Evaluation of the EMG–force relationship of trunk muscles during whole body tilt

https://doi.org/10.1016/j.jbiomech.2007.09.008Get rights and content

Abstract

The study was aimed at the identification of the electromyography (EMG)–force relationship of five different trunk muscles. EMG–force relationships differ depending on changes in firing rate and the concurrent recruitment of motor units, which are linear and S-shaped, respectively. Trunk muscles are viewed as belonging to either the local or global muscle systems. Based on such assumptions, it would be expected that these functionally assigned muscles use different activation strategies.

Thirty-one healthy volunteers (16 women, 15 men) were investigated. Forces on the trunk were applied with the use of a device that gradually tilts the body to horizontal position. Rotation capability enabled investigation of forward and backward as well as right and left sideward tilt directions. Surface EMG (SEMG) of five trunk muscles was taken. Root mean square (rms) values were computed and relative amplitudes, according to the measured maximum amplitudes, were calculated individually.

Back muscles were characterized by a linear SEMG–force relationship during forward tilt. Abdominal muscles showed an S-shaped polynomial SEMG–force relationship for backward tilt direction. Sideward tilt directions evoked lesser SEMG levels with polynomial curve characteristics for all investigated muscles.

Therefore, the SEMG–force relationship possibly is also subject to force vector in relation to fiber direction.

Introduction

To maintain the necessary equilibrium of stability and mobility of the spine trunk muscle co-ordination has to be adequately organized (Gardner-Morse and Stokes, 1998; McGill et al., 2003). During the development of this functional view, functional characteristics were assigned to all trunk muscles. Primarily this assignment was biomechanically determined (Bergmark, 1989) based on whether these muscles insert directly at the spine (local system) or transfer forces between thorax and pelvis (global system). Within this system, local muscles are related to continuous, low-level activation, independent from movement to ensure stability (i.e. deep multifidus (MF) muscle). Global muscles act phasically at medium to high intensities to produce movements (i.e. erector spinae (ES) muscle). More recently, this classification was expanded and the assignment of functional characteristics of global muscles became more differentiated; global stabilizers limit range of motion and therefore mainly act eccentric (i.e. oblique abdominal muscles). Global mobilizers perform concentric contractions to initiate movements (Comerford and Mottram, 2001).

Corrupted trunk muscle co-ordination was found to be related to low back pain (Panjabi, 2003). Therefore, trunk muscle function has been investigated intensively over the last two decades (Hides et al., 1994; Hodges and Richardson, 1996; Kankaanpää et al., 1996; Leinonen et al., 2003; Magnusson et al., 1996), mostly by using surface electromyography (SEMG) techniques.

Any activation of muscles is functionally driven, resulting in random activations of motor units (MUs; Basmajian and De Luca, 1985). Therefore, the resulting SEMG signal has a stochastic characteristic, also called “noise-like interference pattern” (McGill, 2004). It is correlated with global SEMG parameters like root mean square (rms) and median frequency (Bonato, 2001) to name the most common ones. These parameters are correlated with muscle function characteristics (Luttmann et al., 1996) and, therefore, provide insight into muscle function in vivo. Common SEMG applications are directed at the quantification of muscle co-ordination (Brown et al., 2007; Hodges and Richardson, 1999; van Dieen et al., 1996), identification of muscular fatigue (Luttmann et al., 1996) and the estimation of strain levels according to previously defined reference levels (Attebrant et al., 1995; Doorenbosch et al., 2005).

The latter application is used to determine if a certain load induces the predicted strain level, or, which often is of even more importance, to determine strain levels during load situations for which the load level cannot be determined. To approximate muscle strain levels, reliable assumptions about SEMG amplitude–force relationships of the respective muscles are required. Early investigations already revealed differing SEMG amplitude–force relationships between muscles (Lawrence and De Luca, 1983). Further experiments could show a strong dependency of varying SEMG amplitude–force relationships from different recruitment strategies (Solomonow et al., 1990). A linear amplitude–force relationship was evident when all MUs were recruited up to 50% of maximum force and further increase of force resulted from increased firing frequency. If MUs together with increasing firing rates were concurrently recruited throughout the whole force range, a non-linear curve could be demonstrated (Solomonow et al., 1990).

A simulation approach (Fuglevand et al., 1993) including different maximum but fixed firing rates and recruitment ranges of all respective MUs revealed contrasting results; only if MUs were recruited over a broad force range, could a linear SEMG amplitude–force relationship be seen. If all MUs were recruited below 50% of maximum force, non-linear curve shapes were calculated, inconsistent with experimental findings. The authors could not output S-shaped curve characteristics.

Another known variable affecting the SEMG amplitude–force relationship is the muscle length–SEMG relationship (Anders et al., 2004) requiring identical muscle lengths during the normalization and the test situations to avoid false conclusions. Different positions of electrodes with respect to the endplate region also impact measured SEMG levels (Farina et al., 2002; Kleine et al., 2000).

According to the well-established assignment of trunk muscles to either the stabilizing or mobilizing muscle systems (Bergmark, 1989; Comerford and Mottram, 2001), we hypothesized each system would employ different strategies. Local muscles were expected to exhibit variation of firing frequency owing to their continuous activation characteristic (Comerford and Mottram, 2001) and, in contrast, mobilizing muscles were expected to additionally recruit more MUs to reach the high force levels necessary to limit as well as initiate movements (Comerford and Mottram, 2001).

According to what is known from literature, muscles of these two separate systems were expected to show distinct electromyography (EMG)–force relationships; a linear characteristic for local muscles and non-linear curves for the global muscles.

Section snippets

Methods

For this study, 31 healthy volunteers (16 women, mean age 23.1±4.9 S.D.; 15 men, mean age 25.7±4.7 S.D.) were investigated. All subjects volunteered after signing written informed consent. The investigation was performed in a device for trunk muscle diagnosis and treatment (Centaur®, BfMC, Leipzig, Germany; Fig. 1).

This device applies forces on the trunk by tilting the whole body from neutral upright position. Subjects are fixed at their feet, thighs and hips, but the trunk remains unsupported.

Forward/backward tilt

The SEMG amplitude increase with increasing torque differed considerably between abdominal and back muscles. With increasing torque, all abdominal muscles showed a non-linear amplitude increase. In contrast, the back muscles were characterized by a linear increase (Fig. 2). Since amplitude increase was similar for the abdominal muscles on the one hand and the back muscles on the other, SEMG data were pooled according to muscle localization at the front and back side of the body, respectively (

Discussion

All investigated muscles can be identified as either global mobilizing (RA, ES), global stabilizing (OI, OE; Comerford and Mottram, 2001) or local (MF; Bergmark, 1989). These muscle systems are considered to be different in function. Local muscles are permanently active, independent from movement, global stabilizers are characterized by eccentric activation in order to limit range of motion, and global mobilizers initiate movements, associated with concentric activation (Comerford and Mottram,

Conclusions

The recruitment strategies of abdominal and back muscles are different while counteracting backward and forward tilts. An ideal linear EMG–force relationship could be observed for back muscles, if loaded with extension forces. This correlation changed into a non-linear relationship if sideward flexion forces were applied. Abdominal muscles were characterized by a non-linear EMG–force relationship during application of flexion forces. For sideward flexion forces, this characteristic became less

Conflict of interest

Hereby, all authors declare that neither personal nor economic relationship exists with the manufacturer of the device and that they did not receive any sponsoring that could have influenced their work.

Acknowledgment

The authors wish to thank Mrs. Elke Mey for technical assistance and Ms. Marcie Matthews for language correction.

References (37)

  • S.M. McGill et al.

    Coordination of muscle activity to assure stability of the lumbar spine

    Journal of Electromyography and Kinesiology

    (2003)
  • M.M. Panjabi

    Clinical spinal instability and low back pain

    Journal of Electromyography and Kinesiology

    (2003)
  • C. Anders et al.

    Depending from muscle length identical muscular strain causes different muscular activation levels

    Physikalische Medizin, Rehabilitationsmedizin, Kurortmedizin

    (2004)
  • J.V. Basmajian et al.

    Muscles Alive

    (1985)
  • A. Bergmark

    Stability of the lumbar spine. A study in mechanical engineering

    Acta Orthopaedica Scandinavica

    (1989)
  • P. Bonato

    Recent advancements in the analysis of dynamic EMG data

    IEEE Engineering in Medicine and Biology Magazine (M-EMB)

    (2001)
  • A. Denner

    Analyse und Training der Wirbelsäulenstabilisierenden Muskulatur

    (1998)
  • D. Farina et al.

    Influence of anatomical, physical, and detection-system parameters on surface EMG

    Biological Cybernetics

    (2002)
  • Cited by (47)

    • Wireless sEMG-based identification in a virtual reality environment

      2019, Microelectronics Reliability
      Citation Excerpt :

      The MCI can apply the sEMG to identify the type of human motion and convert the sEMG signal into an input command. The sEMG is widely used in the myoelectricity signal detection, which is the first choice for the MCI because of its non-invasive, real-time and easy operation [14]. The sEMG has important significance in clinical diagnosis, rehabilitation medicine and intelligent prosthetic control [15–21].

    • Estimation of loads on human lumbar spine: A review of in vivo and computational model studies

      2016, Journal of Biomechanics
      Citation Excerpt :

      Electrical signals during muscle contractions are detected by surface electrodes on the skin overlying the muscles or by fine wire electrodes inserted deep into muscles. In order to estimate forces in trunk muscles, a relationship between the EMG activity of trunk muscles and their force is however needed (Anders et al., 2008; Cholewicki et al., 1995; Cholewicki and McGill, 1996; Disselhorst-Klug et al., 2009; Granata and Marras, 1995; McGill and Norman, 1986). While this approach is physiological in using EMG signals that account for inter- and intra-subject variability in recruitment, it fails, unlike optimization-driven methods, to satisfy crucial equilibrium equations.

    • Training-related changes in the EMG-moment relationship during isometric contractions: Further evidence of improved control of muscle activation in strength-trained men?

      2015, Journal of Electromyography and Kinesiology
      Citation Excerpt :

      In isometric conditions, previous studies reported linearity of the EMG–force/moment relationship (Bigland and Lippold, 1954; Hof, 1997; Lippold, 1952; Onishi et al., 2000; Pincivero and Coelho, 2000; Shinohara et al., 1998). It was also classified as quasi-linear, non-linear or quadratic (Basmajian and Deluca, 1985; Eloranta, 1989; Marras and Granata, 1997; Metral and Cassar, 1981; Watanabe and Akima, 2009), or even more complex: linear or non-linear depending on the range of force levels (Pincivero and Coelho, 2000; Monod and Flandrois, 2003) or on the force vector in relation to the direction of muscle fibre alignment (Anders et al., 2008). In dynamic conditions, the EMG–force/moment relationship follows either a linear (Bigland and Lippold, 1954; Herman and Bragin, 1967) or a non-linear (Häkkinen and Komi, 1983) relation, but its nature remains much less clear because of the influence of additional factors, such as force–length or force–velocity relationships (Herman and Bragin, 1967; Hof, 1997; Olney and Winter, 1985).

    View all citing articles on Scopus
    View full text