Journal of Biomechanics 48 (2015) 1198–1205
Contents lists available at ScienceDirect
Journal of Biomechanics journal homepage: www.elsevier.com/locate/jbiomech www.JBiomech.com
Short communication
Re-evaluation of the amplitude–force relationship of trunk muscles Agnes Huebner n, Bernd Faenger, Hans-Christoph Scholle, Christoph Anders Clinic for Trauma-, Hand- and Reconstructive Surgery, Division for Motor Research, Pathophysiology and Biomechanics, Jena University Hospital, Bachstrasse 18, 07743 Jena, Germany
art ic l e i nf o
a b s t r a c t
Article history: Accepted 15 February 2015
Amplitude–force relationships of major trunk muscles are established in terms of curve characteristics, but up to now were not normalized with respect to maximum voluntary contraction (MVC) force levels. The study therefore aims at a re-evaluation of trunk muscle amplitude–force relationship data according to MVC. Surface EMG of five major trunk muscles was taken from 50 healthy subjects of both sexes (age 20–40 years). All tasks were performed in a device where submaximal loads on the trunk were applied by gradually tilting the subjects in sagittal plane to horizontal position. MVC flexion and extension forces were determined in upright position using an additional harness over the subject's shoulders. Furthermore, the subject's upper body mass (UBM) was obtained during forward tilt to horizontal. MVC to UBM ratio was calculated, corrected by the actual tilt angle, and these linearly estimated values compared with the measured relative values according to MVC. All abdominal muscles confirmed the known non-linear amplitude–force relationship. At low load levels the linearly estimated values overestimated the measured ones and, at higher load levels, underestimated the true stress levels considerably. Back muscles confirmed the known linear curve shape, but for the longissimus muscle at L1 level measured data was always below estimated values. With increasing load, muscular stress of abdominal muscles changes from overestimated towards considerably underestimated values if expected stress levels are based on linear interpolation. Major back muscles' activation levels are nearly linear, but the amplitude–force relationship values seem overestimated for longissimus. & 2015 Elsevier Ltd. All rights reserved.
Keywords: Surface electromyography (SEMG) Trunk muscles Maximum voluntary contraction (MVC) Amplitude–force relationship Muscle stress level
1. Introduction Trunk muscles are already known to show a localization specific Surface EMG (SEMG) amplitude–force relationship that is non-linear for abdominal muscles and nearly linear for back muscles (Anders et al., 2008). These published results did not include normalization procedures, so up to now amplitude–force data could not be evaluated according to any commonly accepted reference point. In place of normalization procedures, especially for isometric situations, the assessment according to maximum voluntary contraction force (MVC) still serves as the gold standard despite its known drawbacks: it is subject to motivation (McNair et al., 1996) and practice (Bernardi et al., 1996); in the case of investigating many muscles requires extreme effort (Kendall et al., 2005); and moreover, it has limited applicability in patients (O'Sullivan et al., 1997; Roy and Oddsson, 1998). Despite these limitations, in healthy people MVC trials are still preferred to estimate submaximal stress levels (Clarys, 2000).
n
Corresponding author. Tel.: þ 49 3641 934 094; fax: þ 49 3641 934 091. E-mail address: [email protected] (A. Huebner).
http://dx.doi.org/10.1016/j.jbiomech.2015.02.016 0021-9290/& 2015 Elsevier Ltd. All rights reserved.
Normally in SEMG applications, estimates of muscle stress are usually interpolated linearly. However, as mentioned above, for the abdominals a non-linear amplitude–force relationship has been established, preventing credible estimations based on this method. But still the question remains unanswered as to the extent and direction of the assumable deviations between actual data and estimated values: are data permanently misjudged in one direction or do they possibly intersect the estimated ones? This investigation aims at a re-evaluation of the known SEMG amplitude–force relationship of trunk muscles by normalization of amplitudes based on SEMG values during isometric MVC tasks.
2. Materials and methods 2.1. Experimental design For this study 25 healthy subjects of each sex aged between 20 and 40 years (men: mean 29SD77.3, women: mean 26SD76.1) were investigated. Subjects' health status was verified by physical examination and self-reported history. The study followed the ethics requirements for human investigations and was positively
A. Huebner et al. / Journal of Biomechanics 48 (2015) 1198–1205 evaluated by the local ethics committee (2643-08/09). Written informed consent was obtained from every subject. The investigation was performed in a device for trunk muscle diagnosis and training (CTT Centaurs, BfMC, Germany) where subjects' lower body was fixed at thighs and hips while their trunk remained unsupported (Fig. 1). In its common use the device applies submaximal forces on the trunk by tilting the whole body from neutral upright position. During these submaximal tests the subject's task is to stabilize the upper body against gravitational forces while remaining along body axis. Monitoring of upper body position is facilitated by a harness (Fig. 1), positioned over the subject's shoulders and equipped with strain gauges which are used for biofeedback via a small monitor in front of the subject. By this, virtually isometric conditions with invariant muscle positions can be assumed. Forward and backward tilts at 51, 101, 201, 301, 451, 601 and 901 (i.e. 9%, 17%, 34%, 50%, 71%, 87%, 100% of upper body mass (UBM)) were applied. For this study subjects were tilted in sagittal plane only. The different angles and tilt directions were performed in a randomized order to prevent fatigue and order dependent effects. Prior to the submaximal tests, flexion and extension MVC tests were conducted while subjects were first positioned in the device. Held vertically, they were asked to exhibit maximum force against the harness. Subjects were given a habituation trial, followed by three MVC trials, each of five seconds duration with 30 s pause. During task performance subjects held their arms crossed against their chest to avoid effects due to varying arm position.
1199
Individual UBM was determined by tilting the subjects forward to horizontal position while being consciously relaxed: SEMG of the back muscles was measured and residual contractions announced to the subject for correction. To anthropometrically normalize the measured MVC levels a MVC to UBM ratio was determined.
2.2. SEMG data acquisition and analysis SEMG was taken from the following five trunk muscles of both sides: rectus abdominis muscle (RA), obliquus internus abdominis muscle (OI), obliquus externus abdominis muscle (OE), multifidus muscle (lumbar part, at L5 vertebral level, MF) and erector spinae muscle (longissimus, at L1 vertebral level, LO) (Fig. 2). Electrode positions were chosen in accordance with the SENIAM recommendations (Hermens et al., 1999), and, if not available for the respective muscle, according to Ng et al. (1998). We also measured the electrocardiographical (ECG) activation by placing one electrode pair along the heart axis. For SEMG measurements disposable Ag–AgCl electrodes (H93SG, Covidien, Germany) with a circular uptake area of 1.6 cm diameter were used. SEMG was amplified (gain: 1000, Biovision, Germany) using a bipolar montage (inter-electrode distance: 2.5 cm). The signals were analog to digital converted at a rate of 2048/s (Tower of Measurement, DeMeTec, Germany; anti-aliasing filter: 1024 Hz, resolution: 24 bit, software: GJB, Germany). All data was processed equally: elimination of DC components, high-pass filter at 10 Hz, low-pass filter at 300 Hz, and notch filtering at 50 Hz. Elimination of ECGartefacts was accomplished by only analyzing stationary signal sections in a time window of 400 ms starting 100 ms after each detected R wave of the ECG channel (Anders et al., 1991). Except for the MVC tasks where only about three seconds of maximal activity were averaged, the 10 s submaximal situations contained between nine and 15 ECG events. By averaging all respective signal sections a representative RMS value was calculated for every channel and task, respectively. Since the tasks were performed in sagittal plane RMS data from both sides were pooled, resulting in five single mean amplitude values per task. The pooled RMS data were normalized according to the corresponding values during MVC testing, further referred to as measured values. Considering the MVC to UBM ratio, expected SEMG amplitudes were calculated, corrected for the actual tilt angle by applying the sinus function. This procedure resulted in values following a linear trend, further referred to as estimated values. These estimated values were then compared with the measured values, separately for both sexes. Data variability of SEMG data was characterized by an unequal distribution of quartile ranges with less variance in lower quartiles. Hence presented data are specified through their median values and corresponding quartile ranges. Accordingly, non-parametric statistics were applied. In Section 3 tasks will be displayed that directly provoked forces on the respective muscles, i.e. abdominal muscles during backward tasks and back muscles during forward tasks. Furthermore antagonistic muscle activity will be displayed as well to uncover possible influences on the data by involuntary co-contractions, i.e. abdominal muscle activity during forward tilts and back muscle activity during backward tilts.
2.3. Statistical analysis
Fig. 1. Subject performing 301 forward tilt in the device for trunk muscle diagnosis and training (CTT Centaurs).
Prior to pooling of RMS values an ANOVA was calculated to ensure that no significant differences between body sides occurred. Paired student's t-tests were conducted on the force data to determine differences between tilt directions and unpaired student's t-tests to determine differences between sexes. Since SEMG data was not normally distributed, the non-parametric Wilcoxon test for dependent samples was applied for statistical comparison of measured and estimated relative SEMG values. For tests of SEMG data between sexes the non-parametric Mann–Whitney-U-test for independent samples was conducted (Edgeworth approximation).
ECG
rectus abdominis muscle erector spinae muscle (longissimus) obliquus externus abdominis muscle multifidus muscle (lumbar part) obliquus internus abdominis muscle Fig. 2. Electrode positions of investigated trunk muscles.
1200
A. Huebner et al. / Journal of Biomechanics 48 (2015) 1198–1205
Table 1 Ratio of maximal voluntary contraction force (MVC) to upper body mass (UBM) for extension (e) as well as flexion (f) direction and the respective MVC/UBM ratios. Student's t-test results for ratios within one sex as well as between sexes po 0.05. Female
MW SD t-test e/f t-test sex
Male
MVC/UBM extension
MVC/UBM flexion
e/f ratio
MVC/UBM extension
MVC/UBM flexion
e/f ratio
1.98 0.37 o 0.001 o 0.001
1.48 0.39
1.33 0.42
2.37 0.37 o0.001
1.94 0.40
1.23 0.40
o 0.001
0.17
t-test e/f: paired t-test. t-test sex: unpaired t-test.
force direction
100
y = 0.0092x 2 + 0.0616x + 0.7224 R² = 0.9963 y = 0.6586x + 1.4323 R² = 0.9998
50
150 rel. MVC [%]
RA
rel. MVC [%]
150
antagonistic force direction
100 50
0
0
90
y = 0.0034x2 + 0.3902x + 0.5525 R² = 0.9955
60 y = 0.6324x + 3.9212 R² = 0.9996
30
100 rel. MVC [%]
OI
rel. MVC [%]
50
0
60 30 0
50
100
rel. UBM [%]
90
y = 0.0021x2 + 0.5591x + 0.6325 R² = 0.9985
60 y = 0.647x + 2.3587 R² = 0.9995
30
rel. MVC [%]
rel. MVC [%]
0
0
50 y = 0.0003x 2 + 0.3664x + 6.9289 R² = 0.9877
40
y = 0.4746x + 2.8603 R² = 0.9996
20
rel. MVC [%]
rel. MVC [%]
rel. UBM [%]
60 30
100 60
60
0
50
100
rel. UBM [%]
40 20
0
0
60
50 y=
0.0004x2 +
rel. UBM [%] 0.2565x + 9.5056
R² = 0.9968
40
y = 0.4597x + 4.6144 R² = 0.999
20
100 0 60 rel. MVC [%]
0 rel. MVC [%]
100
0 0
LO
50
90
0
MF
100
rel. UBM [%]
90
0
OE
50
50
100
rel. UBM [%]
40 20
0
0 0
50 rel. UBM [%]
estimated values
100
0
50
100
rel. UBM [%]
measured values
Fig. 3. Women's measured and linearly estimated relative amplitude values according to MVC in force direction (left column). Open circles indicate significant differences between both measured and estimated values p o 0.05 (Wilcoxon–Mann–Whitney-Test). Right column: women's measured amplitude values in antagonistic force direction.
A. Huebner et al. / Journal of Biomechanics 48 (2015) 1198–1205
antagonistic force direction
600
100
500
80
rel. MVC [%]
RA
rel. MVC [%]
force direction
400 300 200
60 40 20
100 0
0 0
20
40
60
80
100
0
120 80 40
20
40
60
80
80
100
80
100
80
100
80
100
rel. UBM
0
20
160
rel. UBM
120 80 40
40
60
rel. UBM
120 80 40 0
0
20
100
40
60
80
100
0
20
100
rel. UBM
80 60 40
40
60
rel. UBM
80
rel. MVC [%]
rel. MVC [%]
100
40
0
60 40
20
20
0
0 0
20
100
40
60
80
100
0
20
100
rel. UBM
80
rel. MVC [%]
rel. MVC [%]
80
80
100
rel. MVC [%]
rel. MVC [%]
160
LO
60
0 0
MF
40
120
0
OE
20
160
rel. UBM rel. MVC [%]
rel. MVC [%]
160
OI
1201
60 40 20
40
60
rel. UBM
80 60 40 20
0
0 0
20
40
60
80
100
rel. UBM
0
20
40
60
rel. UBM Fig. 3. (continued)
3. Results
3.2. Abdominal muscles
3.1. Force data
For all abdominal muscles the known non-linear amplitude– force relationship could be confirmed by the measured data (upper three rows in the left columns of Figs. 3 and 4). Compared to estimated values, at low load levels measured amplitudes were significantly lower, but as load levels increased measured amplitudes gradually reached higher values than estimated. Curve shapes of measured data are quite similar between muscles and sexes, but
Significant differences occur for MVC to UBM ratios between extension and flexion tasks with higher values during extension. Likewise the respective extension and flexion ratios differ between sexes, but the extension to flexion ratio does not show any sex differences (Table 1).
1202
A. Huebner et al. / Journal of Biomechanics 48 (2015) 1198–1205
i.e. horizontal position, is due to large variances within sexes. In contrast, for OI and OE no systematic differences in deviation from estimated values occur between sexes.
quantitative differences exist: for both sexes the intersections where abdominal muscle amplitudes equal the estimated values occur at loads larger than 50% UBM, in men these intersections occur at even higher load but almost identical stress levels. According to the differences in force capacity, estimated as well as measured values in men are always lower than in women. Remarkably, at 100% UBM (i.e. horizontal position) women's measured SEMG levels of RA exceeded even those obtained during MVC testing. During forward tilt, i.e. at antagonistic force direction, insignificant and virtually unchanged abdominal contraction levels occurred (upper three rows in the right columns of Figs. 3 and 4). Amplitude levels of men were again lower than in women. In Fig. 5 the deviations from the estimated values (i.e. in force direction) are quantified and sex differences are analyzed: only for RA significant differences between sexes are verifiable and apply to medium loads and up. The absence of significance at 100% UBM,
3.3. Back muscles Both back muscles confirmed the known linear amplitude–force relationship (lower two rows in the left columns of Figs. 3 and 4). For MF only slight and unsystematic differences occur between measured and estimated values. For LO the measured amplitude values are continuously well below the estimated levels, even more so in men. Significant differences between measured and estimated curves occur at almost all load levels, due to different curve slopes. As in the abdominal muscles, the right columns of Figs. 3 and 4 show data of the antagonistic force direction of women and men to
force direction y = 0.0098x2 – 0.3425x + 4.2701 R² = 0.9878
100
y = 0.5257x + 0.6033 R² = 0.9999
50
150 rel. MVC [%]
RA
rel. MVC [%]
150
antagonistic force direction
100 50
0
0 50 rel. UBM [%] y = 0.0063x 2 – 0.0169x + 1.3888 R² = 0.9973
90 60
y = 0.5175x + 1.3861 R² = 0.9998
30
100
30
50
rel. UBM [%] y = 0.0031x2 + 0.3133x –0.8581 R² = 0.997 y = 0.5244x + 0.6904 R² = 0.9998
30
100
30
50 rel. UBM [%] y = -0.0005x 2 + 0.3767x + 7.6736 R² = 0.9913 y = 0.3961x + 2.6892 R² = 0.9981
40 20
100 60 rel. MVC [%]
rel. MVC [%]
60
0
0
50
100
rel. UBM [%]
40 20 0
60
50 rel. UBM+[%] y = 8E-05x 2 + 0.2207x 8.2081 R² = 0.9973
40
y = 0.3726x + 5.3857 R² = 0.999
20
100 0 60 rel. MVC [%]
0 rel. MVC [%]
100
0 0
LO
50 rel. UBM [%]
60
0
MF
0
90
rel. MVC [%]
rel. MVC [%]
60
100
0 0
OE
50 rel. UBM [%]
60
0
90
0
90
rel. MVC [%]
OI
rel. MVC [%]
0
0 0
50 rel. UBM [%]
estimated values
50
100
rel. UBM [%]
40 20
0 100 0
50
100
rel. UBM [%]
measured values
Fig. 4. Men's measured and linearly estimated relative amplitude values according to MVC in force direction (Left column). Open circles indicate significant differences between both measured and estimated values p o 0.05 (Wilcoxon–Mann–Whitney-Test). Right column: men's measured amplitude values in antagonistic force direction.
A. Huebner et al. / Journal of Biomechanics 48 (2015) 1198–1205
antagonistic force direction
250
50
200
40
rel. MVC [%]
RA
rel. MVC [%]
force direction
150 100 50
20
40
60
80
100
80 60 40 20 20
40
60
80
60
80
100
rel. UBM
100 80 60 40
100
0
20
100
rel. UBM
80
rel. MVC [%]
rel. MVC [%]
100
60 40 20
40
60
80
100
80
100
80
100
80
100
rel. UBM
80 60 40 20
0
0 0
20
120
40
60
80
100
0
20
120
rel. UBM
100
rel. MVC [%]
rel. MVC [%]
40
0 0
80 60 40 20
40
60
rel. UBM
100 80 60 40 20
0
0 0
20
200
40
60
80
100
0
20
200
rel. UBM rel. MVC [%]
rel. MVC [%]
20
20
0
LO
0 120
rel. UBM
100
rel. MVC [%]
rel. MVC [%]
120
MF
20
0 0
OE
30
10
0
OI
1203
150 100 50 0
40
60
rel. UBM
150 100 50 0
0
20
40
60
80
100
rel. UBM
0
20
40
60
rel. UBM Fig. 4. (continued)
enable an estimation of co-contraction. Compared to the amplitude values in force direction no relevant co-contraction can be assumed but for women's LO at high load levels. As in the abdominal muscles amplitude levels of men are lower than in women. For both back muscles no systematic differences in deviation from estimated values can be proven between sexes.
4. Discussion Clear differences in amplitude–force relationship are obvious for the SEMG investigated trunk muscles. Curve shapes are representative and allow for a reliable classification between abdominal and
back muscles as described by Anders et al. (2008). With increasing load level measured back muscle amplitude values follow a linear increase while for abdominals curve progressions differ significantly from linearity. From the MVC based calculations of measured values, a direct comparison to estimated values reveals further abdominal and back muscle specific characteristics: for MF only minor discrepancies occur between estimated and measured curves. However, for LO increasing load levels cause measured and estimated values to drift apart considerably with measured values falling short of the expected ones. Therefore, reliance on linearly calculated amplitude– force values leads to an overestimation of actual muscular stress levels. In contrast, for the abdominals actual muscle stress is overestimated at low loads but with increasing load stress levels are
A. Huebner et al. / Journal of Biomechanics 48 (2015) 1198–1205
RA
Differences [%]
1204
20 0 -20 -40 -60 -80
OI
Differences [%]
0
20
40
60
80
100
80
100
80
100
80
100
rel. UBM [%]
20 10 0 -10 -20 -30 0
20
OE
Differences [%]
10
40
60
rel. UBM [%]
0 -10 -20 -30
MF
Differences [%]
0
20
40
60
rel. UBM [%]
10 5 0 -5
relationship. However, our findings show a certain amount of load dependent over- and underestimation of muscle stress levels from linearity that is subject to localization. To show possible co-contractions for the investigated muscles antagonistic force directions are displayed as well. Here, no relevant muscle activation can be found but for women's LO. At this point we do not have an explanation for this sex-specific occurrence of cocontraction. Since the general curve characteristics are very similar between sexes this might only alter the data quantitatively to some extent. Healthy individuals do not differ largely in their muscle response as relative measures of their values during MVC effort. Of course, there are subjects with discrepant results. They are to be judged as divergent in muscle function from our normative data. Even though those subjects were classified as healthy there are other reasons that may explain the outliers, e.g. inexperience with the device, motivation or others. These factors cannot be assessed afterwards. Since we showed significant differences in both groups, subjects with discrepant results represent individual cases and not a systematic deviation of the found tendencies. Our study design includes the mentioned trunk muscles. Of course, we cannot exclude possible influences of deep muscles, e.g. iliopsoas muscle, transversus abdominis muscle or the influence of other superficial muscles like the latissimus dorsi and iliocostalis muscles. Their possible effect on our results cannot be estimated so far.
-10 0
LO
Differences [%]
20
20
40
60
rel. UBM [%]
15
Conflict of interest The study was supported by the Central Innovation Program of the German Federal Ministry of Economics and Technology, Grant KF2150501WD8. All authors disclose any financial and personal relationships with other people or organizations that could inappropriately influence (bias) their work.
10 5 0 -5 0
50
100
rel. UBM [%]
female
male
Fig. 5. Median differences and quartile ranges between estimated and measured relative values for all investigated muscles of both sexes. Positive values stand for an overestimation of stress while negative values represent an underestimation of stress compared to the estimated values. Black circles indicate significant differences between sexes. p o0.05 (U-test).
underestimated considerably. This is true for all measured abdominal muscles, but is most pronounced for RA, reaching amplitudes exceeding even those recorded during MVC trials. The observed intersections between estimated and measured curves always occur at higher load levels in men than in women, but these differences at least for RA and OE seem to be caused by the different force capacities of both sexes since they occur at similar stress levels according to MVC. For estimation of back muscle stress levels during rehabilitation and training applications, linearly interpolated stress levels according to MVC normalizations are adequate. However, for LO a continuous overestimation of actual stress levels seems to happen.
Acknowledgment The study was supported by the Central Innovation Program of the German Federal Ministry of Economics and Technology, Grant KF2150501WD8. The study sponsors had no involvement in the study design, the collection, analysis and interpretation of data, or in the writing of the manuscript. Measurements were carried out at the kindly provided laboratory of the Center for Interdisciplinary Prevention of Diseases related to Professional Activities (KIP) founded and funded by the FriedrichSchiller-University Jena and the German Social Accident Insurance Institution for the foodstuffs and catering industry. The authors would like to thank Ms. Marcie Matthews of polishedwords for language assistance (funded by Central Innovation Program of the German Federal Ministry of Economics and Technology, Grant KF2150501WD8) and Mr. Philipp Schenk and Mr. Eduard Kurz for critical discussions which helped to improve the manuscript considerably.
4.1. Limitations
Appendix A. Supporting information
Determination of MVC was conducted in extension and flexion directions under isometric conditions by subjects producing force against a fixed resistance while being positioned vertically. In contrast, the different submaximal load levels were performed by tilting the subjects in sagittal plane, i.e. deflecting them from vertical position in the gravitational field. This difference in force producing strategy might influence muscle stress levels, but seems unlikely to change the characteristics of the amplitude–force
Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.jbiomech.2015.02.016. References Anders, C., Brose, G., Hofmann, G.O., Scholle, H.C., 2008. Evaluation of the EMG-force relationship of trunk muscles during whole body tilt. J. Biomech. 41 (2), 333–339.
A. Huebner et al. / Journal of Biomechanics 48 (2015) 1198–1205
Anders, C., Schumann, N.P., Scholle, H.C., Witte, H., Zwiener, U., 1991. Quantification of artefacts in surface EMG by validating the lower frequency limit in clinicophysiologic studies. EEG-EMG Z. Elektroenzephalogr., Elektromyogr. Verwandte Geb. 22 (1), 40–44. Bernardi, M., Solomonow, M., Nguyen, G., Smith, A., Baratta, R., 1996. Motor unit recruitment strategy changes with skill acquisition. Eur. J. Appl. Physiol. 74 (1-2), 52–59. Clarys, J.P., 2000. Electromyography in sports and occupational settings: an update of its limits and possibilities. Ergonomics 43 (10), 1750–1762. Hermens, H.J., Freriks, B., Merletti, R., Stegeman, D.F., Blok, J., Rau, G., Disselhorst-Klug, C., Hägg, G., 1999. European recommendations for surface electromyography, results of the SENIAM project. Roessingh Research and Development b.v., Enschede ,vol. 8. Kendall, F.P., Kendall Mc Geary, E., Provance, P.G., Rodgers, M.M., Romani, W.A., 2005. Muscles Testing and Function with Posture and Pain, 4th ed. Lippincott Williams & Wilkins, Philadelphia.
1205
McNair, P.J., Depledge, J., Brettkelly, M., Stanley, S.N., 1996. Verbal encouragement: effects on maximum effort voluntary muscle action. Br. J. Sports Med. 30 (3), 243–245. Ng, J.K., Kippers, V., Richardson, C.A., 1998. Muscle fibre orientation of abdominal muscles and suggested surface EMG electrode positions. Electromyogr. Clin. Neurophysiol. 38 (1), 51–58. O'Sullivan, P., Twomey, L., Allison, G., Sinclair, J., Miller, K., 1997. Altered patterns of abdominal muscle activation in patients with chronic low back pain. Aust. J. Physiother. 43 (2), 91–98. Roy, S.H., Oddsson, L.I., 1998. Classification of paraspinal muscle impairments by surface electromyography. Phys. Ther. 78 (8), 838–851.
©2015 Elsevier
Contents lists available at ScienceDirect
Journal of Biomechanics journal homepage: www.elsevier.com/locate/jbiomech www.JBiomech.com
Short communication
Re-evaluation of the amplitude–force relationship of trunk muscles Agnes Huebner n, Bernd Faenger, Hans-Christoph Scholle, Christoph Anders Clinic for Trauma-, Hand- and Reconstructive Surgery, Division for Motor Research, Pathophysiology and Biomechanics, Jena University Hospital, Bachstrasse 18, 07743 Jena, Germany
art ic l e i nf o
a b s t r a c t
Article history: Accepted 15 February 2015
Amplitude–force relationships of major trunk muscles are established in terms of curve characteristics, but up to now were not normalized with respect to maximum voluntary contraction (MVC) force levels. The study therefore aims at a re-evaluation of trunk muscle amplitude–force relationship data according to MVC. Surface EMG of five major trunk muscles was taken from 50 healthy subjects of both sexes (age 20–40 years). All tasks were performed in a device where submaximal loads on the trunk were applied by gradually tilting the subjects in sagittal plane to horizontal position. MVC flexion and extension forces were determined in upright position using an additional harness over the subject's shoulders. Furthermore, the subject's upper body mass (UBM) was obtained during forward tilt to horizontal. MVC to UBM ratio was calculated, corrected by the actual tilt angle, and these linearly estimated values compared with the measured relative values according to MVC. All abdominal muscles confirmed the known non-linear amplitude–force relationship. At low load levels the linearly estimated values overestimated the measured ones and, at higher load levels, underestimated the true stress levels considerably. Back muscles confirmed the known linear curve shape, but for the longissimus muscle at L1 level measured data was always below estimated values. With increasing load, muscular stress of abdominal muscles changes from overestimated towards considerably underestimated values if expected stress levels are based on linear interpolation. Major back muscles' activation levels are nearly linear, but the amplitude–force relationship values seem overestimated for longissimus. & 2015 Elsevier Ltd. All rights reserved.
Keywords: Surface electromyography (SEMG) Trunk muscles Maximum voluntary contraction (MVC) Amplitude–force relationship Muscle stress level
1. Introduction Trunk muscles are already known to show a localization specific Surface EMG (SEMG) amplitude–force relationship that is non-linear for abdominal muscles and nearly linear for back muscles (Anders et al., 2008). These published results did not include normalization procedures, so up to now amplitude–force data could not be evaluated according to any commonly accepted reference point. In place of normalization procedures, especially for isometric situations, the assessment according to maximum voluntary contraction force (MVC) still serves as the gold standard despite its known drawbacks: it is subject to motivation (McNair et al., 1996) and practice (Bernardi et al., 1996); in the case of investigating many muscles requires extreme effort (Kendall et al., 2005); and moreover, it has limited applicability in patients (O'Sullivan et al., 1997; Roy and Oddsson, 1998). Despite these limitations, in healthy people MVC trials are still preferred to estimate submaximal stress levels (Clarys, 2000).
n
Corresponding author. Tel.: þ 49 3641 934 094; fax: þ 49 3641 934 091. E-mail address: [email protected] (A. Huebner).
http://dx.doi.org/10.1016/j.jbiomech.2015.02.016 0021-9290/& 2015 Elsevier Ltd. All rights reserved.
Normally in SEMG applications, estimates of muscle stress are usually interpolated linearly. However, as mentioned above, for the abdominals a non-linear amplitude–force relationship has been established, preventing credible estimations based on this method. But still the question remains unanswered as to the extent and direction of the assumable deviations between actual data and estimated values: are data permanently misjudged in one direction or do they possibly intersect the estimated ones? This investigation aims at a re-evaluation of the known SEMG amplitude–force relationship of trunk muscles by normalization of amplitudes based on SEMG values during isometric MVC tasks.
2. Materials and methods 2.1. Experimental design For this study 25 healthy subjects of each sex aged between 20 and 40 years (men: mean 29SD77.3, women: mean 26SD76.1) were investigated. Subjects' health status was verified by physical examination and self-reported history. The study followed the ethics requirements for human investigations and was positively
A. Huebner et al. / Journal of Biomechanics 48 (2015) 1198–1205 evaluated by the local ethics committee (2643-08/09). Written informed consent was obtained from every subject. The investigation was performed in a device for trunk muscle diagnosis and training (CTT Centaurs, BfMC, Germany) where subjects' lower body was fixed at thighs and hips while their trunk remained unsupported (Fig. 1). In its common use the device applies submaximal forces on the trunk by tilting the whole body from neutral upright position. During these submaximal tests the subject's task is to stabilize the upper body against gravitational forces while remaining along body axis. Monitoring of upper body position is facilitated by a harness (Fig. 1), positioned over the subject's shoulders and equipped with strain gauges which are used for biofeedback via a small monitor in front of the subject. By this, virtually isometric conditions with invariant muscle positions can be assumed. Forward and backward tilts at 51, 101, 201, 301, 451, 601 and 901 (i.e. 9%, 17%, 34%, 50%, 71%, 87%, 100% of upper body mass (UBM)) were applied. For this study subjects were tilted in sagittal plane only. The different angles and tilt directions were performed in a randomized order to prevent fatigue and order dependent effects. Prior to the submaximal tests, flexion and extension MVC tests were conducted while subjects were first positioned in the device. Held vertically, they were asked to exhibit maximum force against the harness. Subjects were given a habituation trial, followed by three MVC trials, each of five seconds duration with 30 s pause. During task performance subjects held their arms crossed against their chest to avoid effects due to varying arm position.
1199
Individual UBM was determined by tilting the subjects forward to horizontal position while being consciously relaxed: SEMG of the back muscles was measured and residual contractions announced to the subject for correction. To anthropometrically normalize the measured MVC levels a MVC to UBM ratio was determined.
2.2. SEMG data acquisition and analysis SEMG was taken from the following five trunk muscles of both sides: rectus abdominis muscle (RA), obliquus internus abdominis muscle (OI), obliquus externus abdominis muscle (OE), multifidus muscle (lumbar part, at L5 vertebral level, MF) and erector spinae muscle (longissimus, at L1 vertebral level, LO) (Fig. 2). Electrode positions were chosen in accordance with the SENIAM recommendations (Hermens et al., 1999), and, if not available for the respective muscle, according to Ng et al. (1998). We also measured the electrocardiographical (ECG) activation by placing one electrode pair along the heart axis. For SEMG measurements disposable Ag–AgCl electrodes (H93SG, Covidien, Germany) with a circular uptake area of 1.6 cm diameter were used. SEMG was amplified (gain: 1000, Biovision, Germany) using a bipolar montage (inter-electrode distance: 2.5 cm). The signals were analog to digital converted at a rate of 2048/s (Tower of Measurement, DeMeTec, Germany; anti-aliasing filter: 1024 Hz, resolution: 24 bit, software: GJB, Germany). All data was processed equally: elimination of DC components, high-pass filter at 10 Hz, low-pass filter at 300 Hz, and notch filtering at 50 Hz. Elimination of ECGartefacts was accomplished by only analyzing stationary signal sections in a time window of 400 ms starting 100 ms after each detected R wave of the ECG channel (Anders et al., 1991). Except for the MVC tasks where only about three seconds of maximal activity were averaged, the 10 s submaximal situations contained between nine and 15 ECG events. By averaging all respective signal sections a representative RMS value was calculated for every channel and task, respectively. Since the tasks were performed in sagittal plane RMS data from both sides were pooled, resulting in five single mean amplitude values per task. The pooled RMS data were normalized according to the corresponding values during MVC testing, further referred to as measured values. Considering the MVC to UBM ratio, expected SEMG amplitudes were calculated, corrected for the actual tilt angle by applying the sinus function. This procedure resulted in values following a linear trend, further referred to as estimated values. These estimated values were then compared with the measured values, separately for both sexes. Data variability of SEMG data was characterized by an unequal distribution of quartile ranges with less variance in lower quartiles. Hence presented data are specified through their median values and corresponding quartile ranges. Accordingly, non-parametric statistics were applied. In Section 3 tasks will be displayed that directly provoked forces on the respective muscles, i.e. abdominal muscles during backward tasks and back muscles during forward tasks. Furthermore antagonistic muscle activity will be displayed as well to uncover possible influences on the data by involuntary co-contractions, i.e. abdominal muscle activity during forward tilts and back muscle activity during backward tilts.
2.3. Statistical analysis
Fig. 1. Subject performing 301 forward tilt in the device for trunk muscle diagnosis and training (CTT Centaurs).
Prior to pooling of RMS values an ANOVA was calculated to ensure that no significant differences between body sides occurred. Paired student's t-tests were conducted on the force data to determine differences between tilt directions and unpaired student's t-tests to determine differences between sexes. Since SEMG data was not normally distributed, the non-parametric Wilcoxon test for dependent samples was applied for statistical comparison of measured and estimated relative SEMG values. For tests of SEMG data between sexes the non-parametric Mann–Whitney-U-test for independent samples was conducted (Edgeworth approximation).
ECG
rectus abdominis muscle erector spinae muscle (longissimus) obliquus externus abdominis muscle multifidus muscle (lumbar part) obliquus internus abdominis muscle Fig. 2. Electrode positions of investigated trunk muscles.
1200
A. Huebner et al. / Journal of Biomechanics 48 (2015) 1198–1205
Table 1 Ratio of maximal voluntary contraction force (MVC) to upper body mass (UBM) for extension (e) as well as flexion (f) direction and the respective MVC/UBM ratios. Student's t-test results for ratios within one sex as well as between sexes po 0.05. Female
MW SD t-test e/f t-test sex
Male
MVC/UBM extension
MVC/UBM flexion
e/f ratio
MVC/UBM extension
MVC/UBM flexion
e/f ratio
1.98 0.37 o 0.001 o 0.001
1.48 0.39
1.33 0.42
2.37 0.37 o0.001
1.94 0.40
1.23 0.40
o 0.001
0.17
t-test e/f: paired t-test. t-test sex: unpaired t-test.
force direction
100
y = 0.0092x 2 + 0.0616x + 0.7224 R² = 0.9963 y = 0.6586x + 1.4323 R² = 0.9998
50
150 rel. MVC [%]
RA
rel. MVC [%]
150
antagonistic force direction
100 50
0
0
90
y = 0.0034x2 + 0.3902x + 0.5525 R² = 0.9955
60 y = 0.6324x + 3.9212 R² = 0.9996
30
100 rel. MVC [%]
OI
rel. MVC [%]
50
0
60 30 0
50
100
rel. UBM [%]
90
y = 0.0021x2 + 0.5591x + 0.6325 R² = 0.9985
60 y = 0.647x + 2.3587 R² = 0.9995
30
rel. MVC [%]
rel. MVC [%]
0
0
50 y = 0.0003x 2 + 0.3664x + 6.9289 R² = 0.9877
40
y = 0.4746x + 2.8603 R² = 0.9996
20
rel. MVC [%]
rel. MVC [%]
rel. UBM [%]
60 30
100 60
60
0
50
100
rel. UBM [%]
40 20
0
0
60
50 y=
0.0004x2 +
rel. UBM [%] 0.2565x + 9.5056
R² = 0.9968
40
y = 0.4597x + 4.6144 R² = 0.999
20
100 0 60 rel. MVC [%]
0 rel. MVC [%]
100
0 0
LO
50
90
0
MF
100
rel. UBM [%]
90
0
OE
50
50
100
rel. UBM [%]
40 20
0
0 0
50 rel. UBM [%]
estimated values
100
0
50
100
rel. UBM [%]
measured values
Fig. 3. Women's measured and linearly estimated relative amplitude values according to MVC in force direction (left column). Open circles indicate significant differences between both measured and estimated values p o 0.05 (Wilcoxon–Mann–Whitney-Test). Right column: women's measured amplitude values in antagonistic force direction.
A. Huebner et al. / Journal of Biomechanics 48 (2015) 1198–1205
antagonistic force direction
600
100
500
80
rel. MVC [%]
RA
rel. MVC [%]
force direction
400 300 200
60 40 20
100 0
0 0
20
40
60
80
100
0
120 80 40
20
40
60
80
80
100
80
100
80
100
80
100
rel. UBM
0
20
160
rel. UBM
120 80 40
40
60
rel. UBM
120 80 40 0
0
20
100
40
60
80
100
0
20
100
rel. UBM
80 60 40
40
60
rel. UBM
80
rel. MVC [%]
rel. MVC [%]
100
40
0
60 40
20
20
0
0 0
20
100
40
60
80
100
0
20
100
rel. UBM
80
rel. MVC [%]
rel. MVC [%]
80
80
100
rel. MVC [%]
rel. MVC [%]
160
LO
60
0 0
MF
40
120
0
OE
20
160
rel. UBM rel. MVC [%]
rel. MVC [%]
160
OI
1201
60 40 20
40
60
rel. UBM
80 60 40 20
0
0 0
20
40
60
80
100
rel. UBM
0
20
40
60
rel. UBM Fig. 3. (continued)
3. Results
3.2. Abdominal muscles
3.1. Force data
For all abdominal muscles the known non-linear amplitude– force relationship could be confirmed by the measured data (upper three rows in the left columns of Figs. 3 and 4). Compared to estimated values, at low load levels measured amplitudes were significantly lower, but as load levels increased measured amplitudes gradually reached higher values than estimated. Curve shapes of measured data are quite similar between muscles and sexes, but
Significant differences occur for MVC to UBM ratios between extension and flexion tasks with higher values during extension. Likewise the respective extension and flexion ratios differ between sexes, but the extension to flexion ratio does not show any sex differences (Table 1).
1202
A. Huebner et al. / Journal of Biomechanics 48 (2015) 1198–1205
i.e. horizontal position, is due to large variances within sexes. In contrast, for OI and OE no systematic differences in deviation from estimated values occur between sexes.
quantitative differences exist: for both sexes the intersections where abdominal muscle amplitudes equal the estimated values occur at loads larger than 50% UBM, in men these intersections occur at even higher load but almost identical stress levels. According to the differences in force capacity, estimated as well as measured values in men are always lower than in women. Remarkably, at 100% UBM (i.e. horizontal position) women's measured SEMG levels of RA exceeded even those obtained during MVC testing. During forward tilt, i.e. at antagonistic force direction, insignificant and virtually unchanged abdominal contraction levels occurred (upper three rows in the right columns of Figs. 3 and 4). Amplitude levels of men were again lower than in women. In Fig. 5 the deviations from the estimated values (i.e. in force direction) are quantified and sex differences are analyzed: only for RA significant differences between sexes are verifiable and apply to medium loads and up. The absence of significance at 100% UBM,
3.3. Back muscles Both back muscles confirmed the known linear amplitude–force relationship (lower two rows in the left columns of Figs. 3 and 4). For MF only slight and unsystematic differences occur between measured and estimated values. For LO the measured amplitude values are continuously well below the estimated levels, even more so in men. Significant differences between measured and estimated curves occur at almost all load levels, due to different curve slopes. As in the abdominal muscles, the right columns of Figs. 3 and 4 show data of the antagonistic force direction of women and men to
force direction y = 0.0098x2 – 0.3425x + 4.2701 R² = 0.9878
100
y = 0.5257x + 0.6033 R² = 0.9999
50
150 rel. MVC [%]
RA
rel. MVC [%]
150
antagonistic force direction
100 50
0
0 50 rel. UBM [%] y = 0.0063x 2 – 0.0169x + 1.3888 R² = 0.9973
90 60
y = 0.5175x + 1.3861 R² = 0.9998
30
100
30
50
rel. UBM [%] y = 0.0031x2 + 0.3133x –0.8581 R² = 0.997 y = 0.5244x + 0.6904 R² = 0.9998
30
100
30
50 rel. UBM [%] y = -0.0005x 2 + 0.3767x + 7.6736 R² = 0.9913 y = 0.3961x + 2.6892 R² = 0.9981
40 20
100 60 rel. MVC [%]
rel. MVC [%]
60
0
0
50
100
rel. UBM [%]
40 20 0
60
50 rel. UBM+[%] y = 8E-05x 2 + 0.2207x 8.2081 R² = 0.9973
40
y = 0.3726x + 5.3857 R² = 0.999
20
100 0 60 rel. MVC [%]
0 rel. MVC [%]
100
0 0
LO
50 rel. UBM [%]
60
0
MF
0
90
rel. MVC [%]
rel. MVC [%]
60
100
0 0
OE
50 rel. UBM [%]
60
0
90
0
90
rel. MVC [%]
OI
rel. MVC [%]
0
0 0
50 rel. UBM [%]
estimated values
50
100
rel. UBM [%]
40 20
0 100 0
50
100
rel. UBM [%]
measured values
Fig. 4. Men's measured and linearly estimated relative amplitude values according to MVC in force direction (Left column). Open circles indicate significant differences between both measured and estimated values p o 0.05 (Wilcoxon–Mann–Whitney-Test). Right column: men's measured amplitude values in antagonistic force direction.
A. Huebner et al. / Journal of Biomechanics 48 (2015) 1198–1205
antagonistic force direction
250
50
200
40
rel. MVC [%]
RA
rel. MVC [%]
force direction
150 100 50
20
40
60
80
100
80 60 40 20 20
40
60
80
60
80
100
rel. UBM
100 80 60 40
100
0
20
100
rel. UBM
80
rel. MVC [%]
rel. MVC [%]
100
60 40 20
40
60
80
100
80
100
80
100
80
100
rel. UBM
80 60 40 20
0
0 0
20
120
40
60
80
100
0
20
120
rel. UBM
100
rel. MVC [%]
rel. MVC [%]
40
0 0
80 60 40 20
40
60
rel. UBM
100 80 60 40 20
0
0 0
20
200
40
60
80
100
0
20
200
rel. UBM rel. MVC [%]
rel. MVC [%]
20
20
0
LO
0 120
rel. UBM
100
rel. MVC [%]
rel. MVC [%]
120
MF
20
0 0
OE
30
10
0
OI
1203
150 100 50 0
40
60
rel. UBM
150 100 50 0
0
20
40
60
80
100
rel. UBM
0
20
40
60
rel. UBM Fig. 4. (continued)
enable an estimation of co-contraction. Compared to the amplitude values in force direction no relevant co-contraction can be assumed but for women's LO at high load levels. As in the abdominal muscles amplitude levels of men are lower than in women. For both back muscles no systematic differences in deviation from estimated values can be proven between sexes.
4. Discussion Clear differences in amplitude–force relationship are obvious for the SEMG investigated trunk muscles. Curve shapes are representative and allow for a reliable classification between abdominal and
back muscles as described by Anders et al. (2008). With increasing load level measured back muscle amplitude values follow a linear increase while for abdominals curve progressions differ significantly from linearity. From the MVC based calculations of measured values, a direct comparison to estimated values reveals further abdominal and back muscle specific characteristics: for MF only minor discrepancies occur between estimated and measured curves. However, for LO increasing load levels cause measured and estimated values to drift apart considerably with measured values falling short of the expected ones. Therefore, reliance on linearly calculated amplitude– force values leads to an overestimation of actual muscular stress levels. In contrast, for the abdominals actual muscle stress is overestimated at low loads but with increasing load stress levels are
A. Huebner et al. / Journal of Biomechanics 48 (2015) 1198–1205
RA
Differences [%]
1204
20 0 -20 -40 -60 -80
OI
Differences [%]
0
20
40
60
80
100
80
100
80
100
80
100
rel. UBM [%]
20 10 0 -10 -20 -30 0
20
OE
Differences [%]
10
40
60
rel. UBM [%]
0 -10 -20 -30
MF
Differences [%]
0
20
40
60
rel. UBM [%]
10 5 0 -5
relationship. However, our findings show a certain amount of load dependent over- and underestimation of muscle stress levels from linearity that is subject to localization. To show possible co-contractions for the investigated muscles antagonistic force directions are displayed as well. Here, no relevant muscle activation can be found but for women's LO. At this point we do not have an explanation for this sex-specific occurrence of cocontraction. Since the general curve characteristics are very similar between sexes this might only alter the data quantitatively to some extent. Healthy individuals do not differ largely in their muscle response as relative measures of their values during MVC effort. Of course, there are subjects with discrepant results. They are to be judged as divergent in muscle function from our normative data. Even though those subjects were classified as healthy there are other reasons that may explain the outliers, e.g. inexperience with the device, motivation or others. These factors cannot be assessed afterwards. Since we showed significant differences in both groups, subjects with discrepant results represent individual cases and not a systematic deviation of the found tendencies. Our study design includes the mentioned trunk muscles. Of course, we cannot exclude possible influences of deep muscles, e.g. iliopsoas muscle, transversus abdominis muscle or the influence of other superficial muscles like the latissimus dorsi and iliocostalis muscles. Their possible effect on our results cannot be estimated so far.
-10 0
LO
Differences [%]
20
20
40
60
rel. UBM [%]
15
Conflict of interest The study was supported by the Central Innovation Program of the German Federal Ministry of Economics and Technology, Grant KF2150501WD8. All authors disclose any financial and personal relationships with other people or organizations that could inappropriately influence (bias) their work.
10 5 0 -5 0
50
100
rel. UBM [%]
female
male
Fig. 5. Median differences and quartile ranges between estimated and measured relative values for all investigated muscles of both sexes. Positive values stand for an overestimation of stress while negative values represent an underestimation of stress compared to the estimated values. Black circles indicate significant differences between sexes. p o0.05 (U-test).
underestimated considerably. This is true for all measured abdominal muscles, but is most pronounced for RA, reaching amplitudes exceeding even those recorded during MVC trials. The observed intersections between estimated and measured curves always occur at higher load levels in men than in women, but these differences at least for RA and OE seem to be caused by the different force capacities of both sexes since they occur at similar stress levels according to MVC. For estimation of back muscle stress levels during rehabilitation and training applications, linearly interpolated stress levels according to MVC normalizations are adequate. However, for LO a continuous overestimation of actual stress levels seems to happen.
Acknowledgment The study was supported by the Central Innovation Program of the German Federal Ministry of Economics and Technology, Grant KF2150501WD8. The study sponsors had no involvement in the study design, the collection, analysis and interpretation of data, or in the writing of the manuscript. Measurements were carried out at the kindly provided laboratory of the Center for Interdisciplinary Prevention of Diseases related to Professional Activities (KIP) founded and funded by the FriedrichSchiller-University Jena and the German Social Accident Insurance Institution for the foodstuffs and catering industry. The authors would like to thank Ms. Marcie Matthews of polishedwords for language assistance (funded by Central Innovation Program of the German Federal Ministry of Economics and Technology, Grant KF2150501WD8) and Mr. Philipp Schenk and Mr. Eduard Kurz for critical discussions which helped to improve the manuscript considerably.
4.1. Limitations
Appendix A. Supporting information
Determination of MVC was conducted in extension and flexion directions under isometric conditions by subjects producing force against a fixed resistance while being positioned vertically. In contrast, the different submaximal load levels were performed by tilting the subjects in sagittal plane, i.e. deflecting them from vertical position in the gravitational field. This difference in force producing strategy might influence muscle stress levels, but seems unlikely to change the characteristics of the amplitude–force
Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.jbiomech.2015.02.016. References Anders, C., Brose, G., Hofmann, G.O., Scholle, H.C., 2008. Evaluation of the EMG-force relationship of trunk muscles during whole body tilt. J. Biomech. 41 (2), 333–339.
A. Huebner et al. / Journal of Biomechanics 48 (2015) 1198–1205
Anders, C., Schumann, N.P., Scholle, H.C., Witte, H., Zwiener, U., 1991. Quantification of artefacts in surface EMG by validating the lower frequency limit in clinicophysiologic studies. EEG-EMG Z. Elektroenzephalogr., Elektromyogr. Verwandte Geb. 22 (1), 40–44. Bernardi, M., Solomonow, M., Nguyen, G., Smith, A., Baratta, R., 1996. Motor unit recruitment strategy changes with skill acquisition. Eur. J. Appl. Physiol. 74 (1-2), 52–59. Clarys, J.P., 2000. Electromyography in sports and occupational settings: an update of its limits and possibilities. Ergonomics 43 (10), 1750–1762. Hermens, H.J., Freriks, B., Merletti, R., Stegeman, D.F., Blok, J., Rau, G., Disselhorst-Klug, C., Hägg, G., 1999. European recommendations for surface electromyography, results of the SENIAM project. Roessingh Research and Development b.v., Enschede ,vol. 8. Kendall, F.P., Kendall Mc Geary, E., Provance, P.G., Rodgers, M.M., Romani, W.A., 2005. Muscles Testing and Function with Posture and Pain, 4th ed. Lippincott Williams & Wilkins, Philadelphia.
1205
McNair, P.J., Depledge, J., Brettkelly, M., Stanley, S.N., 1996. Verbal encouragement: effects on maximum effort voluntary muscle action. Br. J. Sports Med. 30 (3), 243–245. Ng, J.K., Kippers, V., Richardson, C.A., 1998. Muscle fibre orientation of abdominal muscles and suggested surface EMG electrode positions. Electromyogr. Clin. Neurophysiol. 38 (1), 51–58. O'Sullivan, P., Twomey, L., Allison, G., Sinclair, J., Miller, K., 1997. Altered patterns of abdominal muscle activation in patients with chronic low back pain. Aust. J. Physiother. 43 (2), 91–98. Roy, S.H., Oddsson, L.I., 1998. Classification of paraspinal muscle impairments by surface electromyography. Phys. Ther. 78 (8), 838–851.
©2015 Elsevier
