joint angles (ranging from 45¡ÃƒÂ£to 135¡ÃƒÂ£) during constant-posture, quasi-constant-torque contractions. The contractions ranged between 50% maximum voluntary contractions (MVC) extension and 50% MVC flexion. Advanced EMG amplitude (EMG¦Ãƒâ€™) estimation processors were investigated, and three nonlinear EMG¦Ãƒâ€™-torque models were evaluated. Results show that advanced (i.e., whitened, multiple-channel) EMG¦Ãƒâ€™ processors lead to improved joint torque estimation, compared to unwhitened, single-channel EMG¦Ãƒâ€™ processors. Depending on the joint angle, use of the multiple-channel whitened EMG¦Ãƒâ€™ processor with higher polynomial degrees produced a median error that was 50%-66% that found when using the single-channel, unwhitened EMG¦Ãƒâ€™ processor with a polynomial degree of 1. The best angle-specific model achieved a minimum error of 3.39% MVCF90 (i.e., error referenced to MVC at 90¢X flexion), yet it does not allow interpolation across angles. The best model which parameterizes the angle dependence achieved an error of 3.55% MVCF90. This thesis also summarizes other collaborative research contributions performed as part of this thesis. (1) Decomposition of needle EMG data was performed as part of a study to characterize motor unit behavior in patients with amyotrophic lateral sclerosis (ALS) [with Spaulding Rehabilitation Hospital, Boston, MA]. (2) EMG-force modeling of force produced at the finger tips was studied with the purpose of assessing the ability to determine two or more independent, continuous degrees of freedom of control from the muscles of the forearm [with WPI and Sherbrooke University]. (3) Identification of a nonlinear, dynamic EMG-torque relationship about the elbow was studied [WPI]. (4) Signal whitening preprocessing for improved classification accuracies in myoelectric control of a prosthesis was studied [with WPI and the University of New Brunswick].

Electromyogram (EMG) is an established tool to study operation of neuromuscular systems. In analysing EMG signals, accurate detection of the movement-related events in the signal is frequently necessary. I explored the application of change-point detection algorithm proposed by Moskvina et. al., 2003 to EMG event detection, and evaluated the technique’s performance comparing it to two common threshold-based event detection methods and to the visual estimates of the EMG events performed by trained practitioners in the field. The algorithm was implemented in MATLAB and applied to EMG segments recorded from wrist and trunk muscles. The quality and frequency of successful detection were assessed for all methods, using the average visual estimate as the baseline, against which techniques were evaluated. The application showed that the change-point detection can successfully locate multiple changes in the EMG signal, but the maximum value of the detection statistic did not always identify the muscle activation onset.

Electromyogram (EMG) is an established tool to study operation of neuromuscular systems. In analysing EMG signals, accurate detection of the movement-related events in the signal is frequently necessary. I explored the application of change-point detection algorithm proposed by Moskvina et. al., 2003 to EMG event detection, and evaluated the technique’s performance comparing it to two common threshold-based event detection methods and to the visual estimates of the EMG events performed by trained practitioners in the field. The algorithm was implemented in MATLAB and applied to EMG segments recorded from wrist and trunk muscles. The quality and frequency of successful detection were assessed for all methods, using the average visual estimate as the baseline, against which techniques were evaluated. The application showed that the change-point detection can successfully locate multiple changes in the EMG signal, but the maximum value of the detection statistic did not always identify the muscle activation onset.

Manufacturing and construction workers undertake physically strenuous activities increasing the risk of health problems, disability, and sick leave, leading to lower job attractiveness and job candidate scarcity. In the EU, up to 44 million workers are affected by workplace-related musculoskeletal disorders (MSDs), representing a total annual cost of more than €240 billion. Exoskeleton use could alleviate muscle peak loads and reduce the risks of injury of workers. This work is related to the INTERREG's project "EXSCALLERATE" which aimed to accelerate the adoption of exoskeletons among SMEs. This research presents a comparative study of using exoskeletons by workers while performing different tasks related to their job. The tests evaluate the advantages of using exoskeletons in reducing human muscle activity, thereby, reducing the fatigue and tiredness. The study uses two commercially available exoskeletons, (1) upper body exoskeleton known as Eksovest and (2) lower body exoskeleton known as LegX. For upper body, the study performed drilling tasks at shoulder height and roof drilling positions, whereas, for the lower body, virtual chair position and squatting positions are tested which involved frequent bending of knees. Besides, the experiments based on accuracies of the data collection techniques and compare three volunteer’s body muscle data acquired by EMG sensor. From these comparisons, it is found that the muscle activity can be reduced up to 60% by using these exoskeletons, hence, increasing the work life of the workforce. The results of this study will help create awareness among SMEs towards the adoption of exoskeletons.

This paper demonstrates the implementation of R-CNN in terms of electromyography-related signals to recognize hand gestures. The signal acquisition is implemented using electrodes situated on the forearm, and the biomedical signals are generated to perform the signals preprocessing using wavelet packet transform to perform the feature extraction. The R-CNN methodology is used to map the specific features that are acquired from the wavelet power spectrum to validate and train how the architecture is framed. Additionally, the real-time test is completed to reach the accuracy of 96.48% compared to the related methods. This kind of result proves that the proposed work has the highest amount of accuracy in recognizing the gestures.