An Electromyographic Analysis of

Commercial and Common Abdominal

Exercises: Implications for Rehabilitation
and Training

Associate Professor of Physical Therapy, California State University Sacramento, Sacramento, CA.
Graduate student (at the time of study), Duke University Medical Center, Durham, NC.
3
Student (at the time of study), Duke University Medical Center, Durham, NC.
4
Student (at the time of study), California State University Sacramento, Sacramento, CA.
5
Lieutenant, Medical Service Corps, US Navy, Naval Health Research Center, San Diego, CA.
6
Associate Professor of Kinesiology and Health Science, California State University Sacromento,
Sacromento, CA.
7
Associate Professor of Orthopaedic Surgery, Duke University Medical Center, Durham, NC.
The protocol used in this study was approved by the Institutional Review Board at Duke University
Medical Center, Durham, NC. The authors of this manuscript affirm we have no financial affiliation
(including research funding) or involvement with any commercial organization that has a direct financial
interest in any matter included in this manuscript.
Address correspondence to Rafael Escamilla, Associate Professor of Physical Therapy, California State
University Sacramento, Department of Physical Therapy, 6000 J Street, Sacramento, CA 95819-6020.
E-mail: rescamil@csus.edu
2

Journal of Orthopaedic & Sports Physical Therapy

Conclusions: The Ab Slide and Torso Track

were the most effective exercises in activating
abdominal and upper extremity muscles while
minimizing low back and rectus femoris (hip
flexion) activity. The Ab Doer, Ab Twister, Ab
Rocker, SAM, and bent-knee sit-up may be
problematic for individuals with low back pathologies due to relatively high rectus femoris
activity. J Orthop Sports Phys Ther 2006;36:4557.

Key words: EMG, low back pain,

lumbar spine, rectus abdominis,
sit-up

trong abdominals are important for stabilizing

the trunk and helping
unload stress in the lumbar spine.3,13 Abdominal
muscles (rectus abdominis, external oblique, internal oblique, and
transverse abdominal) are commonly strengthened by actively
flexing the trunk with a concentric
muscle action or by resisting trunk
extension (due to an external
force such as gravity) with an isometric or eccentric muscle action.
45

REPORT

Study Design: A repeated-measures, counterbalanced design.

Objectives: To test the effectiveness of 7 commercial abdominal machines (Ab Slide, Ab Twister,
Ab Rocker, Ab Roller, Ab Doer, Torso Track, SAM) and 2 common abdominal exercises (crunch,
bent-knee sit-up) on activating abdominal and extraneous (nonabdominal) musculature.
Background: Numerous abdominal machine exercises are believed to be effective in activating
abdominal musculature and minimizing low back stress, but there are minimal data to substantiate
these claims. Many of these exercises also activate nonabdominal musculature, which may or may
not be beneficial.
Methods and Measures: A convenience sample of 14 subjects performed 5 repetitions for each
exercise. Electromyographic (EMG) data were recorded for upper and lower rectus abdominis,
external and internal oblique, pectoralis major, triceps brachii, latissimus dorsi, lumbar
paraspinals, and rectus femoris, and then normalized by maximum muscle contractions.
Results: Upper and lower rectus abdominis EMG activities were greatest for the Ab Slide, Torso
Track, crunch, and Ab Roller, while external and internal oblique EMG activities were greatest for
the Ab Slide, Torso Track, crunch, and bent-knee sit-up. Pectoralis major, triceps brachii, and
latissimus dorsi EMG activities were greatest for the Ab Slide and Torso Track. Lumbar paraspinal
EMG activities were greatest for the Ab Doer, while rectus femoris EMG activities were greatest for
the bent-knee sit-up, SAM, Ab Twister, Ab Rocker, and Ab Doer.

RESEARCH

Rafael F. Escamilla, PT, PhD, CSCS 1

Michael S.C. McTaggart, MS 2
Ethan J. Fricklas, MSE 3
Ryan DeWitt, MPT 4
Peter Kelleher, MPT 4
Marcus K.Taylor, PhD 5
Alan Hreljac, PhD 6
Claude T. Moorman, III, MD 7

There are numerous exercises used for abdominal

strengthening. Many of these exercises also activate
extraneous (nonabdominal) muscles, such as the hip
flexors, lumbar paraspinals, or upper extremity musculature, which may or may not be beneficial. For
example, high activation levels from the hip flexors
and lumbar paraspinals tend to generate a force
couple that attempts to anteriorly rotate the pelvis
and increase lumbar lordosis. When coupled with
weak abdominal musculature, activation of these extraneous muscles may increase the risk of low back
pathologies.
Understanding the muscle activation generated by
different abdominal exercises is useful to therapists
and other health care or fitness specialists who
develop specific abdominal exercises for their patients or clients to facilitate their rehabilitation or
training needs and objectives. For example, abdominal exercises that actively flex the trunk may be
problematic for individuals with lumbar disk pathologies due to increased intradiscal pressure18 and lumbar spine compression,3 and for individuals with
osteoporosis due to the risk of vertebral compression
fractures.21 However, these same individuals may be
asymptomatic during abdominal exercises that maintain a relatively neutral spine and pelvis. In contrast,
individuals with facet joint syndrome, spondylolisthesis, and vertebral or intervertebral foramen
stenosis may not tolerate exercises such as the Ab
Slide and Torso Track due to the extended spine
position.
There are numerous commercially available abdominal machines that are believed to be effective in
activating abdominal musculature and minimizing
low back stress, but there are little or no scientific
research data to substantiate these beliefs. While
there are numerous studies that examined muscle
activity during more traditional abdominal exercises,
such as the sit-up or crunch exercises,3,14,16,26,27 there
is a scarcity of data related to the use of abdominal
machines. A limited number of studies compared
select abdominal muscle activity while performing
exercises using the Torso Track, Ab Doer, Ab Shaper,
Ab Flex, and Ab Roller,6,7,12,24,26 but there are no
studies that we are aware of that have quantified
abdominal muscle activity while using the Ab Twister,
Ab Rocker, Super Abdominal Machine (SAM), and
Ab Slide. Moreover, when using abdominal machines,
the extent of recruitment of extraneous musculature,
such as low back or upper and lower extremity
musculature, is currently unknown because there
have been no studies that have reported extraneous
muscle activity while performing these exercises. It is
also unknown how abdominal machines compare to
traditional abdominal exercises, such as the sit-up and
crunch, in recruiting the abdominal musculature.
Abdominal machines use various techniques to
target different muscles. For example, some abdomi46

FIGURE 1. Ab Rocker.

FIGURE 2. Ab Roller.

nal machines allow only uniplanar motions, such as

trunk flexion, while others use multiplanar motions,
such as trunk flexion and rotation. It is commonly
believed that performing simultaneous trunk flexion
and rotation recruits the external and internal oblique musculature to a greater extent compared to
trunk flexion only. However, there is currently no
scientific evidence to support this assertion.
J Orthop Sports Phys Ther Volume 36 Number 2 February 2006

The purpose of this study was to test the effectiveness of 7 popular commercial abdominal machines
and 2 common abdominal strengthening exercises on
activating abdominal and extraneous musculature. It
was hypothesized that significant differences would be
found in the normalized electromyographic (EMG)
data of both abdominal and extraneous muscle activity among exercises.

METHODS
Subjects

FIGURE 4. Ab Twister.

RESEARCH

To optimize the EMG signal, this study was limited

to a convenience sample of 14 healthy, young subjects
(7 male and 7 female) who had normal or below
normal amounts of body fat for their age group.
Baseline skinfold calipers (model 68900; Country
Technology, Inc, Gays Mill, WI) and appropriate
regression equations were used to assess percent body
fat, and standards set by the American College of
Sports Medicine were used to determine normal or
below normal amounts of body fat.4 Mean (SD) age,
mass, height, and percent body fat were 24.1 5.4
years, 58.7 4.9 kg, 166.8 5.9 cm, and 22.7%
1.9%, respectively, for females, and 24.0 7.1 years,
78.6 13.9 kg, 179.8 4.1 cm, and 9.7% 4.1%,
respectively, for males. All subjects provided written
informed consent in accordance with the Institutional
Review Board at Duke University Medical Center.

REPORT

FIGURE 5. Torso Track.

FIGURE 6. Ab Slide.

Individuals were excluded from the study if they had

a history of abdominal or back pain, or were unable
to perform all exercises pain free and with proper
form and technique for 12 consecutive repetitions.

Exercise Descriptions

FIGURE 3. Ab Doer.
J Orthop Sports Phys Ther Volume 36 Number 2 February 2006

The 7 abdominal machine exercises were the Ab

Rocker (Figure 1), Ab Roller (Figure 2), Ab Doer
(Figure 3), Ab Twister (Figure 4), Torso Track
(Figure 5), Ab Slide (Figure 6), and SAM (Figure 7).
The 2 common abdominal exercises tested were the
47

bent-knee sit-up (Figure 8) and crunch (Figure 9).

No subject had prior experience in performing the 7
commercial abdominal exercises, but they had moderate experience in performing the crunch and
bent-knee sit-up.
The Ab Rocker, Ab Twister, Ab Doer, and SAM
exercises started and ended in a seated position with
a neutral spine and pelvis. The 2 common movements advertised for the Ab Rocker and Ab Twister
were the crunch (involving sagittal plane trunk flexion) and the oblique crunch (moving obliquely
across the body by simultaneously flexing and rotating the trunk), and both were tested with rotation
occurring to the left. Three common movements
advertised for the Ab Doer were the body bob
(frontal plane side-to-side motion), body boogie
(moving in a circular motion), and good morning
(involving sagittal plane trunk flexion), and all 3 of
these variations were tested. The movement for the
SAM involved sagittal plane trunk flexion, similar to
how the Ab Rocker (crunch), Ab Twister (crunch),
and Ab Doer (good morning) were performed.
The Ab Roller, crunch, and bent-knee sit-up started
and ended in a supine position. The crunch and Ab
Roller both had 2 variations: a normal crunch involving sagittal plane trunk flexion and an oblique
crunch involving moving obliquely across the body by

FIGURE 7. SAM.
48

FIGURE 8. Bent-knee sit-up.

FIGURE 9. Crunch.

simultaneously flexing and rotating the trunk to the

left. The primary differences between the 2 exercises
were that during the Ab Roller the head was supported by a head pad and the arms were supported
by a supporting bar (Figure 2), while during the
crunch the thumbs were positioned in the ears and
the hands were relaxed against the head (this hand
position was standardized for comfort for both the
crunch and bent-knee sit-up) (Figure 9). Both variations for the crunch and Ab Roller involved a
curling-up motion (trunk flexion or trunk flexion
with left rotation) until both scapulae were off the
ground. During the bent-knee sit-up the thumbs were
positioned in the ears with the hands relaxed against
the head, the feet were supported and held down,
the knees were flexed approximately 90, and from
this supine position the subject simultaneously flexed
the trunk and hips until the elbows were even with
the knees (Figure 8).
The Ab Slide and Torso Track started and ended in
the quadruped position (on hands and knees with
hips and shoulders flexed approximately 90), with a
neutral spine and pelvis. From this position the
subject straightened out the body by rolling forward
in a straight line (Torso Track and Ab Slide straight)
or a curved line to the left (Ab Slide curved), while
maintaining a neutral spine and pelvis (Figures 5 and
6).
J Orthop Sports Phys Ther Volume 36 Number 2 February 2006

Procedures

49

REPORT

J Orthop Sports Phys Ther Volume 36 Number 2 February 2006

RESEARCH

All subjects became familiar with all abdominal

exercises during a pretest session that took place
approximately 1 week prior to the testing session.
During the pretest session, each subject received
instructions explaining how to correctly perform
each of the abdominal exercises (each abdominal
machine came with written or video instructions for
its use). All exercises were performed with a 3-second
cadence (1 second from start of exercise to end
range, 1-second isometric hold at end range,
1-second return to starting position) and a 1-second
rest between repetitions. The subjects practiced all
exercises under the supervision of trained research
personnel. All of the commercial exercises except the
Ab Roller and Ab Slide had adjustable elastic bands
to make the exercise easier or harder. To better
normalize the intensity of each exercise, resistance
was adjusted according to each subjects preference
and the manufacturers recommendation (eg, using a
resistance that was not too hard but hard enough to
allow the execution of at least 15 repetitions)
similar to how each subject would adjust the resistance and use the equipment if they purchased it for
home use. The selected resistance during the pretest
session was also used for that subject during the
testing session. For each exercise, each subject used a
resistance that enabled the subject to correctly perform at least 15 consecutive repetitions using the
3-second cadence described above. It was not possible
to normalize all exercises with exactly the same
relative intensity because the Ab Roller, Ab Slide,
crunch, and bent-knee sit-up used the body only as
an external resistance, while the remaining exercises
used resistance bands in addition to body as external
resistance. In addition, even when the maximum
resistance possible was used for the Ab Doer, Ab
Twister, and Ab Rocker, all subjects indicated that
they were capable of performing these exercises with
more resistance. A metronome (set at 1 beat per
second) was used to help ensure proper cadence
both during the pretest and testing sessions. Once a
subject was able to correctly perform each exercise
with the proper cadence, a testing session was scheduled.
Neuroline (Medicotest Marketing, Inc, Ballwin,
MO) disposable surface electrodes (type 720-00-S)
were used to collect EMG data. These oval-shaped
electrodes (22 mm wide and 30 mm long) were
placed in a bipolar electrode configuration along the
longitudinal axis of each muscle, with a center-tocenter distance of approximately 3 cm between electrodes. Prior to positioning the electrodes over each
muscle, the skin was prepared by shaving, abrading,
and cleaning with isopropyl alcohol wipes to reduce
skin impedance values, which typically were less than
10 k. Electrode pairs were then placed on the

subjects right side (except for the internal oblique,

which was positioned on the subjects left side) for
the following muscles in accordance with procedures
previously described 5,8,17,20 : (a) upper rectus
abdominis, positioned vertically and centered on the
muscle belly (not on tendinous intersection) near the
midpoint between umbilicus and xiphoid process and
3 cm lateral from midline; (b) lower rectus
abdominis, positioned 8 from vertical in
inferomedial direction and centered on the muscle
belly near the midpoint between umbilicus and pubic
symphysis and 3 cm lateral from midline; (c) external
oblique, positioned obliquely approximately 45 (parallel to a line connecting the most inferior point of
the costal margin of the ribs and the contralateral
pubic tubercle) above anterior superior iliac spine
(ASIS) near the level of the umbilicus; (d) internal
oblique, positioned horizontally 2 cm inferomedial to
the ASIS, within a triangle confined by the inguinal
ligament, lateral border of the rectus sheath, and a
line connecting the ASISs (it has been demonstrated
that in this region only the aponeurosis of the
external oblique, and not the external oblique
muscle, covers the internal oblique)20; (e) sternal
pectoralis major, positioned horizontally 2 cm medial
to the axillary fold; (f) triceps brachii long head,
positioned vertically over the long head muscle belly
near midline of the arm approximately halfway between the acromion and olecranon; (g) latissimus
dorsi, positioned obliquely (approximately 25 from
horizontal in an inferomedial direction) 4 cm below
inferior angle of the scapula; (h) rectus femoris,
positioned vertically near midline of thigh approximately halfway between ASIS and proximal patella;
and (i) lumbar paraspinals, positioned vertically 3 cm
lateral to spine and near level of iliac crest between
L3 and L4 vertebrae. A ground (reference) electrode
was positioned over the skin of the right acromion
process. Electrode cables were connected to the
electrodes and taped to skin appropriately to minimize pull on the electrodes and movement of the
cables.
Once the electrodes were positioned, the subject
warmed up and practiced the exercises as needed,
then data collection commenced. EMG data were
collected using a Noraxon 16-channel telemyo EMG
unit (Noraxon USA, Inc, Scottsdale, AZ), and the
amplifier bandwidth frequency was 10 to 500 Hz. The
input impedance of the amplifier was 20 000 k and
the common-mode rejection ratio was 130 dB. EMG
data were sampled at 1000 Hz, recorded by a
transmitter and amplifier, and broadcast to a receiver
interfaced to a computer. The recorded signals were
processed through an analog-to-digital (A/D) converter by a 16-bit A/D board.
EMG data were collected during 5 repetitions for
each exercise, with all exercises performed in a
randomized order. Each repetition was performed in

a slow and controlled manner using the 3-second

cadence previously described and 1-second rest between repetitions. With a relatively low number of
repetitions performed, all subjects acknowledged that
fatigue was minimized. Each testing session took
approximately 45 minutes to complete.
Randomly interspersed within the exercise testing
session, EMG data from each muscle tested were
collected during two 5-second maximum voluntary
isometric contractions (MVICs). After conducting pilot work, we adopted the following protocols for
MVIC testing, which were based on the positions that
elicited the greatest MVIC for each respective muscle
(all MVICs were collected on a plinth with subject in
a prone, supine, or short-sitting position): (a) upper
and lower rectus abdominis, body supine with hips
and knees flexed 90, feet supported, and trunk
maximally flexed (ie, curl-up position), with resistance at the shoulders in the trunk extension direction; (b) external and internal oblique, body supine
with hips and knees flexed 90, feet supported, and
trunk maximally flexed and rotated to the left, with
resistance at the shoulders in the trunk extension
and right rotation directions; (c) sternal pectoralis
major, body supine with right shoulder flexed 90
and internally rotated, the right forearm supinated,
and the right elbow slightly flexed, with resistance at
the right distal arm and forearm in the horizontal
abduction direction; (d) triceps long head, body
prone with right shoulder abducted 90 and right
elbow flexed 45, with resistance at the right distal
forearm in the elbow flexion direction; (e) latissimus
dorsi, body prone with right shoulder abducted 0
and extended maximally, with resistance at the right
distal arm in the direction of shoulder flexion; (f)
lumbar paraspinals, body prone with trunk fully
extended and hands clasped behind head, with resistance at the shoulders in the direction of trunk
flexion; and (g) rectus femoris, body in short-sitting
position with hips and knees flexed 90, with resistance at the distal leg in the knee flexion direction.
The MVICs were collected to normalize the EMG
data collected during the abdominal exercises. Each
subject was given similar verbal encouragement for
each MVIC to help ensure a maximum effort
throughout the 5-second duration, and the subject
was asked after each MVIC if he/she felt it was a
maximum effort. If not, the MVIC was repeated. An
approximately 1-minute rest was given between each
MVIC and an approximately 2-minute rest was given
between each exercise trial.

Data Processing
Raw EMG signals were full-wave rectified, smoothed
with a 10-millisecond moving average window, and
linear enveloped, then averaged over the entire
duration of each exercise repetition. For each repetition the EMG data were normalized for each muscle
50

and expressed as a percentage of a subjects highest

corresponding MVIC trial, which was determined by
calculating throughout the 5-second MVIC the highest average EMG signal over a 1-second time interval.
Normalized EMG data were then averaged over the 5
repetition trials performed for each exercise and
used in statistical analyses.

Data Analysis
A 1-factor repeated-measures analysis of variance
was employed to assess differences in normalized
EMG muscle activity among the different exercise
variations (P .01). Post hoc analyses were performed
using the Bonferroni test to evaluate the significance
of between-exercise pairwise comparisons (P.01).

RESULTS
Normalized EMG data for each muscle and exercise are shown in Table 1. Among all exercises tested,
upper rectus abdominis EMG activities were greatest
for the Ab Slide (straight and curved), Torso Track,
crunch (normal and oblique), and Ab Roller (crunch
and oblique) exercises, and lowest for the Ab Twister
(crunch and oblique), Ab Rocker (crunch and oblique), and Ab Doer (good morning, body boogie,
and body bob) exercises. Lower rectus abdominis
EMG activities were greatest for the Ab Slide (straight
and curved) and Torso Track exercises, and lowest
for the Ab Twister (crunch and oblique), Ab Rocker
(crunch and oblique), and Ab Doer (good morning,
body boogie, and body bob) exercises. Graphical
representations of upper and lower rectus abdominis
activity ranked from highest to lowest among all
exercises are shown in Figures 10 and 11.
The external oblique EMG activity for the crunch
(normal), Ab Roller (crunch), and Ab Doer (good
morning) exercises were significantly lower compared
to the Ab Slide (straight and curved) and bent-knee
sit-up exercises. Internal oblique EMG activities were
greatest for the Ab Slide (straight and curved), Torso
Track, bent-knee sit-up, and crunch (normal and
oblique) exercises, and lowest for the Ab Roller
(oblique), Ab Twister (crunch and oblique), Ab
Rocker (crunch and oblique), and Ab Doer (good
morning) exercises. Graphical representation of external and internal oblique activity ranked from
highest to lowest among all exercises are shown in
Figures 12 and 13.
Sternal pectoralis major EMG activities were greatest for the Ab Slide (straight and curved), Torso
Track, SAM, and Ab Twister (crunch and oblique)
exercises, and lowest for the Ab Rocker (crunch and
oblique), Ab Doer (good morning, body boogie, and
body bob), Ab Roller (crunch and oblique), and
crunch (normal and oblique) exercises. Triceps
brachii long head EMG activities were significantly
J Orthop Sports Phys Ther Volume 36 Number 2 February 2006

67 26
61 24
67 25
51 9
50 15
38 12ab
42 17ab
46 17
49 12
19 8abcde
20 7abcde
15 8abcdefg
14 10abcdefg
14 7abcdefg
12 4abcdefg
7 5abcdefg

72 19
66 19
72 17
50 8ab
39 14ab
44 13ab
50 20ab
42 12ab
36 16ab
19 10abcdefg
22 11abcfg
13 5abcdefg
14 8abcdefg
14 5abcdefg
11 6abcdefg
7 4abcdefg

Lower Rectus
Abdominis*
40
42
32
16
32
41
31
13
20
21
33
22
31
16
24
30

16
17
18
11af
22
16
21
8af
9
12
18
11
18
11af
10
19

External
Oblique*
53
51
58
41
40
49
36
38
25
22
28
24
23
22
31
37

15
15
14
9
11
21
13b
9b
11abf
9abf
11abf
8abf
8abf
13abf
13b
18b

Internal
Oblique*

REPORT

23 7
20 9
20 8
4 3abgi
6 5agi
8 6ag
26 15
7 5agi
5 3abgi
13 11
22 15
7 7agi
6 6agi
4 4abgi
3 2abgi
2 2abgi

Sternal
Pectoralis
Major*
30 12
26 12
26 11
1 1ab
2 2ab
2 2ab
10 6ab
3 2ab
3 2ab
7 3ab
5 4ab
6 4ab
7 4ab
2 1ab
1 1ab
2 1ab

Triceps
Long Head*

* Significant difference (P .001) in EMG activity among abdominal exercises based on a 1-way repeated-measures analysis of variance.
Key to pairwise comparisons (P .01):
a. Significantly less EMG activity compared to the Ab Slide (straight and curved).
b. Significantly less EMG activity compared to the Torso Track.
c. Significantly less EMG activity compared to the crunch (normal).
d. Significantly less EMG activity compared to the crunch (oblique).
e. Significantly less EMG activity compared to the Ab Roller (crunch and oblique).
f. Significantly less EMG activity compared to the bent-knee sit-up.
g. Significantly less EMG activity compared to the SAM.
h. Significantly less EMG activity compared to the Ab Twister (crunch).
i. Significantly less EMG activity compared to the Ab Twister (oblique).
j. Significantly less EMG activity compared to the Ab Rocker (crunch).
k. Significantly less EMG activity compared to the Ab Doer (good morning and body boogie).

Ab Slide (straight)
Ab Slide (curved)
Torso Track
Crunch (normal)
Crunch (oblique)
Bent knee sit-up
SAM
Ab Roller (crunch)
Ab Roller (oblique)
Ab Twister (crunch)
Ab Twister (oblique)
Ab Rocker (crunch)
Ab Rocker (oblique)
Ab Doer (good morning)
Ab Doer (body boogie)
Ab Doer (body bob)

Upper Rectus
Abdominis*

TABLE 1. Average EMG (SD) for each muscle and exercise expressed as a % of maximum isometric voluntary contraction.

RESEARCH

J Orthop Sports Phys Ther Volume 36 Number 2 February 2006

51

10 4
10 3
10 5
5 1g
85
6 3g
12 6
5 2g
6 2g
5 2g
6 2g
6 3g
5 2g
2 2abg
2 1abg
1 1abg

Latissimus
Dorsi*
3 2k
2 2k
2 2k
2 1k
5 3k
4 2k
4 2k
3 2k
3 2k
4 3k
5 6k
4 3k
3 1k
15 7
13 8
83

Lumbar
Paraspinals*
5 3fhj
9 7f
6 5fhj
3 2fhj
3 2fhj
36 16
20 15
1 1fhj
2 2fhj
27 19
24 14
30 21
21 16
12 11f
24 19
16 14f

Rectus
Femoris*

100
90
Normalized EMG (%MVIC)

80
70
60
50
40
30
20
10

SA
nt
-k
M
ne
e
Tw
si
t-u
is
te
p
r
Ab
(o
b
Tw
liq
ue
is
te
)
r(
Ab
cr
R
un
oc
ch
ke
)
Ab
r(
c
R
r
un
oc
Ab
ch
ke
D
)
r
oe
(o
r(
bl
iq
go
ue
Ab
od
)
m
D
oe
or
n
r(
in
bo
g)
d
y
Ab
bo
D
og
oe
ie
r(
)
bo
dy
bo
b)
Ab

Be

ed
)
(n
or
C
ru
m
nc
al
)
h
(o
Ab
bl
R
iq
ol
ue
le
)
r(
Ab
ob
l
i
R
qu
ol
e)
le
r(
cr
un
ch
)
ch

C
ru
n

(c
u

rv

Tr
a
Sl
id

To

rs
o
Ab

Ab

Sl
id
e

(s
t

ra
i

gh

t)

ck

FIGURE 10. Upper rectus abdominis normalized mean (SD) EMG activity among exercises.
100
90
Normalized EMG (%MVIC)

80
70
60
50
40
30
20
10

Ab

Sl
id
e

(s
tra
ig
ht
To
)
rs
o
Ab
Tr
Sl
ac
id
k
e
(c
C
ur
ru
ve
nc
d)
h
(n
or
m
al
)
Be
SA
nt
-k
M
n
ee
Ab
si
R
t-u
ol
le
p
r(
cr
C
un
ru
nc
ch
h
)
(o
Ab
b
liq
R
ol
ue
le
)
Ab
r(
ob
Tw
liq
is
ue
te
)
r(
Ab
ob
Tw
liq
is
ue
te
)
Ab
r(
cr
R
un
oc
Ab
ch
ke
D
)
r
oe
(o
r(
bl
iq
go
ue
od
)
Ab
m
or
R
ni
oc
ng
Ab
ke
)
r(
D
oe
cr
un
r(
ch
bo
)
dy
Ab
b
D
o
og
oe
ie
r(
)
bo
dy
bo
b)

FIGURE 11. Lower rectus abdominis normalized mean (SD) EMG activity among exercises.

greater for the Ab Slide (straight and curved) and

Torso Track exercises compared to all other exercises.
Latissimus dorsi EMG activities were greatest for the
Ab Slide (straight and curved), Torso Track, SAM,
crunch (oblique), and Ab Twister (oblique) exercises,
and lowest for the Ab Doer (good morning, body
boogie, and body bob) and Ab Roller (crunch)
exercises.
Lumbar paraspinal EMG activities were significantly
greater for the Ab Doer (good morning, body
52

boogie, and body bob) exercises compared to all

other exercises. Rectus femoris EMG activities were
greatest for the bent-knee sit-up, SAM, Ab Twister
(crunch and oblique), Ab Rocker (crunch and oblique), and Ab Doer (body boogie) exercises, and
lowest for the Ab Roller (crunch and oblique), Ab
Slide (straight), Torso Track, and crunch (normal
and oblique) exercises. The relative effectiveness of
exercises in muscle recruitment of the trunk, upper
extremity, and hip musculature is shown in Table 2.
J Orthop Sports Phys Ther Volume 36 Number 2 February 2006

70

Normalized EMG (%MVIC)

60
50
40
30
20
10

Ab

Sl

id
e

(c
ur
Be
ve
nt
d)
-k
ne
e
Ab
si
Sl
t-u
id
p
e
Ab
(s
t
r
Tw
ai
gh
is
te
t)
r(
ob
liq
ue
To
)
rs
o
C
Tr
ru
ac
nc
k
h
(o
bl
iq
ue
)
Ab
R
SA
oc
ke
M
r(
Ab
ob
D
liq
oe
ue
r(
Ab
)
b
od
D
oe
y
r(
bo
bo
b)
d
y
Ab
bo
R
og
oc
ie
ke
)
r(
Ab
cr
un
Tw
ch
is
te
)
r
(c
Ab
ru
R
nc
ol
h)
le
r(
ob
C
liq
ru
ue
Ab
nc
)
D
h
oe
(n
or
r(
m
go
al
od
)
m
Ab
or
ni
R
ng
ol
le
)
r(
cr
un
ch
)

FIGURE 12. External oblique normalized mean (SD) EMG activity among exercises.

RESEARCH

80

60
50
40

REPORT

Normalized EMG (%MVIC)

70

30
20
10

Ab

Sl
id
e
Ab
(c
ur
Sl
ve
id
d)
e
(s
tra
ig
To
ht
)
rs
Ab
o
R
T
ol
ra
le
ck
r(
ob
C
ru
liq
nc
ue
h
)
(
no
C
ru
rm
nc
al
h
)
(o
bl
iq
ue
Ab
)
D
oe
SA
r
M
Ab (bo
dy
R
bo
Ab
ol
le
b)
D
r(
oe
c
ru
r(
nc
bo
h)
Ab
dy
bo
Tw
og
is
te
ie
)
r(
cr
Be
u
nt
nc
-k
h)
Ab
ne
e
R
si
oc
t-u
ke
Ab
p
r(
cr
R
oc
un
ch
Ab ker
)
(
ob
Tw
Ab
liq
is
ue
te
D
oe
r(
)
o
r(
bl
go
iq
ue
od
)
m
or
ni
ng
)

FIGURE 13. Internal oblique normalized mean (SD) EMG activity among exercises.

DISCUSSION
Biomechanical Differences Between Flexion and
Extension Exercises
The Ab Slide and Torso Track were the most
effective exercises in activating abdominal musculature, including the upper and lower rectus abdominis
and the external and internal oblique. While performing these exercises, the abdominal muscles contract in a different manner compared to performing
traditional trunk flexion exercises. During the rollJ Orthop Sports Phys Ther Volume 36 Number 2 February 2006

out portion in performing the Ab Slide and Torso

Track, the abdominal musculature contracts eccentrically or isometrically to resist the attempt of gravity to
extend the trunk and rotate the pelvis. During the
return motion, the abdominal musculature contracts
concentrically or isometrically. If the pelvis and spine
are stabilized and maintained in a neutral position
throughout the roll-out and return movements, then
the abdominal musculature primarily would contract
isometrically. While performing these exercises, a
relatively neutral pelvis and spine were maintained
throughout the movement. In contrast, all other
53

exercises activated abdominal musculature by actively

flexing the trunk by concentric contractions during
the initial portion of the motion, an isometric contraction during the middle portion, and an eccentric
contraction during the final portion of the motion.
Understanding biomechanical differences between
exercises is important because trunk flexion may be
contraindicated in certain populations, such as those
with lumbar disk pathologies or osteoporosis. Maintaining a neutral pelvis and spine (such as performing the Ab Slide or Torso Track exercise), rather
than forceful flexion of the lumbar spine (such as
during the bent-knee sit-up), may be desirable for
these individuals. In contrast, an individual with facet
joint pain, spondylolisthesis, and vertebral or
intervertebral foramen stenosis may not benefit from
exercises that maintain the spine and pelvis in a
neutral or extended position, such as when using the
Ab Slide and Torso Track. These exercises may in fact
contribute to the nerve compression. However, trunk
flexion exercises, such as the crunch, bent-knee
sit-up, SAM, Ab Roller, Ab Twister, Ab Rocker, and Ab
Doer may be beneficial.
When the lumbar spine is forcefully flexed, which
may occur when performing many of the commercial
abdominal machines used in the current study, the

anterior fibers of the intervertebral disk are compressed, while the posterior fibers are in tension. In
addition, in extreme lumbar flexion intradiscal pressure may increase several times above the normal
intradiscal pressure from a resting supine position.18,19 While these stresses on the disk may not be
problematic for the normal healthy disk, they may be
detrimental to the degenerative disk or pathologic
spine.
There have only been a few studies that have
compared abdominal machine exercises to the traditional crunch or bent-knee sit-up exercises.6,7,12,24,26
Most of these studies compared the crunch to the Ab
Roller, and like the results of the current study, there
were generally no significant differences in abdominal muscle activity between these 2 exercises. The
biggest difference between these exercises is that the
Ab Roller provides head support, which may make it
more comfortable to perform compared to the
crunch. The only known study to investigate abdominal muscle activity between the crunch and the Torso
Track and Ab Doer (good morning) was by Sternlicht
and Rugg,24 and these authors found similar results
as the current study: that abdominal muscle activity
was significantly greater in the crunch and Torso
Track compared to the Ab Doer (good morning).

TABLE 2. Relative muscle recruitment of the trunk, upper extremity, and hip musculature. Note: the Ab Slide (straight and curved) and
Torso Track were the exercises that produced the greatest activation of the abdominal, oblique, and upper extremity musculature, while
only minimally recruiting the hip flexors.
Abdominal and
Oblique Muscles

Upper Extremity
Muscles

Low Back Muscles

Hip Flexor Muscles

Greatest recruitment

Ab Slide (straight and

curved)
Torso Track

Ab Slide (straight and

curved)
Torso Track
SAM

Ab Doer (good morning and body boogie)

Bent knee sit-up

Ab Rocker (crunch)
Ab Twister (crunch)
Ab Doer (body boogie)
Ab Twister (oblique)

Intermediate recruitment

Crunch (normal and

oblique)
Bent-knee sit-up
SAM
Ab Roller (crunch and
oblique)

Ab Twister (crunch
and oblique)
Ab Rocker (crunch
and oblique)
Crunch (oblique)
Bent-knee sit-up
Ab Roller (crunch and
oblique)

Ab Doer (body bob)

Ab Rocker (oblique)
SAM
Ab Doer (body bob)
Ab Doer (good morning)

Least recruitment

Ab Twister (crunch
and oblique)
Ab Rocker (crunch
and oblique)
Ab Doer (good morning, body boogie, and
body bob)

Crunch (normal)
Ab Doer (good morning, body boogie, and
body bob)

Ab Twister (crunch
and oblique)
Ab Rocker (crunch
and oblique)
Ab Roller (crunch and
oblique)
Bent-knee sit-up
Crunch (normal and
oblique)
SAM
Ab Slide (straight and
curved)
Torso Track

Ab Slide (straight and

curved)
Torso Track
Crunch (normal and
oblique)
Ab roller (crunch and
oblique)

54

J Orthop Sports Phys Ther Volume 36 Number 2 February 2006

Biomechanical Differences Between the Crunch and

Bent-Knee Sit-up

The role of the abdominal muscles, especially the

transverse abdominal and internal oblique, in enhancing spinal and pelvic stabilization and increasing
intra-abdominal pressure (IAP) has been well studied,
but still remains controversial.9,10,15,22,25 IAP has been
shown to unload the spine by generating a trunk
extensor moment and tensile loading to the spine.11
By making the trunk a more solid cylinder by the IAP
mechanism, there is a reduction in spinal axial
compression and shear loads. The attachments of the
transverse abdominal and internal oblique into the
thoracolumbar fascia may enhance spinal and pelvic
stabilization, because when these muscles contract
they tense the thoracolumbar fascia. The transverse
abdominal, which is the deepest of the 4 abdominal
muscles, has been shown to exhibit a similar (within
15%) muscle activation pattern and amplitude as the
internal oblique muscle during many of the same
trunk flexion movements (eg, bent-knee sit-up,
crunch) used in the current study.16,17 The highest
EMG activities from the internal oblique were for the
Ab Slide, Torso Track, crunch, and bent-knee sit-up
exercises, which implies that these exercises may offer
more effective stabilization to the spine and pelvis
compared to the other exercises.
J Orthop Sports Phys Ther Volume 36 Number 2 February 2006

Exercise Intensity
The Ab Slide and Ab Roller were the only 2
commercial exercises in which resistance could not
be adjusted. This may account for more moderate
amounts of muscle activity in the Ab Roller (because
resistance could not be added to make it harder) and
higher amounts of muscle activity in the Ab Slide
(because there was no way to make it easier).
However, the Torso Track, which is performed in the
same manner as the Ab Slide and did have resistance
bands that could be adjusted to make it easier or
harder, had nearly identical muscle activity compared
to the Ab Slide. The subjects used in the current
study were all relatively young, active individuals who
all used the Torso Track in a more difficult resistance
setting. This more difficult resistance may be appropriate for younger more active individuals, but older,
less active, or weaker individuals may not be able to
correctly perform the Ab Slide due to its difficulty
level. In addition, all subjects set the resistance for
the Ab Doer, Ab Rocker, and Ab Twister to the
maximum number of resistance bands that could fit
on each device. Even with maximal resistance, these 3
commercial abdominal devices recorded the lowest
amount of abdominal activity. In contrast, the Ab
Slide, Ab Roller, and Torso Track generated significantly greater abdominal and oblique muscle activity
compared to the Ab Doer, Ab Rocker, and Ab Twister.

Extraneous (Nonabdominal) Muscle Activity

There are no studies that we are aware of that have
reported extraneous muscle activity for abdominal
machine exercises. Of the exercises tested, the Ab
Slide and Torso Track produced the greatest activation of the upper extremity musculature, including
the sternal pectoralis major, triceps brachii, and
55

REPORT

The Role of Abdominal Muscles in Trunk Stability

Despite slightly greater external oblique EMG activities in oblique and curved techniques, mean
abdominal and oblique EMG activities generally were
not significantly different between technique variations for exercises such as the Ab Slide, crunch, Ab
Roller, Ab Twister, and Ab Rocker (eg, normal
crunch versus oblique crunch, straight Ab Slide
versus curved Ab Slide). Because simultaneous trunk
flexion and rotation have been shown to increase the
risk of torsional injury to the annulus fibrosis of the
intervertebral disk, as well as generate relatively high
lumbar compressive forces,3 and because abdominal
and oblique EMG activities were generally not different between uniplanar and multiplanar trunk movements, the additional risks involved when performing
multiplanar trunk flexion and rotation motions are
not warranted for individuals who have lumbar disk
pathologies.

RESEARCH

It should be noted that not all abdominal exercises

involve the same degree of flexion of the lumbar
spine. Halpern and Bleck14 have demonstrated that
lumbar spinal flexion was only 3 during the crunch
but approximately 30 during the bent-knee sit-up. In
addition, the bent-knee sit-up has been shown to
generate greater intradiscal pressure18,19 and lumbar
compression3 compared to exercises similar to the
crunch, largely due to increased lumbar flexion and
hip flexor activity. This implies the crunch may be a
safer exercise to perform than the bent-knee sit-up
for individuals who need to minimize lumbar spinal
flexion or compressive forces due to lumbar pathology.
Although the crunch and bent-knee sit-up were
both effective in recruiting abdominal musculature,
there were some differences. Several studies, including the current study, have shown that external
oblique activity, and to a lesser extent internal oblique activity, are significantly greater in the bentknee sit-up compared to the crunch.1-3,16 However,
upper and lower rectus abdominis activities have
been shown to be greater in the crunch compared to
the bent-knee sit-up.6,14 In addition, like the current
study, hip flexor activity has been shown to be greater
in the bent-knee sit-up compared to the crunch.3,16

Technique Variations

latissimus dorsi. Both the sternal pectoralis major

(lower fibers) and latissimus dorsi contract eccentrically during the initial roll-out phase to control the
rate of shoulder flexion due to gravity, and concentrically in the return phase as the shoulders extend.
Because the elbows typically remain slightly flexed
and at a fixed elbow angle throughout the movement, the triceps brachii primarily contract isometrically throughout the movement. Although the hip
flexors would appear to also contract eccentrically
during the initial roll-out phase to control the rate of
hip extension and concentrically during the return
phase to aid in hip flexion, we unexpectedly found
low rectus femoris activity for both the Ab Slide and
Torso Track exercises. It appears that upper extremity
muscles may have a greater role compared to the hip
flexors in controlling and causing exercise movements during these 2 exercises. Although the activity
of the psoas muscle was not measured in the current
study due to being a deep muscle, it has been
demonstrated that during exercises performed in a
similar position and manner as the Ab Slide and
Torso Track that psoas EMG magnitudes are low and
that psoas EMG magnitudes are typically within approximately 10% of rectus femoris EMG magnitudes.16,17 From these data it can be hypothesized
that psoas activity, like rectus femoris activity, is
relatively low during the Ab Slide and Torso Track.
However, because the Torso Track and Ab Slide are
unique exercises in which psoas activity has not yet
been quantified, additional studies are needed to test
this hypothesis.
Because the Ab Slide and Torso Track exercises
produced the greatest activation of both abdominal
and upper extremity musculature, these exercises may
be beneficial for individuals with limited workout
time and whose goal is to perform exercises that not
only provide an abdominal workout but also an
upper body workout. The greater relative intensity
and number of muscles used during the Ab Slide and
Torso Track exercises implies that these exercises may
also achieve a greater energy expenditure compared
to the other exercises. Moreover, tension in the
latissimus dorsi in addition to the internal oblique
(and presumably the transverse abdominal), which all
tense the thoracolumbar fascia, may enhance trunk
stabilization while performing these exercises. It
should also be emphasized that cocontraction of the
lumbar paraspinal muscles, along with abdominal and
latissimus dorsi musculature, may enhance trunk
stability and spine stiffness. Although excessive activity
from the lumbar paraspinals can cause high compressive and shear (especially at the L5-S1) forces on the
lumbar spine, 3,16,23 the relatively low lumbar
paraspinal activity in all the exercises tested is probably not high enough to by itself cause deleterious
effects to the lumbar spine.
56

Performing exercises that generate high activity

from the hip flexors and lumbar paraspinals may not
be advantageous for those with weak abdominal
muscles or lumbar instability, because the forces
generated when these muscles act to anteriorly rotate
the pelvis may increase the lordotic curve of the
lumbar spine. Individuals with weak abdominal
muscles or lumbar instability may want to avoid the
bent-knee sit-up, SAM, Ab Twister, Ab Rocker, and Ab
Doer exercises due to the relatively high rectus
femoris activity. In exercises performed similarly to
the exercises in the current study, psoas and iliacus
activities have been shown to be similar in magnitude
and recruitment pattern as rectus femoris activity.1,2,17
The psoas muscle, by its attachments into the lumbar
spine, acts to hyperextend the spine as it flexes the
hip during the bent-knee sit-up and similar types of
hip flexion exercises, which may be detrimental to
individuals with lumbar instability. It has also been
demonstrated that the psoas muscle can generate
compression of the lumbar spine and anterior shear
force at L5-S1,16,23 which may be problematic for
individuals with lumbar disk pathologies. In addition,
the role of gravity in generating L5-S1 shear forces in
some exercises, such as the Torso Track and Ab Slide,
should also be considered when examining injury risk
to the low back. Unfortunately, it is unknown how
much L5-S1 shear force is generated while performing the Torso Track and Ab Slide, and whether or
not these forces are high enough to be problematic
in some patients with low back pathologies.

Effects of Electrode Placement on EMG Crosstalk

The electrode positions used in the current study
have been shown to minimize EMG crosstalk from
other muscles.5,8,20 This is especially true for the
internal oblique, which was the only muscle tested
that is not a superficial muscle. The internal oblique
normally lies deep to the external oblique, and
therefore is susceptible to considerable EMG crosstalk
from this muscle. However, it has been shown that
the internal oblique is only covered by the
aponeurosis of the external oblique, and not the
external oblique muscle, within the triangle confined
by the inguinal ligament, lateral border of the rectus
sheath, and a line connecting the ASISs.20 Therefore,
surface electrodes are appropriate to use for the
internal oblique when electrode placement is within
this area, especially when clinical questions are being
discussed and if a small percentage of EMG crosstalk
is acceptable. In fact, it has been shown that when
performing trunk flexion exercises similar to those in
the current study, mean internal and external oblique EMG data from surface electrodes (similarly
located as in the current study) were only approximately 10% different compared to mean internal and
external oblique EMG data from intramuscular electrodes.17 These authors demonstrated that appropriJ Orthop Sports Phys Ther Volume 36 Number 2 February 2006

ately placed surface electrodes accurately reflect

(within 10%) the muscle activity within the internal
or external oblique muscles.

CONCLUSIONS
The exercises in the current study activated abdominal muscles by actively flexing the trunk
(crunch, bent-knee sit-up, SAM, Ab Roller, Ab
Twister, Ab Rocker, Ab Doer) or by resisting trunk
extension (Ab Slide and Torso Track). The Ab Slide
and Torso Track exercises produced the highest
activation of the abdominal and upper extremity
muscles while minimizing low back and hip flexion
activity. Both the bent-knee sit-up and crunch exercises demonstrated similar amounts of abdominal
activation, while the Ab Twister, Ab Rocker, SAM, Ab
Doer, and bent-knee sit-up exercises exhibited the
greatest rectus femoris activity. The Ab Doer (good
morning and body boogie) exhibited the greatest
amount of lumbar paraspinal activity.

ACKNOWLEDGEMENTS

REFERENCES

J Orthop Sports Phys Ther Volume 36 Number 2 February 2006

57

REPORT

1. Andersson EA, Ma Z, Thorstensson A. Relative EMG

levels in training exercises for abdominal and hip flexor
muscles. Scand J Rehabil Med. 1998;30:175-183.
2. Andersson EA, Nilsson J, Ma Z, Thorstensson A. Abdominal and hip flexor muscle activation during various
training exercises. Eur J Appl Physiol Occup Physiol.
1997;75:115-123.
3. Axler CT, McGill SM. Low back loads over a variety of
abdominal exercises: searching for the safest abdominal
challenge. Med Sci Sports Exerc. 1997;29:804-811.
4. Balady GJ, Franklin BA, Whaley MH, Howley ET.
ACSMs Guidelines for Exercise Testing and Prescription. 6th ed. Philadelphia, PA: Lippincott Williams &
Wilkins; 2000.
5. Basmajian J, Blumenstein R. Electrode Placement in
EMG Biofeedback. Baltimore, MD: Williams & Wilkins;
1980.
6. Beim GM, Giraldo JL, Pincivero DM, Borror MJ, Fu FH.
Abdominal strengthening exercises: a comparative EMG
study. J Sport Rehab. 1997;6:11-20.
7. Clark KM, Holt LE, Sinyard J. Electromyographic comparison of the upper and lower rectus abdominis during
abdominal exercises. J Strength Cond Res. 2003;17:475483.
8. Cram JR, Kasman GS. Introduction to Surface
Electromyography. Gaithersburg, MD: Aspen Publishers,
Inc; 1998.
9. Cresswell AG, Grundstrom H, Thorstensson A. Observations on intra-abdominal pressure and patterns of abdominal intra-muscular activity in man. Acta Physiol
Scand. 1992;144:409-418.

RESEARCH

We would like to acknowledge Mike Andrawes,

Tracy Lowry, and Mark Adams for all their help in
data collection and analyses in this project.

10. Cresswell AG, Blake PL, Thorstensson A. The effect of

an abdominal muscle training program on intraabdominal pressure. Scand J Rehabil Med. 1994;26:7986.
11. Daggfeldt K, Thorstensson A. The role of intraabdominal pressure in spinal unloading. J Biomech.
1997;30:1149-1155.
12. Demont RG, Lephart SM, Giraldo JL, Giannantonio FP,
Yuktanandana P, Fu FH. Comparison of two abdominal
training devices with an abdominal crunch using
strength and EMG measurements. J Sports Med Phys
Fitness. 1999;39:253-258.
13. Gardner-Morse MG, Stokes IA. The effects of abdominal
muscle coactivation on lumbar spine stability. Spine.
1998;23:86-91; discussion 91-82.
14. Halpern AA, Bleck EE. Sit-up exercises: an
electromyographic study. Clin Orthop Relat Res.
1979;172-178.
15. Hodges PW, Richardson CA. Contraction of the abdominal muscles associated with movement of the
lower limb. Phys Ther. 1997;77:132-142; discussion
142-134.
16. Juker D, McGill S, Kropf P, Steffen T. Quantitative
intramuscular myoelectric activity of lumbar portions of
psoas and the abdominal wall during a wide variety of
tasks. Med Sci Sports Exerc. 1998;30:301-310.
17. McGill S, Juker D, Kropf P. Appropriately placed surface
EMG electrodes reflect deep muscle activity (psoas,
quadratus lumborum, abdominal wall) in the lumbar
spine. J Biomech. 1996;29:1503-1507.
18. Nachemson A. Lumbar intradiscal pressure. In: Jayson
MIV, ed. The lumbar Spine and Back Pain. Edinburg,
Scotland: Churchill Livingstone; 1987:191-203.
19. Nachemson AL. The lumbar spine: an orthopaedic
challenge. Spine. 1976;1:59-71.
20. Ng JK, Kippers V, Richardson CA. Muscle fibre orientation of abdominal muscles and suggested surface EMG
electrode positions. Electromyogr Clin Neurophysiol.
1998;38:51-58.
21. Ralston SH, Urquhart GD, Brzeski M, Sturrock RD.
Prevalence of vertebral compression fractures due to
osteoporosis in ankylosing spondylitis. BMJ.
1990;300:563-565.
22. Richardson CA, Snijders CJ, Hides JA, Damen L, Pas
MS, Storm J. The relation between the transversus
abdominis muscles, sacroiliac joint mechanics, and low
back pain. Spine. 2002;27:399-405.
23. Santaguida PL, McGill SM. The psoas major muscle: a
three-dimensional geometric study. J Biomech.
1995;28:339-345.
24. Sternlicht E, Rugg S. Electromyographic analysis of
abdominal muscle activity using portable abdominal
exercise devices and a traditional crunch. J Strength
Cond Res. 2003;17:463-468.
25. Thomson KD. Estimation of loads and stresses in
abdominal muscles during slow lifts. Proc Inst Mech
Eng [H]. 1997;211:271-274.
26. Warden SJ, Wajswelner H, Bennell KL. Comparison of
Abshaper and conventionally performed abdominal exercises using surface electromyography. Med Sci Sports
Exerc. 1999;31:1656-1664.
27. Willett GM, Hyde JE, Uhrlaub MB, Wendel CL, Karst
GM. Relative activity of abdominal muscles during
commonly prescribed strengthening exercises. J Strength
Cond Res. 2001;15:480-485.