Measurement and Estimation of Muscle Contraction Strength Using Mechanomyography Based on Artificial Neural Network Algorithm

Abstract

Muscle contraction strength estimation using mechanomyographic (MMG) signal is typically calculated by the root mean square (RMS) amplitude. Raw MMG signal is processed by rectification, low-pass filtering, and mapping. In this work, beside RMS amplitude, another significant parameter of MMG signal, i.e. frequency variance (VAR), is introduced and used for constructing an algorithm for estimating the muscle contraction strength. Seven participants produced isometric contractions about the elbow while MMG signal and generated torque (resultant of muscle contraction strength) of biceps brachii were recorded. We found that MMG RMS increased monotonously and VAR decreased under incremental voluntary contractions. Based on these results, a two-layer neural network was utilized for the model of estimating the muscle contraction strength from MMG RMS and VAR. Experimental evaluation was performed under constant posture and sinusoidal contractions at 0.5 Hz, 0.25 Hz, 0.125 Hz, and random frequency. The results of the proposed algorithm and MMG RMS linear mapping were also compared. The proposed algorithm has better accuracy than linear mapping for all contraction frequencies. The mean absolute error decreased 6% for the 0.5Hz contraction, 43% for 0.25 Hz contraction, 52% for 0.125 Hz contraction, and 30% for random frequency contraction.

Estimation of Knee Extension Force Using Mechanomyography Signals Based on GRA and ICS-SVR

“…In particular, compared with BPNN, ELM, and CS-SVR models, the estimated results of the proposed model had less fluctuation. Meanwhile, in terms of numerical results, the accuracy of the muscle force estimation of the proposed model in this paper was optimal compared with the results of the literature [25][26][27][28][29]. Furthermore, it is shown that our proposed model had unparalleled applicability and prediction accuracy in knee joint extension force estimation.…”

“…Additionally, since MMG signals are mechanical signals, their detection process is also not affected by the change in skin impedance [21]. Compared with sEMG, MMG signals have unparalleled advantages, have attracted extensive attention from many researchers in recent years, and have also been applied in many fields, such as prosthetic hand control [9], muscle fatigue assessment [22], human movement recognition [13], and human kinetic parameter estimation [23][24][25][26][27]. However, because of the characteristics of weak signal strength, low frequency, and strong randomness, it also poses a challenge to decode MMG signals and use them to estimate and predict muscle force.…”

“…With further research, the researchers found the presence of a nonlinear relationship between MMG and muscle force, indicating that MMG can be used to estimate muscle force. Therefore, subsequent researchers also started to extract time-domain features and frequency-domain features from MMG signals and fed them into a machine learning model to achieve a nonlinear MMG-force relationship [25][26][27]. Although some achievements have been made in muscle force estimation, there are still some drawbacks such as fewer features considered and lower accuracy of the estimated models.…”

Advances and perspectives of mechanomyography

“…Once the fresh muscle starts to fatigue, new muscle fiber recruitment occurs ( Lei et al (2013) with the experimental set up. Through an artificial neural network, the author correlated the force response with the MMG system output.…”

“…The results showed that same-subject validation tests were significantly greater than those of the cross-subject validation tests. Lei et al (2013) estimated the muscular contraction strength using an MMG-based ANN (Figure 2). Their experimental set up consisted of: (i) dynamometer to monitor the torque, (ii) MMG system, (iii) display for biofeedback, and (iv) computer with a display output to acquire the torque and MMG signals.…”