A COMPARISON OF TECHNIQUES FOR ESTIMATING
TRAINING-INDUCED CHANGES IN MUSCLE
CROSS-SECTIONAL AREA
JASON M. DEFREITAS, TRAVIS W. BECK, MATT S. STOCK, MICHAEL A. DILLON, VANESSA D. SHERK,
JEFFREY R. STOUT, AND JOEL T. CRAMER
Department of Health and Exercise Science, University of Oklahoma, Norman, Oklahoma
ABSTRACT
DeFreitas, JM, Beck, TW, Stock, MS, Dillon, MA, Sherk, VD,
Stout, JR, and Cramer, JT. A comparison of techniques for
estimating training-induced changes in muscle cross-sectional
area. J Strength Cond Res 24(9): 2383–2389, 2010—The
ability to accurately estimate changes in muscle cross-sectional
area (CSA) could be a useful tool for strength and conditioning
practitioners to assess the effectiveness of a resistance training
program. The purpose of this study was twofold: (a) to compare
the reliability of 2 separate anthropometric-based field estima-
tions of thigh muscle CSA with that of a more accurate,
sophisticated imaging technique (peripheral quantitative com-
puted tomography [pQCT] scanner) and (b) to determine if the
field methods would be sensitive enough to detect changes in
CSA during a resistance training program. Twenty-five healthy,
untrained men completed 8 weeks of resistance training.
Cross-sectional area testing occurred twice before the start of
training, for reliability and again every 2 weeks during the study.
Testing consisted of a pQCT scan of the right thigh followed by
circumference and skinfold measurements. Two separate
equations (Moritani and deVries [M + D] and Housh multiple
regression [HMR]) were used to estimate CSA from the an-
thropometric data. The M + D and HMR methods demonstrated
intraclass correlations of 0.983 and 0.961, respectively, but
both significantly underestimated thigh muscle CSA when
compared to the pQCT. This error was consistent, however,
and consequently, the field methods were able to demonstrate
increases in muscle CSA with a pattern similar to those from the
pQCT. Thus, these equations can be useful tools to evaluate an
Address correspondence to Jason M. DeFreitas, defreitas@ou.edu.
24(9)/2383–2389
Journal of Strength and Conditioning Research
Ó 2010 National Strength and Conditioning Association
Copyright © National Strength and Conditioning Association Unauthorized reproduction of this article is prohibited.
athlete’s progress toward the goal of increasing muscle CSA. It
is the authors’ hope that the present study will increase
awareness among practitioners of these useful field methods
for estimating training-induced changes in muscle CSA.
KEY WORDS anthropometry, pQCT, hypertrophy
INTRODUCTION
I
t is well established that strength, which is one of the
determinants of performance, has a direct relationship
with the mass of skeletal muscle (6,7,9,12). The ability
to accurately estimate changes in muscle cross-
sectional area (CSA), therefore, could be a useful tool for
the strength and conditioning practitioner to assess the
effectiveness of their resistance training program. Currently,
magnetic resonance imaging (MRI) and computed tomog-
raphy (CT) are considered to be the gold standards for
in vivo estimation of muscle CSA (2–4,8). However, these
methods are expensive and typically inaccessible for the
coaching community. Anthropometric-based methods may
provide the cheapest and most easily accessible alternative
for estimating muscle CSA.
The uses of anthropometry for estimating body compo-
sition and determination of body build characteristics
are fairly well known. However, its use for muscle size
estimation is less familiar to the majority of the practitioners.
In an early attempt to quantify muscular performance, Martin
(10) proposed using height, weight, and chest circum-
ference measurements as an index to predict physical
efficiency. Rasch and Morehouse (14) took this concept 1
step further and used upper-arm circumference to track
muscle hypertrophy during a 6-week training program.
The limitation of girth measurements, however, is that they
are unable to distinguish between fat and muscle. To our
knowledge, Moritani and deVries (11) were the first to
attempt to account for subcutaneous fat with an anthropo-
metric-based estimation of muscle size. They accomplished
this by measuring the circumference of the upper arm,
calculating its radius, and then adjusting the radius
by subtracting the average of 4 skinfolds. Theoretically, the
VOLUME 24 | NUMBER 9 | SEPTEMBER 2010 | 2383
 Field Methods for Estimating Muscle CSA

TABLE 1. A summary of the study design.

Week Day 1 Day 3 Day 5 Day 7

0 (Pretesting) Testing (PRE 1)* Testing (PRE 2)*

1 Training Training Training
2 Training Training Testing*
Training
3 Training Training Training
4 Training Training Testing*
Training
5 Training Training Training
6 Training Training Testing*
Training
7 Training Training Training
8 Training Training Training Testing (posttesting)*
*Testing session.

CSA calculated from this fat-adjusted radius includes only
the remaining lean tissue (muscle and bone). Housh et al. (8)
took a different approach and used MRI to develop an
anthropometric-based multiple regression equation to pre-
dict thigh muscle CSA.
Despite these advancements in anthropometric-based
methodology, CSA is still an inaccessible measurement for
most practitioners. The purpose of this study was twofold: (a)
to compare the reliability of 2 separate anthropometric-based
field estimations of thigh muscle CSA with that of
a more accurate, sophisticated imaging technique (peripheral
quantitative computed tomography [pQCT] scanner) and (b)
to determine if the field methods would be sensitive enough
to detect changes in CSA during a resistance training
program.
METHODS
Experimental Approach to the Problem
To investigate the test–retest reliability of the anthropometric
and pQCT measurements, each subject performed 2 sessions of
pretesting before the resistance training, separated by at least 48
hours. To observe the patterns of response for the different CSA
measurement techniques, the subjects performed an 8-week
resistance training program designed to stimulate hypertrophy
in the leg extensor muscles. The training was 3 d
wk21, and the
sessions were always 48 hours apart. After training began, the

TABLE 2. An example of a subject’s first workout.*†

Pretesting Set 1 Set 2 Set 3

Exercise 1RM 80% Weight (lb) Reps Weight (lb) Reps Weight (lb) Reps

Leg press 360 288 290 6 270 10 270 6

Bench press 120 96 95 12 95 6 95 4
Leg extension n/a n/a 110 15 120 14 130 11
*1RM = 1-repetition maximum.
†The leg press and bench press starting weights were approximately 80% of the subject’s 1RM. The leg extension 1RM was not
tested. Every set was performed to failure. The weight was adjusted on a set-by-set basis with the goal of having the subject fail between
8 and 12 repetitions. For example, the subject did not meet the minimum of 8 reps on the first set of leg press, so the weight was
decreased for the second set. For the leg extension, the subject performed .12 repetitions, so the weight was raised for the subsequent
set. If a subject showed the ability to perform the appropriate number of repetitions in 1 set, but not in a following one, then the weight was
not adjusted (e.g., the bench press). This is because the failure to perform the desired repetitions was because of a lack of local muscular
endurance and not strength.

the TM

2384 Journal of Strength and Conditioning Research

Copyright © National Strength and Conditioning Association Unauthorized reproduction of this article is prohibited.
 the TM

Journal of Strength and Conditioning Research | www.nsca-jscr.org

Figure 1. Analyzing each subject’s whole muscle cross-sectional area (CSA, cm2) was a 3-part process. A) A sample scan of a subject’s dominant thigh. B) A filter
was used to distinguish between the density of fatty tissue and the density of lean tissue (which includes both muscle and bone). C) A separate filter was then
used to distinguish between the density of bone tissue and the densities of fat and lean tissue. The CSA of the bone tissue was then subtracted from the CSA of
the lean tissue (B–C) to provide the CSA of the thigh muscles.

subjects were tested again every 2 weeks throughout the study.
During the third visit of each testing week, the subjects were
tested for CSA, and they then performed their training. With
a few minor exceptions, all of the testing was performed at the
same time of the day for each subject. A summary of the study
design can be seen in Table 1.
Subjects
Twenty-five healthy men (mean 6 SD age = 21.5 6 3.6 years;
stature = 1.81 6 0.01 m; and mass = 76.5 6 13.2 kg)
volunteered to participate in this investigation. Each
participant completed an informed consent and a pre-
exercise health and exercise status questionnaire. The
questionnaire had to indicate no current or recent (within
the past 6 months) neuromuscular or musculoskeletal
problems to the knees, hips, or lower back for the subject
to be considered eligible for the study. In addition, each
subject had to be untrained in resistance exercise (i.e., no
participation in an organized weight training program for at
least the last 6 months before the study). The study was
approved by the University
Institutional Review Board for
Human Subjects before testing
and took place during both the
Fall and Spring semesters on
the University of Oklahoma
campus.

Procedures
Resistance Training and Testing.
The training program consisted
of the bilateral incline leg-press,
leg extension and bench-press
exercises performed 3 d
wk21
for 8 weeks. For each exercise,
3 sets to failure were per-
formed, with approximately
2 minutes of rest between each
set. The training load continu-
ally increased as the subjects
became stronger to follow the
Figure 2. Mean 6 SD muscle cross-sectional area (CSA; cm2) across the training program for the peripheral
quantitative computed tomography (pQCT), Moritani and deVries (M + D) (11), and Housh multiple regression
principles of progressive over-
(HMR) (8) methods. load. Therefore, the weight was
constantly adjusted to assure

VOLUME 24 | NUMBER 9 | SEPTEMBER 2010 | 2385

Copyright © National Strength and Conditioning Association Unauthorized reproduction of this article is prohibited.
 Field Methods for Estimating Muscle CSA

maximums (1RMs) for each

participant. During the second
pretesting session (PRE2), the
subjects performed the 1RM
test. Before the 1RM test,
each subject performed 3
warm-up sets with loads of
60% (6–8 repetitions), 70%
(3–5 repetitions), and 80%
(1–2 repetitions) of their esti-
mated 1RM. The weight
was then increased after each
successful lift until the partici-
pant failed. The 1RM was
usually established within 5
attempts to the nearest 5-lb
increment. Approximately 80%
of each subject’s 1RM (to the
Figure 3. Relationship between the pQCT and Moritani and deVries (M + D) (11) methods for estimating cross- nearest 5-lb. increment) was
2
sectional area (CSA; cm ).
used as their starting weight for
the first training session. The
same 1RM testing procedures
were used for the posttesting,
that the subject was failing between 8 and 12 repetitions (i.e., except that the warm-ups were based on pretesting values,
if the subject performed 16 repetitions, the weight was rather than on estimations.
increased accordingly before the next set). This set-by-set
adjustment system is shown in Table 2. During the first Muscle Cross-Sectional Area. The CSA of the right thigh
pretesting session (PRE1), the subjects were familiarized muscles was measured using a pQCT (XCT 3000, Stratec
with the leg press and bench press exercises. This allowed Medizintechnik GmbH, Pforzheim, Germany) scanner.
the inexperienced participants to become accustomed to Cramer et al. showed that when compared to the
proper lifting technique and also allowed the investigator gold standard MRI, the pQCT was a valid and reliable
to find an approximate estimate of the 1-repetition measurement of muscle CSA (3). The authors tested
the pQCT’s validity using 2
separate investigators and cal-
culated a correlation between
MRI and pQCT. Their R2
values were 0.979 and 0.983.
For the present study, the sub-
ject sat upright with their leg
fully extended (180°), and the
scan was taken at the midpoint
of the thigh (i.e., 50% of
the distance between the greater
trochanter and lateral epicon-
dyle of the femur). The voxel
resolution for each scan was 0.4
mm. Muscle CSA was calculated
using the software provided by
the manufacturer. This proce-
dure can be seen in Figure 1.
In addition to using the
pQCT to directly measure
Figure 4. Relationship between the pQCT and Housh multiple regression (HMR) (8) methods for estimating cross- changes in the muscle CSA, 2
sectional area (CSA; cm2). separate techniques of estimat-
ing CSA from anthropometric
the TM

2386 Journal of Strength and Conditioning Research

Copyright © National Strength and Conditioning Association Unauthorized reproduction of this article is prohibited.
 the TM

Journal of Strength and Conditioning Research | www.nsca-jscr.org

Statistical Analyses
A 2-way repeated measures
(method 3 time) analysis of
variance (ANOVA) was used
to compare the 3 different
muscle CSA measurements
across the training program.
When appropriate, follow-up
analyses included 1-way re-
peated measures ANOVAs
and Bonferroni post hoc com-
parisons. An alpha level of 0.05
was used for all compar-
isons. Test–retest reliability
between the 2 pretesting ses-
sions was calculated using
Pearson correlations, 3 sepa-
rate paired samples t-tests and
Figure 5. Relationship between the Moritani and deVries (M + D) (11) and Housh multiple regression (HMR) (8) 2-way, fixed-effect intraclass
methods for estimating cross-sectional area (CSA; cm2).
correlations (ICCs; model
3,1), and standard error of
measurements were used. The scan location of the pQCT was measurement (SEM), and the minimal difference needed
marked on the thigh to ensure that the anthropometric for a change to be considered real (15).
measurements were taken at the same position. The
measurements included thigh circumference and 4 skinfolds RESULTS
(anterior, posterior, medial, and lateral). The first technique Muscle Cross-Sectional Area
estimated CSA with the following equation as provided by The mean 6 SD thigh muscle CSAs for the 3 measurement
Moritani and deVries (M + D) (11): methods can be seen in Figure 2. The 2-way repeated
2 4 32 measures ANOVA revealed a significant method 3 time
+ skf i interaction. Three separate 1-way repeated-measures
6Circumference i ¼1 7
CSA ¼ p4  5 ; ANOVAs established that for each of the 3 methods, there
2p 4
was a significant increase in CSA over time. Six additional
1-way repeated-measures ANOVAs showed that there were
where Circumference was the circumference of the thigh, and
significant differences among the CSAs from the 3 methods
skf were the thigh skinfold thicknesses at each of the 4 sites.
at each time point. Bonferroni pairwise comparisons for all
The second technique used the following multiple regression
1-way ANOVAs are shown in Figure 2. Correlations among
equation as provided by Housh et al (Housh multiple
the 3 methods are demonstrated in Figures 3–5. The test–
regression [HMR]) (8):
retest reliability results for each method are shown in Table 3.
CSA ¼ ð4:683CircumferenceÞ  ð0:643skfA Þ  22:69; The pretesting (PRE 2) to posttesting (Week 8) percent
changes in CSA for the pQCT, M + D, and HMR methods
where skfA was the anterior thigh skinfold. were 9.1, 16.6, and 9.9%, respectively.

TABLE 3. Test–retest reliability for the 3 methods of estimating muscle cross-sectional area.*

Method Pre 1 Mean 6 SD (cm2) Pre 2 Mean 6 SD (cm2) r ICC3,1 SEM (cm2) MD (cm2) pValue

pQCT 145 6 26.1 145.6 6 26.2 0.995 0.995 1.76 4.89 0.20
M+D 121 6 25 120.6 6 25 0.983 0.983 3.25 9.01 0.64
HMR 114.8 6 20.6 115.5 6 20.6 0.961 0.961 4.05 11.22 0.52

*pQCT = peripheral quantitative computed tomography method; M + D = Moritani and deVries method (11); HMR = Housh multiple
regression method (8); ICC3,1 = intraclass correlation coefficient; SEM = standard error of measurement; MD = minimal difference
needed for a change to be considered real.

VOLUME 24 | NUMBER 9 | SEPTEMBER 2010 | 2387

Copyright © National Strength and Conditioning Association Unauthorized reproduction of this article is prohibited.
 Field Methods for Estimating Muscle CSA

DISCUSSION different in the 2 studies. Moritani and deVries’s (11) subjects

Previous investigations have shown that the pQCT can performed unilateral training of the forearm flexors using 2
accurately determine muscle CSA (3,4). However, the sets of 10 repetitions at approximately 67% of their
maximum. The subjects of the present study followed
technique is expensive and requires extensive training by the
a program in which the variables matched recommendations
investigator. Therefore, it is also impractical and prohibitive for
by the National Strength and Conditioning Association
strength and conditioning practitioners and some researchers.
(NSCA) with the goal of stimulating hypertrophy (1).
The results from this study indicated that anthropometric field
Therefore, they performed bilateral training of the leg
methods can not only be used to estimate thigh muscle CSA
extensors using 3 sets of 8–12 repetitions at approximately
before a training program, as previously investigated (3) but
80% of their maximum while following the principles of
can also estimate the changes in CSA that occur during
progressive overload. Additionally, because they also per-
a resistance training program.
formed the bench press exercise, they recruited more muscle
We hypothesized that the pQCTwould be more sensitive to
mass than the Moritani and deVries (11) study and
changes in muscle CSA than the other 2 methods that are
incorporated both the upper- and lower-body. This may
based on anthropometric measurements. For example, small
be noteworthy because Hansen et al. (5) have shown there is
changes in muscle CSA at the beginning of the training
a significantly greater growth hormone response when
program might be detected with the pQCT but not with
a combination of upper- and lower-body exercises is
the other 2 methods. To our surprise, not only did the
performed, and Mulligan et al. (13) reported that exercise
anthropometric measurement-based methods show high
routines with greater training volumes elicit greater hor-
degrees of precision with ICCs of 0.983 and 0.961, compared
monal and metabolic responses than those with lower
to 0.995 for the pQCT, but they also showed similar sensitivity
volumes. Third, Moritani and deVries’s (11) subjects only
to change. All 3 methods showed a significant increase in trained the agonist muscle. Because the leg press exercise
muscle CSA after only 2 weeks of training. It should be noted used in the present study involves thigh extension, the
that although the anthropometric-based methods provided antagonist (hamstrings) muscles were also trained. Although
a similar pattern of response for CSA over time as the pQCT, the thigh extensors did not undergo the same volume of
they consistently underestimated the absolute value of muscle training as the leg extensors, it is possible that they did
CSA. The overall percentage change in CSA because of influence the muscle CSA changes. However, they would
training from the HMR method (9.9%) was very similar to that have influenced the CSA measurements of the pQCT and
of the pQCT (9.1%), but less than that from the M + D method anthropometric-based methods equally.
(16.6%). The absolute CSA values for the HMR method were
also more highly correlated with those of the pQCT (r = 0.95)
than those from the M + D method (r = 0.91). However, as PRACTICAL APPLICATIONS
shown in Figure 2, the values estimated by the M + D
For a coach aiming to increase the muscle mass of his or her
method did not underestimate the CSA as much as the HMR
athletes, the anthropometric equations that were used in the
method. At each time point, the M + D CSA values were
present study can be useful tools to evaluate their progress
significantly closer to those from the pQCT than were those
toward that goal. These equations are simple, inexpensive,
from the HMR. Therefore, each anthropometric-based
and reliable methods for estimating training-induced changes
method has advantages and disadvantages, but both provide
in muscle CSA. It is the authors’ hope that the present study
reliable and reasonably valid estimates of CSA.
not only increases the awareness among practitioners of these
Moritani and deVries (11) did not report any hypertrophic
potentially useful field methods for assessing muscle CSA but
changes (based on the Efficiency of Electrical Activity [EEA]
also justifies their use to assess progress and track training–
technique) until roughly 4–6 weeks into their training
induced changes in muscle CSA.
program. The authors (11) did not, however, have access
to the peripheral imaging techniques that are available today.
This led to the expectation of less sensitivity to change for the
ACKNOWLEDGMENTS
M + D method when compared to the pQCT technique in
this study. The disparity in the time course of hypertrophic The funding for this study was received by the NSCA
changes between Moritani and deVries (11) and the present Foundation through their Master’s Student Research Grant.
investigation could be because of a number of reasons. First,
Moritani and deVries (11) investigated the biceps brachii,
whereas the present study examined the quadriceps femoris. REFERENCES
It is possible that the difference in the architecture between 1. Baechle, TR, Earle, RW, and Wathen, D. Resistance training. In:
Essentials of Strength Training and Conditioning. Champaign, IL:
the 2 muscle groups (fusiform vs. pennate) may have Human Kinetics. 2000, pp. 395–426.
influenced the disparity in the rate that the CSAs increased. 2. Bemben, MG, Sato, Y, and Abe, T. The use of anthropometry for
Second, the volume of training used was significantly assessing muscle size. Int J Kaatsu Train Res 1: 33–36, 2005.
the TM

2388 Journal of Strength and Conditioning Research

Copyright © National Strength and Conditioning Association Unauthorized reproduction of this article is prohibited.

3. Cramer, JT, Palmer, IJ, Ryan, ED, Herda, TJ, Bemben, DA, Bemben, MG,
and Stratemeier, PH. Validity and reliability of a peripheral quanti-
tative computed tomography scanner for measuring muscle cross-
sectional area. In: American College of Sports Medicine–Annual Meeting.
New Orleans, LA: 2007. Med Sci Sports Exer 39: S225–S226, 2007.
4. Gordon, CL, Webber, CE, and Beaumont, LF. Accuracy and
precision error of muscle cross-sectional area measured using
peripheral quantitative computed tomography in adults. In:
American Society for Bone and Mineral Research–Annual Meeting.
Minneapolis, MN: Wiley Blackwell, 2003.
5. Hansen, S, Kvorning, T, Kjaer, M, and Sjogaard, G. The effect of
short-term strength training on human skeletal muscle: the
importance of physiologically elevated hormone levels. Scand J Med
Sci Sports 11: 347–354, 2001.
6. Haxton, HA. Absolute muscle force in the ankle flexors of man.
J Physiol 103: 267–273, 1944.
7. Hettinger, T. Physiology of Strength. Springfield, IL: Thomas, 1961.
8. Housh, DJ, Housh, TJ, Weir, JP, Weir, LL, Johnson, GO, and Stout, JR.
Anthropometric estimation of thigh muscle cross-sectional area.
Med Sci Sports Exerc 27: 784–791, 1995.
Copyright © National Strength and Conditioning Association Unauthorized reproduction of this article is prohibited.
the TM
Journal of Strength and Conditioning Research | www.nsca-jscr.org
9. Ikai, M and Fukunaga, T. Calculation of muscle strength
per unit cross-sectional area of human muscle by means
of ultrasonic measurement. Int Z Angew Physiol 26: 26–32,
1968.
10. Martin, EG. Tests of muscular efficiency. Physiol Rev 1: 454, 1921.
11. Moritani, T and deVries, HA. Neural factors versus hypertrophy in
the time course of muscle strength gain. Am J Phys Med 58:
115–130, 1979.
12. Morris, CB. The measurement of the strength of muscle relative to
the cross section. Res Q 19: 295–303, 1948.
13. Mulligan, SE, Fleck, SJ, Gordon, SE, Koziris, P, Triplett-McBride, T,
and Kraemer, WJ. Influence of resistance exercise volume on serum
growth hormone and cortisol concentrations in women. J Strength
Cond Res 10: 256–262, 1996.
14. Rasch, PJ and Morehouse, LE. Effect of static and dynamic
exercises on muscular strength and hypertrophy. J Appl Physiol
11: 29–34, 1957.
15. Weir, JP. Quantifying test–retest reliability using the intraclass
correlation coefficient and the SEM. J Strength Cond Res 19: 231–240,
2005.
VOLUME 24 | NUMBER 9 | SEPTEMBER 2010 | 2389