Professional Documents
Culture Documents
From the *Gait Analysis Laboratory, Department of Orthopedic Communications should be addressed to:
Surgery, and †Department of Physical Therapy, Hvidovre University T. Bandholm; Gait Analysis Laboratory (Section 247), Department of
Hospital, Copenhagen, ‡Department of Radiology X 3023, Orthopedic Surgery; Hvidovre University Hospital; Kettegaard Allé
Rigshospitalet, Copenhagen, §Department of Pediatrics, Hvidovre 30; DK-2650 Hvidovre, Denmark.
University Hospital, Copenhagen, and ¶Department of Exercise and E-mail: thomas.bandholm@hvh.regionh.dk
Sport Sciences, University of Copenhagen, Copenhagen, Denmark. Received March 12, 2007; accepted May 30, 2007.
© 2007 by Elsevier Inc. All rights reserved. Bandholm et al: Muscle Volume Estimates 263
doi:10.1016/j.pediatrneurol.2007.05.019 ● 0887-8994/07/$—see front matter
cles to be quantified in a quick, low-cost manner. Further-
more, as objective measurements of spasticity using iso-
kinetic dynamometry are emerging [6,7,9-11], the degree
of spasticity reduction could be assessed objectively in
muscle volume-dose studies.
Previous attempts were made to estimate arm and leg
muscle group volume and muscle mass based on anthro-
pometry [12,13]. Others combined anthropometric and
ultrasound measurements to establish prediction equations
for limb muscle volume [14]. In this way, muscle-volume
prediction equations of the elbow flexor and extensor
[1,15], quadriceps femoris [16], and tibialis anterior [17]
muscles were developed. For the plantar flexors specifi-
cally, Miyatani et al. established a muscle-volume predic-
tion equation for the plantar flexor muscle group, using
muscle group thickness and lower leg length as indepen-
dent variables [14]. To our knowledge, the accuracy of
estimating the volume of individual plantar flexor muscles
using ultrasound has not been addressed. Furthermore,
muscle-volume prediction equations have not usually been
validated against the true muscle volume measured as the
amount of displaced water when the muscle is lowered
into a water-filled cylinder, because this technique can be Figure 1. Measurement sites used for ultrasound measurements (top)
and an example of an ultrasound scan (bottom). mGA, measurement site
performed in cadavers only. for the medial gastrocnemius muscle; IGA, measurement site for the
The primary objective of the present study was to lateral gastrocnemius muscle; SOL, measurement site for the soleus
muscle; GA, gastrocnemius; SOL, soleus.
investigate the suitability of using ultrasonography mus-
cle-thickness measurements and anthropometry to esti-
mate gastrocnemius and soleus muscle volumes in cadav- Lines connecting the MTJs of the medial and lateral gastrocnemius head
eric human legs. We tested the hypothesis that the volume (mGA and IGA, respectively), and the insertion of the Achilles tendon
of gastrocnemius and soleus muscles can be estimated with the corresponding 25%, 75%, and 50% marks indicated on the
posterior circumferential knee epicondylar line, were then drawn on the
using muscle thickness and anthropometric measurements.
skin, and distances were measured (Fig 1). Half of the measured distance
was indicated on the skin for each line, leaving one measurement site for
each of the gastrocnemius muscle heads and one measurement site for the
Methods
soleus muscle (SOL) at approximately 50% of the muscle-belly length.
The subcutaneous adipose tissue-muscle to muscle-fascia (mGA and
Materials IGA) and muscle-fascia to muscle-fascia (SOL) interface distances
(muscle thickness) were then determined from the ultrasound image (Fig 1).
Eleven whole, frozen, cadaveric human legs were used after thawing Muscle thickness was determined for the mGA, IGA, and SOL as a mean
overnight. No information could be obtained concerning age, sex, body of three scans for each measurement site. The same investigator per-
weight, and cause of death of the deceased because of medical-school formed all ultrasound measurements, and the mean difference of the three
routines. recordings at each measurement site was ⬍0.4 mm. Gastrocnemius
muscle thickness was expressed as a mean of the mGA and IGA muscle
thickness. During scans, great care was taken not to compress the dermal
Muscle Thickness Measured by Ultrasound surface.
Simple regression
Gastrocnemius y ⫽ 0.987x ⫹ 147.1 0.373 (0.046) 24.4 (11.7)
Soleus y ⫽ 0.238x ⫹ 154.9 0.518 (0.012) 33.2 (12.7)
Multiple regression
Gastrocnemius Y ⫽ 85.3X1 ⫹ 5.0X2 ⫺ 104.9 0.497 (0.016) 21.9 (10.5)
Soleus Y ⫽ 73.6X1 ⫹ 10.6X2 ⫺ 284.4 0.650 (0.003) 28.4 (10.8)
Abbreviations:
LL ⫽ Lower leg length
MT ⫽ Muscle thickness
r2 ⫽ Validity correlation between estimated and measured muscle volume
SEE ⫽ Standard error of the estimate
x ⫽ ⫻ (MT/2)2 ⫻ LL
X1 ⫽ Muscle thickness (cm)
X2 ⫽ Lower leg length (cm)
y ⫽ Estimated muscle volume (cm3), using simple regression equation
Y ⫽ Estimated muscle volume (cm3), using multiple regression equation
respectively, of the variation in gastrocnemius muscle standard errors of estimates between 10-13% of the
volume, and 28% and 38%, respectively, of the variation measured mean muscle volume, using ultrasonography
in soleus muscle volume (P ⬍ 0.05). The muscle-volume and anthropometry. The established muscle-volume pre-
prediction equations, based on multiple regression analy- diction equations were based on simple and multiple
ses, are given in Table 1. Using these equations to estimate regression analyses, using muscle thickness and lower leg
muscle volume resulted in an r2 and SEE of 0.497 (P ⫽ length as independent variables.
0.016) and 21.9 cm3 (10.5%) for the gastrocnemius The accuracy of the established muscle-volume predic-
muscle, and an r2 and SEE of 0.650 (P ⫽ 0.003) and 28.4 tion equations in predicting the measured muscle volume
cm3 (10.8%) for the soleus muscle when validated against differed somewhat between equations based on simple and
the measured muscle volume (Fig 2C,D). No systematic multiple regression analyses. Equations based on multiple
differences existed between the estimated and measured regression analyses produced higher r2 and lower SEEs
muscle volume (P ⬎ 0.05), and the Bland and Altman than equations based on simple regression analyses when
plots showed no significant correlation between muscle- validated. This implies that the assumption of proportion-
volume differences and the means for any of the two ality with squared muscle thickness (simple regression
muscles (r ⬍ 0.246, P ⬎ 0.465, Fig 3C,D). analyses) is not entirely valid for the gastrocnemius and
soleus muscles. By performing multiple regression analy-
Discussion ses, the accuracy of muscle-volume prediction equations
in predicting the measured muscle volume improved when
The main finding of the present study was that gastroc- compared with simple regression analyses. This outcome
nemius and soleus muscle volume could be estimated with was also found by Miyatani et al., who established simple
and multiple regression equations based on ultrasound davers to living humans should always be undertaken with
measurements to estimate the muscle volume of the whole caution. Freezing of human tissue seems to increase the
plantar flexor muscle group determined from multiple extracellular area due to cell shrinkage [19]. This increase
magnetic resonance imaging scans [14]. Using only mus- is likely caused by a loss of water from the cells to the
cle thickness in the simple regression equation resulted in surrounding medium (extracellular media). Hence, abnor-
a SEE of 9.2%, whereas using muscle thickness and lower mal fluid shifts within the investigated muscles might have
leg length in the multiple regression equation resulted in a taken place because of the freeze-thaw cycle. It is un-
SEE of 6.6% in a validation sample of 14 healthy men known, however, if these shifts had any impact on muscle-
[14]. A likely reason for this between-study difference in thickness measurements by changing the total muscle
results for the SEE of muscle-volume estimates might be water content. If so, this was likely a systematic error.
that the shape of the individual plantar flexor muscle Bearing this issue in mind, using the muscle-volume
varies to a greater extent between individuals than that of prediction equations developed in the present study to
the entire plantar flexor muscle group. Furthermore, a few predict gastrocnemius and soleus muscle volume in living
precautions must be taken when interpreting the present humans may influence the SEE. The calculated SEE is
results. valid only for the sample in this validation study, as the
First, the selection of investigated legs was based on variation in the prediction equations likely varies from
availability. No information about cause of death, age, sex, sample (cadaver) to sample (living human).
and physical-activity status of the deceased could be Despite these apparent limitations, SEEs of 10.5% and
obtained. Secondly, extrapolating results obtained in ca- 10.8% for the respective gastrocnemius and soleus muscle