Independent Component Analysis of High-Density Electromyography in Muscle Force Estimation

Abstract

Abstract-Accurate force prediction from surface electromyography (EMG) forms an important methodological challenge in biomechanics and kinesiology. In a previous study (Staudenmann et al., 2006), we illustrated force estimates based on analyses lent from multivariate statistics. In particular, we showed the advantages of principal component analysis (PCA) on monopolar high-density EMG (HD-EMG) over conventional electrode configurations. In the present study, we further improve force estimates by exploiting the correlation structure of the HD-EMG via independent component analysis (ICA). HD-EMG from the triceps brachii muscle and the extension force of the elbow were measured in 11 subjects. The root mean square difference (RMSD) and correlation coefficients between predicted and measured force were determined. Relative to using the monopolar EMG data, PCA yielded a 40% reduction in RMSD. ICA yielded a significant further reduction of up to 13% RMSD. Since ICA improved the PCA-based estimates, the independent structure of EMG signals appears to contain relevant additional information for the prediction of muscle force from surface HD-EMG.

Heterogeneity of muscle activation in relation to force direction: A multi-channel surface electromyography study on the triceps surae muscle

“…Note that some roll force occurred during this representative contraction. thus, common features (Staudenmann et al, 2006); the last few modes (20 ± 16) with very small eigenvalues only contained measurement noise (Staudenmann et al, 2007), (4) full-wave rectifying and low-pass filtering of the projected modes (Butterworth, bi-directional, cut-off 2 Hz) to obtain EMG-envelopes for each individual channel, (5) compensating for an overall electromechanical delay (EMD) between force and the sum of the EMGenvelopes (maximizing cross-correlation of the entire trapezoidal pattern), and, finally (6) estimating clusters using a K-means clustering algorithm (we defined distances via correlation and allowed for 100 search replications to reduce the risk of misinterpretation because of local minima). This algorithm not only allocates channels to the clusters, but also allows for projecting the data onto these clusters resulting in ''representative" time series per cluster (referred to as the centroid of a cluster's EMG).…”

Methodological aspects of SEMG recordings for force estimation – A tutorial and review

“…When the original EMG signals are transformed by PCA into ranked modes, the higher modes show an ongoing noise pattern, whereas the lower modes covary with force modulations (Staudenmann et al, 2007a(Staudenmann et al, , 2006. This approach could be further developed to remove noise from a set of EMG signals.…”

EMG–force modeling using parallel cascade identification

“…It is believed that lower time resolution results in more precise SEMG amplitude estimates (Staudenmann et al, 2007a). Decimating (down-sampling following lowpass filtering) was used to decrease time resolution and to reduce the number of data points thereby accelerating the training process of the PCI models.…”

“…Muscle force estimation using surface EMG (SEMG) has been widely researched in recent decades (Hof and Van Den Berg, 1981;Cholewicki and McGill, 1994;Lloyd and Besier, 2003;Staudenmann et al, 2007a). Active SEMG sensors are inexpensive and relatively small, and can be considered as a potential substitute for force sensors for muscle force estimation.…”