Improved evaluation of back muscle SEMG characteristics by modelling
Introduction
Surface EMG (SEMG) is a valuable tool in the functional studies of movement co-ordination. Because this method is non-invasive and painless, it avoids an irritation of the muscles and therefore it does not influence the activation patterns. Furthermore, it is possible to apply a grid or matrix of SEMG electrodes, denoted as multi-channel SEMG. In addition to the neurological applications of multi-channel SEMG (see, e.g. [1], [2]), the interpolation of the SEMG power gives information about intra- and inter-muscular co-ordination [3]. It has been shown that SEMG maps of low back pain patients and healthy subjects differ [4]. Root mean square (RMS) is typically used to roughly estimate muscle force and the activity distribution. A more detailed interpretation of SEMG is complicated due to volume conduction of muscle tissue, fat, and skin decreasing the spatial and temporal resolution of the surface EMG signal. Spatial (e.g. [5], [6]) or temporal [7] filtering is able to increase the spatial resolution again.
If a row of SEMG channels is located parallel to the muscle fibre direction, the propagation of motor unit action potentials (MUAP) should cause time shifts between signals of these channels (Fig. 1), which can be estimated from the distance between channels and the muscle fibre potential conduction velocity (MFCV, about 4–6 m/s). These time shifts can be investigated by means of different methods. The action potential propagation can be observed directly (see, e.g. [8]) as it causes both time shifts in the cross covariance function (see, e.g. [9], [7]) and the linear behaviour of the cross phase calculated between these channels [10]. In back muscles MUAP propagation has been shown in the m. latissimus [10]. Similar experiments for the m. erector spinae do not demonstrate such propagation of action potentials, and the cross phase has no systematic phase behaviour [11]. Also, Farina et al. [12] found experimentally that the signal is dominated by the non-propagating components; propagating action potentials can be observed, but they are rare. In other words, the erector muscle seems to show a more complicated behaviour. To investigate this phenomenon further, we applied in the present study the method of high pass filtering of cross covariance functions [7] to SEMG data from the back region.
Section snippets
Subjects and methods
Surface EMG has been recorded from 16 subjects during a static isometric contraction of their back muscles. The location of electrodes is shown in Fig. 2. The subjects performed two exercises: one exercise was an unsupported horizontal trunk extension in prone position [13] to evoke the isolated activation of the m. erector spinae. For this investigation of the erector muscle a row of 16 monopolar surface electrodes was applied along a line parallel to the spinal column, on the belly of m.
Results and discussion
The correlation peaks from the m. latissimus showed typical time shifts indicating action potential propagation (see Fig. 3A). Occasionally different populations of motor units could be distinguished by means of their discrete endplate positions (Fig. 3B), similar to the m. biceps brachii [7]. The fact that the population of motor units with neuromuscular junctions between channels 4 and 5 was surpassed by the other EMG components revealed some information about the position of the sources:
Modelling
The anatomical structure of the m. erector spinae is very complex (see, e.g. [14]). The muscle is covered with an aponeurosis (ap in Fig. 6A). This is the beginning of an outer layer of muscle fibres (FB) which spirals around some muscle bundles (see Fig. 6B). These muscle bundles have their own complex structure. They contain f. i. mm. multifidii et rotatores connecting first, second, and third neighbouring vertebras.
This structure can be described by simplified models; we assume finite muscle
Simulation results and discussion
The cross covariance functions from model A-D (deep fibres) showed correlation peaks with a slow spatial sign reversal and without time shifts which would indicate the propagation of action potentials. Such behaviour is independent from the position of the reference channel. Cross covariance functions from the other models have different characteristics. Model A-S (superficial fibres) shows covariance peaks with a time shift which can be explained by action potential propagation (Fig. 9). The
Conclusion
Surface EMG signals from the back extensors show no signs of action potential propagation (or only rarely). This behaviour, which is known from the literature and is also found in correlation functions from our data, can be explained by a model assuming short, deep muscle fibres having end effects according to Dumitru et al. [15], and with overlapping positions parallel to the fibre direction. The region of very deep fibres begins at a depth of 1/3 of half the fibre length. This condition is
References (17)
- et al.
Estimation of the relationship between the noninvasively detected activity of single motor units and their characteristic pathological changes by modelling
J. Electromyogr. Kinesiol.
(1998) - et al.
Spectral EMG mapping of human masseter under different functional conditions: a topographically oriented EMG analysis in healthy subjects
Arch. Oral Biol.
(1994) - et al.
Simulation analysis of the ability of different types of multi-electrodes to increase selectivity of detection and to reduce cross-talk
J. Electromyogr. Kinesiol.
(2003) - et al.
Selective spatial information from surface EMG after temporal filtering. The application to interference EMG using cross covariance analysis
Clin. Neurophysiol.
(2003) - et al.
Assessment of low back muscle fatigue by surface EMG signal analysis: methodological aspects
J. Electromyogr. Kinesiol.
(2003) - et al.
Multichannel surface EMG: basic aspects and clinical utility
Muscle Nerve
(2003) - et al.
Quantitative topographic and temporal characterization of myoelectrical activation pattern—new diagnostic possibilities in neurology, physiotherapy and orthopaedics
Funct. Neurol.
(1994) - et al.
Spatial filtering of noninvasive multielectrode EMG: Part I—introduction to measuring techniques and applications. Part II—filter performance in theory and modelling
IEEE Trans. Biomed. Eng.
(1987)
Cited by (3)
Peculiarities of extracellular potentials produced by deep muscles. Part 2: Motor unit potentials
2013, Medical and Biological Engineering and ComputingPeculiarities of extracellular potentials produced by deep muscles. Part 1: Single fibre potential fields
2013, Medical and Biological Engineering and ComputingDose response effect of cement dust on respiratory muscles competence in cement mill workers
2006, International Journal of Environmental Health Research