The muscles of the abdominal wall (rectus abdominis, external oblique, internal oblique, transverse abdominis) and psoas play a fundamental role for the normal functioning of the lumbar spine. Examination of their role as movers and dynamic stabilizers has been greatly assisted by electromyographic techniques although past work on the deeper muscles such as psoas(1,9) and the deeper layers of the abdominal wall, using intramuscular electrodes, has been sparse and qualitative. The objective of this work was to obtain normalized activation amplitudes of these deep muscles to understand their role in spine mechanics, more specifically, for use in rehabilitation and investigations of the injury process.
Earlier EMG studies of psoas were not normalized to some form of standardized contraction effort, precluding physiological interpretation(e.g.,2,9-11,17,20,22). Furthermore, many low back problems have been hypothesized to be associated with (or a result of) muscular weakness and/or imbalance(6). For this reason, an important aim in rehabilitation programs is to train for strength and endurance of the abdominals muscles. Ideally, the objective is to challenge the abdominals while imposing a minimal load penalty to the lumbar spine given the link between injury, spine posture, and compressive loading (15). Since psoas has been shown to be a major source of spine loading (3,23) we were motivated to measure the activity of psoas and the abdominal wall over a variety of tasks and flexion-dominated exercises. In this way, rehabilitation exercises and other tasks could be ranked to identify the most effective ones for utilization and training of the various abdominals and psoas.
The psoas major is the largest muscle in cross-section at the lower levels of the lumbar spine (12). A review of the literature reveals that little is known about its mechanical capacity and activity with respect to the lumbar spine. This is because of its location within the body, which precludes routine access with indwelling myoelectric electrodes. Often, psoas is not considered separately from the iliopsoas complex(2) during investigations; however, it is quite different in architecture and attachments when compared with iliacus(23). Bogduk et al. (3) concluded that the psoas major is not a prime mover of the lumbar spine and that the resultant compression and shear forces from its activation were substantial but were considered a large price paid (by the lumbar spine) for a well-designed hip flexor. Santaguida and McGill (23) demonstrated in a three-dimensional mechanical modeling study, based on MRI measurement, that psoas has the potential to stabilize the lumbar spine with compressive loading from bilateral activation, laterally flex it, and can create large anterior shear forces but only at L5-S1.
The specific purpose of this work was to quantify psoas involvement with normalized EMG over a wide variety of tasks that challenge the low back and hips. Several hypotheses and subquestions were of interest: How is the psoas maximally activated? Is psoas activity primarily linked to hip flexion or internal/external rotation? What is the best way to train the abdominals while minimizing psoas involvement? Is psoas active during squat style lifting?
Data collection. Five men (age [horizontal bar over]x = 25.8 yr, SD = 1.3; height [horizontal bar over]x = 1.77 m, SD = 0.08; weight[horizontal bar over]x = 72 kg, SD = 8 and three women (age [horizontal bar over]x = 23.3 yr, SD = 2.3; height [horizontal bar over]x = 1.68 m, SD = 0.03; weight - = 60 kg, SD = 6) were recruited from a student population. All subjects were fit, healthy, and had never experienced disabling low back pain. The experiments took approximately 3 h to ensure that fatigue was not a confounding variable. The electrodes were in the body for approximately 4 h. The experimented protocol was approved by the ethics committee of the Faculty of Medicine, University of Bern.
There were two sets of tasks performed in this study: the first set consisted of a series of maximal effort isometric exertions intended to produce the largest possible amplitudes of myoelectric activity to provide a basis for normalization. Subjects were given sufficient practice and provided with feedback in the form of myoelectric signals displayed in real time on the computer monitor. Real time monitoring also provided an instant check on data quality. After pilot work the following protocols were adopted. Maximum activation of psoas (left side) was obtained by asking subjects to stand on the right leg, raise the left knee to approximately 90 °, where both hands pushed down on the thigh while both maximal effort hip flexion and spine lateral bending efforts were performed. Maximum activation for rectus femoris was performed with subjects seated on a test bench while they performed simultaneous hip flexion and knee extension against resistance. The spine extensors were normalized during an exertion where subjects lay prone on a bench with the upper body unsupported out over the end of the table. The feet were restrained with a tight-fitting self-stick strap. While in this position, subjects started with a slightly flexed lumbar region and then slowly extended the lumbar spine against a matched resistance on the upper back provided from an assistant. Maximal abdominal activation was obtained with the subjects starting in a bent knee sit-up posture (knees at 90 °) with the feet restrained by a strap. Hands were placed on the opposite shoulder while an assistant provided a matched resistance to the shoulders. Three trials of each normalization exertion were performed. The instructions for the exertion were to combine maximal flexor effort with simultaneous slow isometric twisting efforts. On occasion, some maximal activation muscle signals were obtained during other maximal exertions: for example, rectus abdominis activity may be slightly larger during the maximum psoas routine. The largest amplitude for each muscle was used regardless of the activity from which it was obtained. All these tasks will be referred to as the maximal voluntary contraction (MVC) trials.
A wide variety of tasks (a second set) was then selected to test the several hypotheses stated earlier. Tasks were grouped into subsections: those primarily for examining the mechanics of torso flexion (including bent leg and straight leg sit-ups and leg raises, curl-ups, push-ups), lateral bend(standing lateral bends, side supports), extension (and lifting loads up to 70 kg), and hip internal/external rotation (seated and standing). Another set of tasks consisted of different gymnastic exercises (e.g., to train the flexors) used in training and rehabilitation programs (see Fig. 1). All dynamic tasks were performed slowly and smoothly to minimize inertial effects and EMG movement artifact.
Both surface and intramuscular electrodes were used for muscles on the left side of the body. Fine wire (76-μm wire, 100-μm outer diameter gage, purity 99.9% silver, insulation: heavy polymide Pyre ML Dupont, Fa. Corradi, Milan, Italy) bipolar electrodes were made by running the wire down the outside of a 0.9-mm bore hypodermic needle where the uninsulated tips (2 mm) were inverted into the canula bore. The tips were 3 mm apart. In this way the wires and canula were inserted into the muscle, and the canula was removed, leaving the wires with the hooked end in the desired location. Wires were guided to the appropriate muscle under ultrasound images (Toshiba Tosbee SSA-240A, 7.5 MHz probe). Two pairs of intramuscular electrodes were inserted into psoas at the level of L3 approximately 10 cm from the midline passing the lumbodorsal fascia and quadratus lumborum muscle and at an angle of approximately 45 ° to the sagittal plane. Subjects lay on their sides while the psoas electrodes were inserted. One percent Xylocaine was injected to anesthetize the skin and fascia at entry. Intramuscular electrodes were also placed in external oblique, internal oblique, and transverse abdominis midway between the linea semi lunaris and the midline laterally and at the transverse level of the umbilicus. Surface electrodes (Beckmann bipolar, Ag Ag-Cl, with a center-to-center distance of 2.5 cm) were placed on the following muscles (see Fig. 2): rectus abdominis: 3 cm lateral to the umbilicus; external oblique: approximately 15 cm lateral to the umbilicus and at the transverse level of the umbilicus; internal oblique: below the external oblique electrodes and just superior to the inguinal ligament; lumbar erector spinal: 3 cm lateral to the L3 spinous process; rectus femoris: over the belly of the muscle approximately 8 cm below the inguinal ligament.
All raw myoelectric signals were preamplified (gain = 1000, common mode rejection ratio greater than 100 dB at 50 Hz, filtered to produce a band width of 4 to 20,000 Hz). Signals were further filtered (10-500 Hz to minimize artifact) and amplified (2 to 8 times) with custom made analog instrumentation to produce signals of approximately ±5 V. The sagittal plane view of the subjects was also filmed on video tape and scaled so that joint coordinates could be obtained for input into a biomechanical model to calculate moments about the lumbar spine and hips.
Data reduction. All myoelectric signals were A/D converted (12 bit resolution) at 1024 Hz. Signals were then digitally full wave rectified and low pass filtered (single pass, Butterworth) at a cutoff frequency of 3 Hz, and then normalized to the maximum activity observed during the MVC trials. The low pass filter created the electromechanical delay between the electrical event (EMG) and the mechanical event (muscle force). The cutoff frequency of 3 Hz was chosen in the following way: Olney and Winter(21) reported the frequency response of the rectus femoris to be between 1.0 and 2.8 Hz during walking, whereas Milner-Brown Brown et al. (16) reported approximately 3 Hz in the first dorsal interosseous. In addition, the 3-Hz cutoff produced an impulse response (time to peak) of 53 ms, which is comparable with the 30-90-ms contraction times reported by Buchthal and Schmalbruch(4)
Reaction moments about the lumbar spine (during sit-ups, lifting tasks, etc.) and hips (during leg raises) were estimated from a static, rigid linked segment biomechanical model (WATBAK) (19). The most demanding part of each task was chosen for analysis (i.e., maximum effort in isometric tasks, beginning of situp in situp tasks, at point of weight moving off the ground during lifting, etc.). Input included subject height, weight, gender, and x-y coordinates of major body joints (13). The myoelectric data was then analyzed where amplitudes that corresponded to specific events of each task were extracted.
This study was invasive, which restricted subject numbers, and therefore was not designed for statistical analysis. However, statistical analysis was attempted using ANOVA tests and paired t-tests, even with limited data.
To facilitate interpretation of muscle activation and to indicate the overall challenge of each task (at the point of greatest challenge), sagittal moments about the low back (L4/L5 unless otherwise indicated) were computed for the following tasks: bent knee sit-ups: males 88 N·m, females 64 N·m; straight leg situps: males 87 N·m, females 62 N·m; press heels situps: males 85 N·m, females 68 N·.m; bent knee leg raise (calculated about the hip): males 28 N.m, females 18 N·m; straight leg raise (calculated about the hip): males 74 N.m, females 62 N·m; curlups (calculated about T11-T12): males 19 N.m, females 14 N·m; push-ups from the feet: males 60 N·m, females 66 N.m; isometric side support would be a similar moment to pushups although about L4/L5 in the frontal plane (i.e., lateral bending moment).
Because of the quantity of data, an overview (Table 1) is presented followed by an assessment of specific and major clinical questions listing the peak EMG activity. The standard deviation is also provided to indicate variability. Myoelectric variability is normal given differences between people in the way that they respond to prescribed challenges, and even between repeated trials where the motor control system chooses to activate muscles in a different way. Even highly skilled athletes have difficulty repeating simple tasks.
Choosing the most appropriate abdominal flexion exercise. For years, one objective of many abdominal exercises has been to minimize concomitant psoas activity, but without the benefit of quantification. Flexion exercises were ranked such that minimal psoas activation and maximal abdominal muscle activation was considered the desired objective and therefore given the highest rank (Table 2). Time histories of muscle activity of a typical subject are shown in Figures 3 and 4 to contrast a curl-up and bent knee sit-up. It appears that the exercise to challenge both rectus abdominis and the abdominal wall (external and internal oblique, transverse abdominis) and to minimize the lumbar compressive and shear penalty of psoas activation is the combination of curl-up (or cross curl), and either isometric and/or dynamic side support.
Is psoas active during lifting? Two types of lifting were analyzed: dynamically lifting a barbell from the floor using a squat-lift style, and statically holding loads in the hands with arms at sides in an upright standing posture (bucket hold). Lifting a barbell of 20 kg only produced activity of 9 (±10)% (mean ± (SD)) and 3 (± 4)% MVC activity in the two psoas locations while increasing the barbell load to 50-100 kg (depending on the consent of the subject) only increased psoas activation to 16 (±18)% and 5 (±6)% MVC, respectively (seeFigs. 5 and 6). During upright standing with symmetric loads in the hands (bucket hold) the moments about the three axes of the lumbar spine were close to zero since increasing the hand loads simply increased externally applied spine compression. (Weight position and upper body and pelvis alignment were monitored by an assistant to minimize the moment about the low back). Psoas activation was uniformly low as the weights were increased: 2% MVC (±2) and 0 (±1) with no hand load; 2(±4) and 1 (±1) with 20 kg total hand load; 3 (±4) and 1(±1) with 30 kg; 3 (±5) and 1 (±1) with 40 kg, suggesting that psoas plays virtually no role in modulating spine stability in this posture and at these relatively low spine compression levels. In contrast, quadratus lumborum has been demonstrated to act as the dominant spine stabilizer in this task (14).
What are the muscle activation profiles during maximal isometric twist efforts? The abdominal wall (external and internal oblique, transverse abdominis) are active twist moment generators because of their orientation and differential activation. For example, during CW efforts the external oblique was activated at 52 (±13)% MVC, while the internal oblique was 15 (±11)% and transverse abdominis was 18 (±19)% MVC; however, the reverse trend occurred during CCW efforts. Only the activation of the external and internal oblique muscles were significantly different between CW and CCW efforts (paired t-tests: P< 0.0001 and P = 0.0006, respectively) as no other muscle demonstrated significant activation differences. This is perhaps because of the fact that during twisting efforts the abdominals also generate flexor moment (being anterior to the spine) which must be balanced with extensor forces. The erector spinae group appears to perform this flexion/extension moment balancing role given their constant activity with no difference between CW 13 (±8)% and CCW 14 (±6)% MVC efforts. Psoas may be involved in the generation of both twist moment generation and moment balance about other axes because of the approximately 30% difference in myoelectric amplitude between CW and CCW directions.
Is psoas an internal or external hip rotator? During seated isometric internal/external hip rotation efforts, psoas was less active in internal (19-21% MVC seated; 10-21% MVC standing) than external rotation(25-32% MVC seated; 22-27% MVC standing) (ANOVA Psoas1, P = 0.05; Psoas2, P = 0.06). Thus, during sitting psoas activity was an additional 40% higher during external rotation, while psoas activity increased an additional 58% for external rotation efforts in a standing posture.
What task causes psoas to be maximally activated? After comparing psoas activation during torso flexion, torso lateral bending, and hip flexion maneuvers it was found that the best method to obtain maximal activation was to adopt a standing posture, raise the left knee flexing the hip (for left psoas), and push down with the hands placed on the thigh. This created both maximal hip flexion moment together with lumbar lateral bending moment required to elevate the hip and assist contralateral hip abductors. For example, removing the hip elevation component and performing the isometric hand-to-knee exertions while lying on the back reduced psoas activation from 100% MVC to 57 (±20)% on average.
Does the “press heels” sit-up technique disable psoas? Spring (24) hypothesized that activating the hamstrings (extensors of the hip) during bent knee sit-ups would inhibit psoas. This hypothesis was not supported by our results, and in fact psoas activity was increased (28 (±23)% and 34 (±18)% MVC for the press heels style of situp) compared with that during bent knee sit-ups (17(±10)% and 28 (±7)% MVC) (paired t-test - Psoas1,P = 0.0035), compared with that during curl-ups (Psoas1 and Psoas2,P < 0.0001) and compared with that during straight leg sit-ups(Psoas2 - P = 0.057).
Also, external oblique activity increased (from 43 (12)% to 5 (14)% MVC)(P = 0.0065) and transverse abdominis increased (from 10 (7)% to 20(13)% MVC) (P = 0.0025) when the “press heels” technique was performed. Interestingly, rectus femoris showed no changes between the two techniques suggesting that psoas increased activation to balance the extensor hip moment resulting from the increased hamstring activity.
The present study constitutes one of the first quantitative examinations of lumbar psoas activity, to our knowledge, with normalized-scaled intramuscular EMG to facilitate comparison of muscle activation levels in a physiological way. The major clinical findings are: psoas appears not to be involved in lifting and lumbar stabilization-rather it is primarily active to flex the hip. The quantitative activation data of this study agree with the conclusions from the several qualitative studies(2,9,17,22) and the most recent study by Andersson et al. (1) who reported both psoas and iliacus activity and that the psoas activity is most closely linked with the generation of hip flexion moment. In addition, Bogduk et al.(3) and Santaguida and McGill (23) reported that the morphology and geometry of psoas demonstrated that its mechanical advantage gives highest priority to flex the hip. Maximal myoelectric activity of psoas was produced by maximal effort isometric hip flexion (hip at 90 °) in standing posture and pushing down the flexed thigh with both hands combined with the lateral bending moments created by standing on the contralateral leg.
While it is often hypothesized and stated by clinicians that psoas activity is reduced in bent knee sit-ups compared with that in straight leg sit-ups, this appears to be untrue as both psoas and the abdominals have higher levels of activity during the bent knee style of sit-ups (but not statistically significant by the extremely limited analysis performed here). Although the neural activation to psoas may be larger, the muscle is shorter during bent knee positions and force production is likely modulated (i.e., reduced) by the forcelength relationship. Furthermore, curl-ups always produced psoas activity lower than 10% MVC, whereas performing sit-ups and leg raises always increased activity to levels that exceeded 15% MVC. Some authors(24) have suggested that activation of the hamstrings should inhibit the psoas and have emphasized “pressing the heels down” during sit-up exercises. Our data do not support this notion. It appears that subjects, when pulling the heels up toward the buttocks, recruit the hip flexors, thus accounting for the increased psoas activity. In addition, during push-ups, psoas activity is higher than 25% MVC. Testing this exercise was motivated by the authors' clinical experience where a high incidence of pain and discomfort during push-ups have been noted. Observations of psoas activity of 12 to 24% MVC during push-ups must be regarded as contributing to substantial loading of the spine. For example, assuming a reasonable maximum muscle stress of 30 N·cm-2, and the psoas geometry collected on 15 young men by Santaguida and McGill(23), a bilateral activation of 20% MVC during pushups will impose 226 N of compression and 179 N of anterior shear on the lumbosacral junction.
It appears that the side support exercise is the most effective way to train the abdominal obliques with only moderate psoas activity. Furthermore, there is no single best exercise to simultaneously train all abdominal muscles as demonstrated by ranking of individual muscle activation by task (shown inTable 2). For those concerned with spine loading during the performance of exercises to enhance muscle strength and endurance, curl-ups, cross-curl-ups combined with side supports (isometric/dynamic) appear to be the preferred combination of exercises. Both of these exercises are suited for longer duration higher repetition cycles to optimize endurance together with strength gains. Clearly, any form of sit-up is contraindicated in the unstable spine, such as during the first period after surgery, for those with radiological evidence of segmental instability or after injury in the lower lumbar segments, and should be reserved for a healthy athletic spine. Both leg raises and sit-ups do not strongly challenge the abdominal wall but are characterized by higher psoas activity. While this may be of primary concern to those in rehabilitation programs, therapists and/or coaches will realize that a young healthy athlete may require a more rigorous training program.
During lifting, even of loads up to 100 kg, psoas showed much lower activity than expected (less than 16% MVC). Similar observations were made during the isometric bucket-holding tasks with rising external loads. On the one hand, Nachemson (18) suggested that psoas activity was needed to stabilize the spine, while on the other hand psoas can only exert small moments in the sagittal plane (3,23). Perhaps such relatively low activity might be enough to create sufficient stiffness and stability of the lumbar spine for these demanding low back tasks. Specifically, compression from activation will increase segmental stiffness (7) although other muscles have been shown to have a greater role in this role (5), and possibly the splinting action of psoas acting as a muscular sleeve will support shearing translations. Samuel et al. (22) likened this architecture to a composite beam of bone and muscle.
Interestingly, psoas activity was higher during unsupported sitting (both upright and slouched) when compared with quiet, unloaded upright standing. Perhaps, activity of hip extensors inhibit activation of the flexors (i.e., psoas). Alternatively, during sitting perhaps the psoas creates hip flexor moments to balance the elastic extensor moments caused by ligamentous hip capsule strain, and passively stretched extensor muscles and other tissues, as the hip-thighs are close to end range thigh flexion. The opposite occurs in upright standing where the hip capsule is strained at the opposite end of range of motion, causing flexion moment of the thigh requiring gluteal activity. Furthermore, psoas was more active during upright sitting than in a relaxed slouch (t-test: psoas1, P = 0.08; psoas2,P = 0.03). Possibly, psoas helps in upright sitting to extend the upper lumbar spine and flex the lower lumbar spine, even if it acts with only very small moments (3,23). Also, similar to the argument presented in the previous hypothesis, during upright sitting the pelvis rotates forward increasing the amount of thigh flexion and thus increasing the passive extensor elastic moment required an active source of hip flexion. Psoas appears to be the candidate for this role.
Finally, there is some debate as to whether psoas primarily internally or externally rotates the hip. Psoas activity was higher during external rotation, in both seated and standing postures, than during internal rotation. While psoas activity was biased toward external hip rotation, the cocontraction would suggest that the psoas is not the prime mover in either of these rotation movements. Possibly the psoas is active to create a rotation moment about the hip joint but is also active to create a stable base (pelvis, lumbar spine) for the upper body during this maximal isometric effort.
The conclusions of this study are limited by the following methodological restrictions: As the study involved invasive procedures, subject numbers are low. This precluded designing the study for subsequent statistical analysis even though this was attempted. In most cases differences in muscle activation were interpreted based on their mean magnitudes and standard deviation values. Furthermore the subject pool was quite homogeneous (they were young, healthy, and very physically active) and may not represent a wide range of the population. As well, the tasks were not performed over a full range of effort, i.e., the isometric twist tasks were performed with maximum effort against external resistance, whereas the sit-ups were performed in a smooth and slow way with no additional external resistance-only the subjects' own body weight. Muscle instantaneous length is also a known modulator of muscle force. While we did not account for changing muscle length in this study, generally length changes are smaller in the torso muscles than in the extremities. Furthermore, muscle lengths are constant between comparable tasks such as the various types of flexion exercises, lifts, etc. Finally, all of the usual limitations of using fine wire electrodes pertain to this data such as monitoring only small portions of the entire muscle. For example, the psoas electrodes could not be placed in a specific layer of the muscle, as these could not be discriminated by ultrasound imagery. Future experiments should include EMG mapping to evaluate the possible importance of non-homogeneous electrical activity. Also, we were unable to prevent the possibility of wire migration during muscle contraction or to control the possibility of internal electrodes affective movement patterns (as has been seen in children walking(25)). Finally, the number of intramuscular EMG channels were restricted by the invasive nature of the study.
In conclusion, the normalized and scaled EMG data collected over a wide variety of tasks will assist clinicians, bioengineers, physical therapists, and fitness professionals in understanding the roles of these deep muscles to optimize rehabilitation and training programs.
1. Andersson, E., L. Oddsson, H. Grundstrom, and A. Thorstensson. The role of the psoas and iliacus muscles for stability and movement of the lumbar spine, pelvis and hip. Scand. J. Med. Sci. Sports
2. Basmajian, J. V. Electromyography
of iliopsoas.Anatom. Rec.
3. Bogduk, N., M. Pearcy, and G. Hadfield. Anatomy and biomechanics of psoas major. Clin. Biomech.
4. Buchthal, F. and H. Schmalbruch. Contraction times and fibre types of intact human muscle. Acta. Physiol. Scand.
5. Cholewicki, J. and S. M. McGill. Mechanical stability of the vivo lumbar spine: Implications for injury and chronic low back pain.Clin. Biomech.
6. Janda, V. Muscles, motor regulation and back problems. In: Neurobiologic Mechanisms in Manipulative Therapy,
I. M. Korr(Ed.). New York: Plenum Press, 1978, p. 27.
7. Janevic, J., J. A. Ashton-Miller, and A. B. Schultz. Large compressive pre-loads decrease lumbar motion segment flexibility.J. Orthop. Res.
8. Kapandji, I. A. Funktionelle Anatomie der Gelenke
. Enke, Stuttgart, 1992.
9. Keagy, R. D., J. Brumlik, and J. J. Bergan. Direct electromyography
of the psoas major muscle in man. J. Bone Joint Surg.
10. Laban, M. M., A. D. Raptou, and E. W. Johnson, Electromyographic Study of Function of Iliopsoas Muscle. Arch. Physiol. Med. Rehab.
11. Linden, P. and L. Delhez. Modalités d'activité electrique du psoas-iliaque en conditions statiques et dynamiques. Med. Phys.
12. McGill, S. M., N. Patt, and R. W. Norman. Measurement of the trunk musculature of active males using CT scan radiography: implications for force and moment generating capacity about the L4/L5 joint.J. Biomech.
13. McGill, S. M. and R. W. Norman. Dynamically and statically determined low back moments during lifting. J. Biomech.
14. McGill, S. M., D. Juker, and P. Kropf. Quantitative intramuscular
myoelectric activity of quadratus lumborum during a wide variety of tasks. Clin. Biomech.
15. McGill, S. M. The Biomechanics of low back injury: Implications on current practice in industry and the clinic (ISB keynote lecture). J. Biomech.
16. Milner-Brown, M. S., R. B. Stein, and R. Yemm. The contractile properties of human motor units during voluntary isometric contractions. J. Physiol.
17. Nachemson, A. Electromyographic studies on the vertebral portion of the psoas muscle. Acta Orthop. Scand.
18. Nachemson, A. The possible importance of the psoas muscle for stabilization of the lumbar spine. Acta. Orthop. Scand.
19. Norman, R. W., S. M. McGill, W. Lu, and M. Frazer. Improvements in biological realism in an industrial low back model: 3D WATBAK.Proc. 12th Triennial Congress of the International Ergonomics Association,
Vol. 2, Toronto, August 15-19, 1994, pp. 299-301.
20. Okamura, S. Electromyographic studies on muscular activities during spinal movement-especially the kinetic function of the psoas and iliacus muscles. Nichidai Igaku Zasshi
21. Olney, S. J. and D. A. Winter. Predications of knee and ankle moments of force in walking from EMG and kinematic data. J. Biomech.
22. Samuel, J., M. Revel, J. C. Andres, J. Jusserand. Etudeéléctrocinésiologique des muscles paravertébraux lombaires. Rev. Rhum.
23. Santaguida, P. L. and S. M. McGill. The Psoas major muscle: a three dimensional geometric study. J. Biomech.
24. Spring, H. K. Theorie und Praxis
. Stuttgart-New York: Thieme, 1990.
25. Young, C. C., S. E. Rose, E. N.Biden, M. P.Wyatt, D. M. Sutherland. The effect of surface and internal electrodes on the gait of children with cerebral palsey, spastic diplegic type. J. Orthop. Res.