基于表面肌电信号的人体动作识别算法研究进展

表面肌电信号(Surface Electromyography,sEMG)是通过相应肌群表面的传感器记录下来的一维时间序列非平稳生物电信号,不但反映了神经肌肉系统活动,对于反映相应动作肢体活动信息同样重要。而模式识别是肌电应用领域的基础和关键。为了在应用基于表面肌电信号模式识别中选取合适算法,本文拟对基于表面肌电信号的人体动作识别算法进行回顾分析,主要包括模糊模式识别算法、线性判别分析算法、人工神经网络算法和支持向量机算法。模糊模式识别能自适应提取模糊规则,对初始化规则不敏感,适合处理s EMG这样具有严格不重复的生物电信号;线性判别分析对数据进行降维,计算简单,但不适合大数据;人工神经网络可以同时描述训练样本输入输出的线性关系和非线性映射关系,可以解决复杂的分类问题,学习能力强;支持向量机处理小样本、非线性的高维数据优势明显,计算速度快。比较各方法的优缺点,为今后处理此类问题模式识别算法选取提供了参考和依据。     

Extraction and optimization of exopolysaccharide from Lactobacillus sp. using response surface methodology and artificial neural networks

Abstract

The microbial polysaccharides secreted and produced from various microbes into their extracellular environment is known as exopolysaccharide. These polysaccharides can be secreted from the microbes either in a soluble or insoluble form.Lactobacillus sp. is one of the organisms that have been found to produce exopolysaccharide. Exo-polysaccharides (EPS) have various applications such as drug delivery, antimicrobial activity, surgical implants and many more in different fields. Medium composition is one of the major aspects for the production of EPS from Lactobacillus sp., optimization of medium components can help to enhance the synthesis of EPS . In the present work, the production of exopolysaccharide with different medium composition was optimized by response surface methodology (RSM) followed by tested for fitting with artificial neural networks (ANN). Three algorithms of ANN were compared to investigate the highest yeild of EPS. The highest yeild of EPS production in RSM was achieved by the medium composition that consists of (g/L) dextrose 15, sodium dihydrogen phosphate 3, potassium dihydrogen phosphate 2.5, triammonium citrate 1.5, and, magnesium sulfate 0.25. The output of 32 sets of RSM experiments were tested for fitting with ANN with three algorithms viz. Levenberg–Marquardt Algorithm (LMA), Bayesian Regularization Algorithm (BRA) and Scaled Conjugate Gradient Algorithm (SCGA) among them LMA found to have best fit with the experiments as compared to the SCGA and BRA.     

Adaptations to isolated shoulder fatigue during simulated repetitive work. Part I: Fatigue

Upper extremity muscle fatigue is challenging to identify during industrial tasks and places changing demands on the shoulder complex that are not fully understood. The purpose of this investigation was to examine adaptation strategies in response to isolated anterior deltoid muscle fatigue while performing simulated repetitive work. Participants completed two blocks of simulated repetitive work separated by an anterior deltoid fatigue protocol; the first block had 20 work cycles and the post-fatigue block had 60 cycles. Each work cycle was 60 s in duration and included 4 tasks: handle pull, cap rotation, drill press and handle push. Surface EMG of 14 muscles and upper body kinematics were recorded. Immediately following fatigue, glenohumeral flexion strength was reduced, rating of perceived exertion scores increased and signs of muscle fatigue (increased EMG amplitude, decreased EMG frequency) were present in anterior and posterior deltoids, latissimus dorsi and serratus anterior. Along with other kinematic and muscle activity changes, scapular reorientation occurred in all of the simulated tasks and generally served to increase the width of the subacromial space. These findings suggest that immediately following fatigue people adapt by repositioning joints to maintain task performance and may also prioritize maintaining subacromial space width.     

Effect of low frequency fatigue on human muscle strength and fatigability during subsequent stimulated activity

Fatiguing contractions of the adductor pollicis muscle were produced by intermittent supramaximal stimulation of the ulnar nerve in a set frequency pattern, in six normal subjects. At the end of an initial fatiguing contraction series, low frequency fatigue (LFF) had been induced and persisted at 15 min of recovery. Stimulated fatiguing activity was then repeated in an identical fashion to the initial series. At high frequencies, declines in force were similar for both series. At low frequencies, declines in force were greater during the second series despite similar changes in compound muscle action potential amplitude. This confirmation that LFF persists during subsequent stimulated activity, and reduces low but not high frequency fatigue resistance, suggests that the impaired endurance of fatigued muscle during voluntary activity primarily results from peripheral changes at low frequency. These findings also have implications for therapeutic electrical stimulation of muscle.     

Surface EMG signal processing during isometric contractions

This paper provides an overview of techniques suitable for the estimation, interpretation and understanding of time variations that affect the surface electromyographic (EMG) signal during sustained voluntary or electrically elicited contractions. These variations concern amplitude variables, spectral variables and muscle fiber conduction velocity, are interdependent and are referred to as the ‘fatigue plot'. The fatigue plot provides information suitable for the classification of muscle behavior. In addition, the information obtainable by means of linear electrode arrays is discussed, and applications of mathematical models for the interpretation of array signals are presented. The model approach provides additional ways for the classification of muscle behavior.     

Microcirculation in the upper trapezius muscle during sustained shoulder load in healthy women an endurance study using percutaneous laser-Doppler flowmetry and surface electromyography

Microcirculation in the upper portion of the trapezius muscle was measured percutaneously in a group of 16 healthy women of different ages by continuous laser-Doppler flowmetry (LDF) in relation to electromyography (EMG) during an endurance test. During the measurements the subject kept her arms straight and elevated at 45° in the scapular plane and held a 1-kg load in each hand as long as possible. This was followed by rest with the arms hanging and carrying no load. The 10-min recording period comprised 1-min initial rest followed by the endurance test and then recovery. Signal processing was done by computer on line. The LDF and root-mean-square (rms) EMG signals were normalized. Spectrum analyses of EMG mean power frequency (MPF) were performed. The amount of load produced was on average 2,267 (SD 939) N · m · s, i.e. shoulder torque × time expressed as Newton meter seconds, and the endurance time was 4.3 (SD 1.20) min. The rms-EMG as well as the LDF increased significantly during endurance, both when related to endurance time and to amount of load. The MPF showed no significant changes. The mean total increase in muscle blood flow was 175% of that recorded in the initial rest period. The average increase per each 10 s of contraction was 2.9%. Maximum was reached during the 1st min of recovery followed by a fall to the base level that was reached within 77 s on average. The amount of load produced and the blood flow increase was smaller than that found in a separate study of men, indicating a lower functional capacity. This may be of importance for the development of neck-shoulder disability in women.     

Factors affecting the turns analysis of the interference EMG signal

This study was performed to evaluate the relative significance of changes typical for muscle fatigue on quantitative parameters obtained from turns analysis of simulated intramuscular and surface interference electromyographic (EMG) signals. Effects of reduction of firing rate of motor units (MUs) and changes of intracellular action potential (IAP) profile along active fibers were analyzed. A new analytic function was proposed to simulate changes in IAP shape at different stages of muscle fatigue. In intramuscular EMG, both the decrease in firing rate of MUs and the changes in IAP profile led to reduction in the number of turns per second (NTs) and mean turn amplitude (MTA). The development of fatigue and especially the changes in IAP profile could explain why NTs increased up to only about 50% of maximal voluntary contraction, and remained unchanged above that level of efforts or even decreased. These effects should be especially pronounced in patients with myopathy whose IAP and muscle fatigability are expected to be abnormal. In surface EMG, the MTA increased considerably with fatigue; the sensitivity of NTs to reduction in firing rate (or number of discharges) was low. Thus, the benefits of the turns analysis of surface EMG signals should be lower not only in diagnosis of myopathy but also neuropathy.     

Feasibility of using combined EMG and kinematic signals for prosthesis control: A simulation study using a virtual reality environment

Transhumeral amputation has a significant effect on a person’s independence and quality of life. Myoelectric prostheses have the potential to restore upper limb function, however their use is currently limited due to lack of intuitive and natural control of multiple degrees of freedom. The goal of this study was to evaluate a novel transhumeral prosthesis controller that uses a combination of kinematic and electromyographic (EMG) signals recorded from the person’s proximal humerus. Specifically, we trained a time-delayed artificial neural network to predict elbow flexion/extension and forearm pronation/supination from six proximal EMG signals, and humeral angular velocity and linear acceleration. We evaluated this scheme with ten able-bodied subjects offline, as well as in a target-reaching task presented in an immersive virtual reality environment. The offline training had a target of 4° for flexion/extension and 8° for pronation/supination, which it easily exceeded (2.7° and 5.5° respectively). During online testing, all subjects completed the target-reaching task with path efficiency of 78% and minimal overshoot (1.5%). Thus, combining kinematic and muscle activity signals from the proximal humerus can provide adequate prosthesis control, and testing in a virtual reality environment can provide meaningful data on controller performance.     

EMG median power frequency in an exhausting exercise

EMG median power frequency of the calf muscles was investigated during an exhausting treadmill exercise. This exercise was an uphill run, the average endurance time was 1.5 min. Median power frequency of the calf muscles declined by more than 10% during this exercise. In addition EMG median power frequency of isometric contractions of the same muscles was measured before and in one minute intervals for 10 min after this run. Immediately after the run isometric median power frequency had declined by less than 5% for the soleus muscle, more than 10% for the gastrocnemius medialis and gastrocnemius lateralis muscles. In the 10 min following exercise the isometric median power frequency increased to pre-execise levels. Maybe the median power frequency shift to lower frequencies during dynamic exercise can be interpreted as a sign of local muscle fatigue.     

Cross-correlation of bilateral differences in fatigue during sustained maximal voluntary contraction

Maximal isometric force and electromyograph (EMG) activity of biceps brachii muscle during bilateral sustained elbow flexion were followed in 25 right-handed oarsmen. The percentage decline in force was greater for the left than for the right arm. Also, the mean power frequency (MPF) and the root mean square (rms) value of the EMG amplitude decreased more for the left than for the right arm. It was hypothesized that a common drive would indicate that the two forces curves would be highly correlated during the nonfatigued period, but the level of cross-correlation would decline during muscle fatigue. For the first 4 s of the contraction, the cross-correlation between the right and left force was high (r = 0.99), but thereafter it declined rapidly to a constant level. The decline of the cross-correlation was accompanied by a similar decrease in the correlation between the right and left EMG activations (MPF and rms). Thus, the decline in the cross-correlation level of force accompanied by a similar decrease in the correlation level of EMG would suggest a fatigue-induced neural derangement of the common drive.     

Modelling evolutionary cell behaviour using neural networks: Application to tumour growth

In this paper, we present a modelling framework for cellular evolution that is based on the notion that a cell’s behaviour is driven by interactions with other cells and its immediate environment. We equip each cell with a phenotype that determines its behaviour and implement a decision mechanism to allow evolution of this phenotype. This decision mechanism is modelled using feed-forward neural networks, which have been suggested as suitable models of cell signalling pathways. The environmental variables are presented as inputs to the network and result in a response that corresponds to the phenotype of the cell. The response of the network is determined by the network parameters, which are subject to mutations when the cells divide. This approach is versatile as there are no restrictions on what the input or output nodes represent, they can be chosen to represent any environmental variables and behaviours that are of importance to the cell population under consideration. This framework was implemented in an individual-based model of solid tumour growth in order to investigate the impact of the tissue oxygen concentration on the growth and evolutionary dynamics of the tumour. Our results show that the oxygen concentration affects the tumour at the morphological level, but more importantly has a direct impact on the evolutionary dynamics. When the supply of oxygen is limited we observe a faster divergence away from the initial genotype, a higher population diversity and faster evolution towards aggressive phenotypes. The implementation of this framework suggests that this approach is well suited for modelling systems where evolution plays an important role and where a changing environment exerts selection pressure on the evolving population.     

Investigating the relationship between proteomic,compositional, and histologic biomarkers and cartilage biomechanics using artificial neural networks

A thorough understanding of the relationship between the biological and mechanical functions of articular cartilage is necessary to develop diagnostics and treatments for arthritic diseases. A key step in developing this understanding is the establishment of models which utilize large numbers of biomarkers to create comprehensive models of the interplay between cartilage biology and biomechanics, which will more accurately demonstrate the complex etiology and progression of tissue adaptation and degradation. It is the goal of this study to demonstrate the ability of artificial neural networks (ANNs) to utilize biomarkers to create predictive models of articular cartilage biomechanics, which will provide a basis for more sophisticated research in the future. Osteochondral plugs were collected from patients undergoing total knee arthroplasty, cultured, then analyzed to collect proteomic, compositional, and histologic biomarker data. Samples were subjected to stress relaxation testing as well as computational simulations using finite element analysis (FEA) modeling and optimization to determine key mechanical properties. The acquired data was fed into an ANN to generate a model which predicts the biomechanical properties of cartilage from given biomarkers. Using all significant inputs, the developed neural network predicted the ground substance modulus with a moderate degree of accuracy, but had difficulty predicting the collagen fiber modulus and cartilage permeability. Using only clinically attainable biomarkers, the best-performing model produced comparably accurate and more consistent predictions of all three mechanical properties. These models demonstrate the potential for ANNs to be included in clinical studies of articular cartilage.     

EMG-based muscle fatigue assessment during dynamic contractions using principal component analysis

A novel approach to fatigue assessment during dynamic contractions was proposed which projected multiple surface myoelectric parameters onto the vector connecting the temporal start and end points in feature-space in order to extract the long-term trend information. The proposed end to end (ETE) projection was compared to traditional principal component analysis (PCA) as well as neural-network implementations of linear (LPCA) and non-linear PCA (NLPCA). Nine healthy participants completed two repetitions of fatigue tests during isometric, cyclic and random fatiguing contractions of the biceps brachii. The fatigue assessments were evaluated in terms of a modified sensitivity to variability ratio (SVR) and each method used a set of time-domain and frequency-domain features which maximized the SVR. It was shown that there was no statistical difference among ETE, PCA and LPCA (p > 0.99) and that all three outperformed NLPCA (p < 0.0022). Future work will include a broader comparison of these methods to other new and established fatigue indices.     

Reliability of muscle fibre conduction velocity distribution estimation from surface EMG

Most of the neuromuscular diseases induce changes in muscle fibre characteristics. For example, Duchenne dystrophy is characterized by a specific loss of fast fibres, and an increase in small diameter fibres. These morphological changes may lead to large modifications in the distribution of fibre diameters, possibly producing bimodal distributions. It has already been shown that it is possible to reveal these morphological modifications through the distribution of muscle fibre conduction velocity (MFCV) as estimated from needle electromyography (EMG) recordings. In this paper, we investigate whether such changes can be extracted from surface EMG signals.

Simulation allows generation of surface EMG signals in which features are well described especially at a morphological level. Therefore, we generated a database of simulated signals both in voluntary and electrically elicited contraction conditions using a bimodal distribution of muscle fibre diameters. MFCV distributions were computed using two short-term methods based on cross-correlation and peak-to-peak techniques for voluntary contraction signals, and using a deconvolution method in time domain for electrically elicited signals. MFCV distributions were compared with true ones, as generated from modelling.

This work reveals that estimating MFCV distribution through these methods does not appear yet as precise and robust enough to accurately characterize changes in redistribution of various muscle fibre diameters. However, it opens to new experimental protocols that can be explored in order to improve the robustness of MFCV distribution estimation for the follow-up of patients suffering from neuromuscular disorders.     

Estimation of muscle fiber conduction velocity from two-dimensional surface EMG recordings in dynamic tasks

The aims of this study are (1) to demonstrate that multi-channel surface electromyographic (EMG) signals can be detected with negligible artifacts during fast dynamic movements with an adhesive two-dimensional (2D) grid of 64 electrodes and (2) to propose a new method for the estimation of muscle fiber conduction velocity from short epochs of 2D EMG recordings during dynamic tasks. Surface EMG signals were collected from the biceps brachii muscle of four subjects with a grid of 13 × 5 electrodes during horizontal elbow flexion/extension movements (range 120–170°) at the maximum speed, repeated cyclically for 2 min. Action potentials propagating between the innervation zone and tendon regions could be detected during the dynamic task. A maximum likelihood method for conduction velocity estimation from the 2D grid using short time intervals was developed and applied to the experimental signals. The accuracy of conduction velocity estimation, assessed from the standard deviation of the residual of the regression line with respect to time, decreased from (range) 0.20–0.33 m/s using one column to 0.02–0.15 m/s when combining five columns of the electrode grid. This novel method for estimation of muscle fiber conduction velocity from 2D EMG recordings provides an estimate which is global in space and local in time, thus representative of the entire muscle yet able to track fast changes over the execution of a task, as is required for assessing muscle properties during fast movements.     

Sex-specific myoelectric manifestations of localized fatigue during a multi-joint repetitive task

We have previously demonstrated that fatigue at different locations impacts joint angles, angular variability, and coordination variability differently. However, the neuromuscular control aspects underlying these kinematic changes have never been demonstrated. Seventeen young adults (8 males) were recruited. Electromyographic electrodes were placed on: upper trapezius, pectoralis major, anterior and middle deltoid, biceps and triceps brachii, and left and right erector spinae. Subjects performed the repetitive pointing task (RPT) at 1 Hz for 30 s before and after localized fatigue tasks, which consisted of one shoulder, one elbow and one lower back isometric fatiguing protocols until exhaustion in randomized order. Electromyographic amplitude (RMS), variability (SD) and mean power frequency (MnPF) were calculated for each of the pre-fatigue and post-fatigue RPT trials. There were sex × fatigue location interaction effects on upper trapezius RMS (p = 0.038) with males’ values increasing the most after shoulder fatigue. Females’ triceps brachii RMS was greater compared to males after shoulder, elbow, and trunk fatigue (p = 0.003, p = 0.001 and p = 0.007 respectively). There were sex × fatigue location effects on left erector spinae MnPF (p = 0.011) with males and females’ values decreasing the most after trunk fatigue, but more so in males. Results demonstrate that males and females compensate differently during a repetitive pointing task when their elbows, shoulders and trunks are locally fatigued, which could have implications on sex-specific workplace injury risks. See Table 1 for acronyms.