Decomposition of surface electrode-array electromyogram of biceps brachii muscle in voluntary isometric contraction

Abstract

Abstruct-The study of the electric signals generated by contracting muscle fibers (EMG) is relevant to neurophysiological research and to the diagnosis of motoneuron diseases. Classical diagnosis methods make use of invasive, painful needle electrodes. We explored a non-invasive alternative to these methods by developing a specific algorithm that uses Independent Component Analysis OCA) and template-matching techniques to decompose surface EMG (s-EMG). An experiment was carried out with two healthy subjects performing isometric contractions at different force levels. We measured s-EMGs with an electrode array and applied to them our decomposition algorithm. The obtained motor-unit firing pattern is in agreement with results obtained with needle electrodes. Kepordssurface-EMG, ICA, template matching

Decomposition of Synthetic Multi-channel Surface-Electromyogram Using Independent Component Analysis

“…This might be the reason why when using real s-EMG signals we do not get results as accurate as those obtained when using synthetic s-EMG [17]. For some of the independence measures it was necessary to estimate both joint and marginal probability density function (p.d.f.)…”

“…In order to measure the MUAP waveforms enhancement of a target MUAPT compared to the other MUAPTs obtained by applying ICA, we designed and applied to all the data obtained an additional measure; the peak-to-peak signal-to-interference ratio (SIR pp ) [17]. This measure was defined as the mean value of a MUAPT N firings peak in its closest channel (where it appeared the strongest) divided by the mean of the N strongest peaks belonging to the other MUAPTs present in the same channel.…”

“…We have recently developed an algorithm able to decompose eight-channel s-EMG signals into their constitutive MUAP trains (MUAPTs) at high contraction levels (up to 30 and even 60% of the subjects' maximum voluntary contraction -MVC) [17]. This algorithm is based on the application of signal-conditioning filters, Independent Component Analysis (ICA) [18][19][20], and a template-matching technique [21].…”

EMG Decomposition and Artefact Removal

“…If a rectangular window is used then Equation 3 is an LPD filter. Similarly to the LPD filter the WLPD is very easy to implement and considerably fast for real-time applications, but results presented in [17,46] show that phase distortion may be present in the filtered signal and depending on the level of noise false spikes may be generated.…”

“…In fact, when studying the state of deep muscles this is the only possible choice. However, many studies [5,11,17,43,49] have shown that surface EMG electrodes with a sufficiently small contact surface may be successfully applied to the detection of MUAP activity in superficial muscles.…”