Content uploaded by François Castonguay
Author content
All content in this area was uploaded by François Castonguay on Jan 19, 2016
Content may be subject to copyright.
M. Thériault, C. Pomar and F. W. Castonguay
at different measuring sites in lamb
Accuracy of real-time ultrasound measurements of total tissue, fat, and muscle depths
doi: 10.2527/jas.2008-1002 originally published online Jan 16, 2009;
2009.87:1801-1813. J Anim Sci
http://jas.fass.org/cgi/content/full/87/5/1801
the World Wide Web at:
The online version of this article, along with updated information and services, is located on
www.asas.org
at Bibliotheque De L'Univ. Laval Section Des Acquisitions on April 27, 2009. jas.fass.orgDownloaded from
ABSTRACT: Accuracy of live ultrasound measure-
ments to evaluate the total tissue depth (GR), as well
as fat and LM depths at different scanning sites, was
studied in 96 purebred Suffolk and Dorset lambs of both
sexes slaughtered between 36 and 54 kg of BW. Before
slaughter, 7 real-time ultrasound measurements were
taken on lambs: fat and LM depths between the 12th
and 13th ribs (transverse) and between the 3rd and 4th
lumbar vertebrae (transverse and longitudinal), and
GR. After slaughter, the measurements equivalent to
ultrasound measurements were taken on digitized im-
ages of the cuts on the left half carcass of each lamb.
Ultrasound GR and fat depth measurements were close-
ly correlated with the corresponding carcass measure-
ments (0.76 ≤ r ≤ 0.81). Ultrasound GR measurement
exhibited a large error of central tendency, but the level
of error due to the disturbance (ED) was comparable
with fat depth measurements (ED = 8.5%; residual SD
= 2.24 mm; CV of residuals = 9.5%). Ultrasound fat
depth measurements were more accurate between the
12th and 13th ribs (error due to regression = 1.20; ED
= 0.82) than between the 3rd and 4th lumbar vertebrae
(error due to regression = 5.58 and 5.4; ED = 1.10 and
0.93, transverse and longitudinal, respectively), mainly
due to image interpretation errors in the lumbar region.
Measurements of LM depth demonstrated low variabil-
ity in the population under study (SD = 2.6 mm), and
these ultrasound measurements showed low correlation
with the corresponding carcass measurements (0.34 ≤
r ≤ 0.43). The results of this study demonstrated that
ultrasound measurements were more accurate for eval-
uating fat depth and the GR measurements than for es-
timating LM depths. Ultrasound GR measurement is a
promising measurement, especially where carcass grad-
ing systems are based on this carcass measurement.
Key words: fat depth, lamb, live measurement, loin muscle depth, total tissue depth measurement, ultrasound
©2009 American Society of Animal Science. All rights reserved. J. Anim. Sci. 2009. 87:1801–1813
doi:10.2527/jas.2008-1002
INTRODUCTION
In hogs and cattle, ultrasound fat and muscle depths
have been used for several years in genetic selection
programs for improving carcass quality, and much re-
search has been published on this topic (Moeller, 2002;
Williams, 2002). In sheep, few scientific data are avail-
able to evaluate and compare the accuracy of the dif-
ferent scanning sites, and conclusions on their useful-
ness were often conflicting (Houghton and Turlington,
1992). Besides, it is difficult to compare results because
of the various statistical methods used.
The most common site to evaluate transverse fat and
LM depths in lamb is at the 12th-13th ribs (Wilson,
1992). Some researchers obtained greater correlation
between ultrasound and carcass measurements at the
3rd-4th lumbar vertebrae (Fernández et al., 1998; Silva
et al., 2006). In hogs, longitudinal measurements, par-
allel to the backbone, are commonly used in Canada,
but this method is rarely reported in lambs (Berg et al.,
Accuracy of real-time ultrasound measurements of total tissue, fat,
and muscle depths at different measuring sites in lamb1
M. Thériault,*† C. Pomar,* and F. W. Castonguay*†2
*Dairy and Swine Research and Development Centre, Agriculture and Agri-Food Canada, Sherbrooke, Québec,
Canada, J1M 1Z3; and †Département des sciences animales, Université Laval, Québec,
Québec, Canada, G1V 0A6
1
Financial support for this project was provided by the Conseil
des recherches en pêche et en agroalimentaire of the Ministère de
l’Agriculture, des Pêcheries et de l’Alimentation du Québec (Qué-
bec, Canada), Agriculture and Agri-Food Canada (AAC), the Fonds
pour la formation des chercheurs et l’aide à la recherche du Qué-
bec (Québec, Canada), the Coopérative fédérée (Montreal, Québec,
Canada), Intervet Canada Ltd. (Whitby, Ontario, Canada), and the
Société des éleveurs de moutons de race pure du Québec (Pont-
Rouge, Québec, Canada). The authors thank Jean-Paul Daigle at
the Centre de développement du porc du Québec (Québec, Canada)
for his guidance and expertise in ultrasound measurement, Francis
Goulet (Université Laval) for professional assistance, Giovanny Leb-
el for taking care of the lambs (Lamb Test Station, St-Jean-de-Dieu,
Québec, Canada), Carol Bernier and all the staff at the Luceville
abattoir (Luceville, Québec, Canada), Marcel Marcoux (AAC) and
Julie Mercier (Université Laval) for coordinating and carrying out
the carcass fabrication and image digitization, and Steve Méthot
(AAC) for his assistance in the statistical analyses.
2
Corresponding author: francois.castonguay@fsaa.ulaval.ca
Received March 4, 2008.
Accepted January 9, 2009.
1801
at Bibliotheque De L'Univ. Laval Section Des Acquisitions on April 27, 2009. jas.fass.orgDownloaded from
1996) and has never been compared with the transverse
measurement. Total tissue depth over the 12th rib at 11
cm from the midline of the carcass (GR) is a measure-
ment included in carcass grading systems to predict
lean and fat yield in Canada (Jones et al., 1996), Aus-
tralia (Hopkins, 1994), and New Zealand (Kirton and
Johnson, 1979). Despite the proven usefulness of car-
cass GR measurement, few researchers demonstrated
interest in this ultrasound scanning site in live lambs
(McEwan et al., 1989; Ramsey et al., 1991; Hopkins et
al., 1993).
This investigation aimed to determine the better
scanning sites, in regard to accuracy and feasibility,
to validate and refine the evaluation of carcass quality
traits in genetic selection programs for lamb. Specifi-
cally, the objectives of this study were to assess the ac-
curacy of ultrasound fat and LM depths at 2 scanning
sites and the accuracy of ultrasound GR, as well as to
compare transverse and longitudinal ultrasound mea-
surements.
MATERIALS AND METHODS
Care and handling of the lambs used in this study
were conformed to the guidelines established by the
Canadian Council on Animal Care (1993).
Animal Sampling and Husbandry Conditions
A total of 144 purebred Suffolk (SU; n = 72) and
Dorset (DP; n = 72) lambs were selected at weaning,
at around 55 d of age, from 9 Quebec sheep producers.
Lambs representative to their respective breed in terms
of weaning weight were retained. The Dorset breed is a
maternal type and is characterized by moderate growth
and a greater fat content. Conversely, the Suffolk is a
leaner, fast-growing terminal breed.
Lambs entered the test station at around 65 d of age
and were assigned according to sex (male and female),
breed (SU and DP), and slaughter weight classes (36 to
39 kg, 41 to 44 kg, 46 to 49 kg, and 51 to 54 kg) to a 2
× 2 × 4 factorial design arranged in 8 randomized com-
plete blocks, each block consisting of 1 pen of each sex.
Treatments were used for studying growth and tissue
deposition of heavy lambs as part of another research
project (F. W. Castonguay, unpublished data). Lambs
were fed ad libitum a pelleted, complete grower diet
(18% CP; 2.76 Mcal/kg of ME) to approximately 35 kg
of BW and then a finisher diet (15% CP; 2.79 Mcal/
kg of ME) until slaughter. High quality hay also was
available for ad libitum intake during the entire experi-
ment.
Live Measurements
Body weight and the ultrasound measurements were
recorded less than 48 h before slaughter. Ultrasound
measurements were taken by an experienced operator
using a real-time ultrasound device (Ultrascan50, Al-
liance Médicale Inc., Montreal, Canada) with a 120-
mm, 3.5-MHz linear probe. Lambs were restrained
and measured in a standing position on a preparation
table to minimize errors related to movement and tis-
sue compression. Before each ultrasound session, the
different scanning sites were sheared with a surgical
blade (0.1 mm) and a conductive solution (mineral oil
or P-net, DGF, Pintendre, Canada) was applied. De-
pending on the scanning site, a flat or curved gel pad
(Superflab, Mick Radio Nuclear Instruments, Bronx,
NY) was placed under the probe coated with ultra-
sound gel (Ecogel200, Eco-Med Pharmaceutical Inc.,
Mississauga, Canada). The flat gel pad was fitted with
guides (at 4 cm and 11 cm) to assist in the longitudinal
measurements.
Ultrasound measurements were taken on the left
side at 4 sites on the live animal: total tissue depth
(GRus) between the 11th and 12th ribs, 11 cm lateral
to the spine and parallel to it (longitudinal measure-
ment, flat gel pad; Figure 1); fat depth (FD12us) and
LM depth (LD12us) between the 12th and 13th ribs
perpendicular to the body midline (transverse mea-
surement, curved gel pad); fat depth (FD3Tus) and
LM depth (LD3Tus) between the 3rd and 4th lumbar
vertebrae, taken perpendicular to the spine (transverse
measurement, curved gel pad); fat depth (FD3Lus) and
LM depth (LD3Lus) between the 3rd and 4th lumbar
vertebrae, taken parallel to the body midline (longitu-
dinal measurement, flat gel pad). The same operator
performed the ultrasound measurements throughout
the experiment. The images were captured at each site,
and measurements were taken immediately using the
cursor of the device. For the transverse measurements,
the probe was placed perpendicular to the backbone
capturing the entire lamb chop from which the maximal
height of the LM, perpendicular to the surface, and
the fat depth over this height was assigned respectively
to muscle and fat depths (Pálsson, 1939). Longitudinal
measurements, parallel to the backbone, captured an
image of the LM over its length. In this case, the muscle
depth corresponded to maximal height between trans-
verse processes and fat depth was the fat cover over this
muscle depth (Figure 2). Skin depth was included in all
the ultrasound fat measurements because this tissue is
not easily distinguishable from the fat layer. Skin layer
was thin (2.5 to 3.0 mm) and showed little between-
animal variation (Gooden et al., 1980; Cameron and
Bracken, 1992). Our measurements and analyses of skin
thickness performed in this experiment at around 110 d
of age corroborated these observations (3.5 ± 0.4 mm;
data not shown).
Slaughter and Grading
Lambs were slaughtered weekly on a fixed day for
the entire duration of the experiment. After feed with-
drawal for at least 12 h, the BW was recorded before
slaughter. Lambs were slaughtered in a commercial ab-
attoir. Particular attention was given to the pelt re-
Thériault et al.
1802
at Bibliotheque De L'Univ. Laval Section Des Acquisitions on April 27, 2009. jas.fass.orgDownloaded from
moval to keep the subcutaneous fat intact. Hanging
carcasses were weighed before chilling at 4°C. After 24
h of chilling, the carcasses were graded according to the
method of Agriculture Canada (1992). Using a metal
ruler, total tissue depth was measured over the 12th
rib, 11 cm from the midline (GR measurement). Final-
ly, the carcasses were split longitudinally, and the left
half-carcasses were sent to Agriculture and Agri-Food
Canada’s Dairy and Swine Research and Development
Centre.
Cuts and Carcass Measurements
Five days after slaughter, the half-carcasses were cut
into primary cuts (shoulder, loin, leg, and flank). Two
cuts parallel to the ribs were made in the loin region,
one between the 12th and the 13th ribs and one behind
the 13th rib, to extract the last chop. This chop was
digitized using an image digitizer (Scanmaker 2, Mi-
crotek, Taiwan) at a resolution of 100 pixels per inch.
Similarly, a cut was made between the 3rd and 4th lum-
bar vertebrae, and an image of this surface (posterior
region of the loin) was digitized. Measurements corre-
sponding to the live ultrasound measurements, fat depth
(FD12: between the 12th and 13th ribs; FD3: between
the 3rd and 4th lumbar vertebrae), LM depth (LD12:
between the 12th and 13th ribs; LD3: between the 3rd
and 4th lumbar vertebrae), and LM area (LMA12: be-
tween the 12th and 13th ribs; LMA3: between the 3rd
and 4th lumbar vertebrae), were evaluated using image
analysis software (Pomar et al., 2001).
Statistical Analyses
Analyses were performed on animals having valid
data for all the studied variables. Pearson’s correla-
tion coefficients between the ultrasound measurements
and the corresponding carcass measurements were cal-
culated using the CORR procedure (SAS Inst. Inc.,
Cary, NC), indicating the intensity of the relationship
between these 2 sets of variables. Coefficients of corre-
lation (r) and determination (r2) are, however, strongly
influenced by the population distribution (Houghton
and Turlington, 1992). For the purpose of comparing
studies, it is therefore preferable to refer to the residual
SD (RSD) of the relationship between the carcass and
ultrasound measurements. The relationship between ul-
trasound measurements (dependent variable) and mea-
surements taken on the digitized images (independent
variable) was studied using the SAS REG procedure.
The inverse relationship was also determined, with the
same SAS procedure, to compare our results with those
of other studies. Outliers and data having undue in-
fluence were identified using influence diagnostics and
graphic analysis.
Additionally, error decomposition was used to deter-
mine the accuracy of the ultrasound measurements, in
terms of trueness and precision (ISO, 1993). According
to the method described by Theil (1966), the total mea-
surement error is equal to the mean square prediction
error (MSPE). In our case, the error is the difference
between measurements taken on the carcass and the
value obtained using ultrasound imaging; that is,
MSPE carcassultrasound
n
ii
=-
å()
.
2
The MSPE also is equal to the square of the root mean
square error (RMSE) as described by Herring et al.
(1994). Graphically, the error represents the difference
between each point (ultrasound measurement) and the
line of identity (perfect match between carcass and ul-
trasound measurements). The MSPE can be broken
down into 3 components: error of central tendency
(ECT), error due to regression (ER) and error due
to disturbance (ED), as proposed by Benchaar et al.
(1998) and Pomar and Marcoux (2005). The ECT eval-
uates the closeness of the agreement between the mean
value obtained using an instrument and the accepted
reference value. The ECT is equal to the square of the
bias of the ultrasound measurements (bias = mean dif-
ference between the ultrasound and carcass measure-
ments), as used in studies of measurement precision
(Moeller and Christian, 1998; Greiner et al., 2003).
Figure 1. Longitudinal ultrasound image of total tissue depth be-
tween 11th and 12th ribs at 11 cm from the midline (GR) in lamb.
Ribs are designated by letters R.
Accuracy of ultrasound measurements in lamb 1803
at Bibliotheque De L'Univ. Laval Section Des Acquisitions on April 27, 2009. jas.fass.orgDownloaded from
ECT bias mean mean
carcassultrasou
carcassultrasound
i
== -=
-
22
()
(nnd
n
i).
å
æ
è
ç
ç
ç
ç
ç
ö
ø
÷
÷
÷
÷
÷
2
The ER refers to the difference between the slope of
the regression line between the ultrasound and carcass
measurements and the slope (b = 1) of the identity line.
The ED is the component of the error that cannot be
explained by the regression. It represents the dispersion
of the points around the regression line, the random er-
ror. The ED is the square of the RSD.
In the present study, the trueness of the measure-
ments is evaluated as the sum of ECT and ER, where-
as the precision is evaluated by the ED. Presence of
bias does not mean that the measurement is not useful
(ECT > 0). The ECT can easily be corrected by adding
the value of the bias to the ultrasound measurement.
Although it also can be corrected by regression, the ER
implies that the bias is not constant and varies depend-
ing on the magnitude of the measurements. The ED,
on the other hand, cannot be corrected and needs to
be minimized.
Few authors use the SE of prediction (SEP) for
evaluating the precision of ultrasound measurements
(Herring et al., 1994; Moeller and Christian, 1998). The
SEP is similar to the RMSE of Herring et al. (1994):
SEP carcassultrasoundbias
n
ii
=--
-
å()
.
2
1
Thus, an SEP of 1.5 mm for fat depth indicates that,
in 68% of cases, an ultrasound measurement will be
within 1.5 mm of the carcass measurement (Moeller and
Christian, 1998). Within the SEP, the ER and the ran-
dom error are merged when n is very large, because
ER ED carcassultrasoundbias
n
ii
+= --
å()
.
2
The error decomposition method proposed here gives
additional information about the type of errors made
with ultrasound imaging. The SEP was presented for
the purposes of comparison with previous published re-
sults. All error calculations were performed using SAS
software.
RESULTS AND DISCUSSION
Data from 96 lambs (44 SU and 52 DP) were used
in this study. Information from other animals was re-
moved from the data set because of mortality of the
animal, disease, or because carcasses were not properly
split. Average daily gain of the lambs was 417 g/d and
ranged between 272 and 620 g/d (data not shown).
Lambs were slaughtered at an average age of 130.9 d
with an average fasted BW of 47.0 kg and HCW of 24.4
kg (Table 1). The high variability showed in Table 1 for
carcass fat depth reflects the discrepancy in the pattern
of fattening of the 2 breeds of lamb used in this study
(maternal and terminal types).
Relationship Between Live Ultrasound
and Carcass Measurements
Total Tissue Depth. Ultrasound GR measure-
ments (GRus) and those measured on the carcass with
a ruler (GR) were correlated (Table 2; r = 0.83; P <
0.001) in agreement with other studies (McEwan et al.,
1989; Ramsey et al., 1991; Hopkins et al., 1993). The
GR also was correlated with ultrasound fat depth mea-
surements (FD3Tus, FD3Lus, FD12us; 0.76 ≤ r ≤ 0.81;
P < 0.001).
Fat Depth. Correlations between the various ultra-
sound and corresponding carcass fat depth measure-
ments on digitized images were high (Table 2; r = 0.82,
0.78, and 0.82 for FD12us vs. FD12, FD3Tus vs. FD3,
and FD3Lus vs. FD3, respectively; P < 0.001) and were
similar to the coefficients of correlation reported else-
where (Thompson et al., 1977; Delfa et al., 1991; Fer-
nandes, 2000) but greater than those below 0.6 obtained
by Turlington (1990) and Hopkins et al. (1996).
Figure 2. Longitudinal ultrasound image between the 3rd and 4th
lumbar vertebrae at 4 cm to the midline in lamb. Letters S, F, and M
represent skin, fat, and muscle depths, respectively. Transverse pro-
cesses are identified by letter T.
Thériault et al.
1804
at Bibliotheque De L'Univ. Laval Section Des Acquisitions on April 27, 2009. jas.fass.orgDownloaded from
The RSD obtained in our study were between 1.39
and 2.31 mm (data not shown; ultrasound = indepen-
dent variable). Silva et al. (2006) reported better preci-
sion than our (0.77 ≤ RSD ≤ 0.95 mm) with greater
probe resolution (5 and 7.5 MHz). In ultrasound imag-
ing, it is recommended to use probes with focal depths
close to the tissue of interest (Ginther, 1986). For lambs
with BW of 35 to 55 kg, the depth of fat measurement,
including the thickness of the gel pad (20 mm), ranged
between 25 and 42 mm. For a probe of 3.5 MHz, the
focal depth is around 80 mm. A probe of 5.0 MHz and
having a focal distance about 40 mm improves fat layer
image definition.
LM Depth. For LM depths measured between the
12th and 13th ribs, the coefficients of correlation be-
tween ultrasound and carcass measurements reported
in the literature were generally less than those calculat-
ed for the fat depth measurements (0.4 ≤ r ≤ 0.7; For-
tin and Shrestha, 1986; McEwan et al., 1989; Hopkins
et al., 1996). In our study, a correlation of 0.34 (Table
2; P < 0.001) was found between LD12us and the cor-
responding carcass measurement on digitized images.
In the lumbar region, the coefficients of correlation be-
tween the LD3Tus and LD3Lus with the LD3 measure-
ments were 0.43 and 0.42, respectively. Values reported
by Fernández et al. (1998) and Fortin and Shrestha
(1986) for measurement at this site ranged from 0.49
to 0.76. Coefficient of correlation between measured
and reference values for a given measuring device are
dependent on both the precision of the device (RSD)
and the SD of the population under study. In fact, for
a given device and measured variable, the coefficient of
correlation increases with the increase of the popula-
tion variation. Therefore, the low muscle depth vari-
ability observed in the studied population (Table 1; SD
= 2.6 mm) could explain the reduced correlations ob-
served in our study compared with those greater than
0.85 observed by Silva et al. (2006) and Binnie et al.
(1995). In both studies, SD of muscle depth was greater
than 5 mm. Even with the high correlation, RSD values
observed by Silva et al. (2006) were greater (2.27 ≤
RSD ≤ 4.09 mm) than the 1.74 mm obtained by Binnie
et al. (1995) and ours (2.38 ≤ RSD ≤ 2.42 mm; data
not shown). Despite our low r-value, the RSD obtained
for ultrasound and carcass muscle depth regressions are
comparable with or less than RSD between 2.4 and
2.8 mm observed by McEwan et al. (1989) and Hopkins
et al. (1996).
Errors of Live Ultrasound Measurements
A measurement taken using a given device corresponds
to the sum of the true value and the measurement er-
ror. Magnitude of the error will vary depending on the
accuracy of the device; it is this error that is presented
in Table 3. However, it is important to note that in
Table 1. Means, SD, CV, and minimum and maximum values for live lambs and carcass traits (n = 96)
Trait Mean SD CV, % Minimum Maximum
Live, before slaughter
Age, d 130.9 19.5 14.9 96.0 177.0
Empty BW, kg 47.0 5.4 11.5 36.0 55.8
Total tissue depth, mm 23.6 4.0 16.9 14.5 31.6
Ultrasound fat depth, mm
12th-13th rib 8.5 1.6 18.8 5.9 12.0
3rd-4th lumbar vertebrae transverse 8.8 1.7 19.3 5.2 13.1
3rd-4th lumbar vertebrae longitudinal 8.9 1.7 19.1 5.9 13.8
Ultrasound LM depth, mm
12th-13th rib 31.3 2.1 6.7 27.4 35.9
3rd-4th lumbar vertebrae transverse 31.5 2.0 6.3 26.6 36.1
3rd-4th lumbar vertebrae longitudinal 30.8 2.3 7.5 24.5 36.3
Ultrasound sum of depths, mm
12th-13th rib 39.8 3.0 7.5 34.0 46.5
3rd-4th lumbar vertebrae transverse 40.2 2.9 7.2 33.7 46.9
3rd-4th lumbar vertebrae longitudinal 39.8 3.2 8.0 31.1 47.9
Carcass
HCW, kg 24.4 3.0 12.3 18.1 29.5
Total tissue depth, mm 16.4 4.4 26.8 8.0 25.0
Fat depth, mm
12th-13th rib 6.2 2.4 38.7 2.1 12.7
3rd-4th lumbar vertebrae 7.9 3.7 46.8 1.8 17.9
LM depth, mm
12th-13th rib 33.7 2.6 7.7 27.4 40.1
3rd-4th lumbar vertebrae 34.0 2.6 7.6 28.6 41.9
Sum of depths, mm
12th-13th rib 39.8 3.3 8.3 31.8 47.3
3rd-4th lumbar vertebrae 41.9 3.7 8.8 33.2 49.5
LM area, mm2
12th-13th rib 1,628.0 169.4 10.4 1,133.0 1,929.0
3rd-4th lumbar vertebrae 1,682.0 207.1 12.3 1,172.0 2,213.0
Accuracy of ultrasound measurements in lamb 1805
at Bibliotheque De L'Univ. Laval Section Des Acquisitions on April 27, 2009. jas.fass.orgDownloaded from
Table 2. Simple coefficient correlation between live ultrasound measurements and corresponding carcass measurements in lamb (n = 96)
Variable1GRus FD12us FD3Tus FD3Lus LD12us LD3Tus LD3Lus GR FD12 FD3 LD12 LD3 LMA12 LMA3
GRus 1.00 0.74*** 0.76*** 0.78*** 0.52*** 0.49*** 0.57*** 0.83*** 0.72*** 0.71*** 0.00 0.03 0.13 0.08
FD12us 1.00 0.89*** 0.91*** 0.35*** 0.26** 0.30** 0.78*** 0.82*** 0.82*** −0.08 −0.16 −0.06 −0.18
FD3Tus 1.00 0.97*** 0.33*** 0.25* 0.32*** 0.76*** 0.76*** 0.78*** −0.10 −0.14 −0.01 −0.12
FD3Lus 1.00 0.34*** 0.25* 0.32*** 0.81*** 0.80*** 0.82*** −0.09 −0.15 0.01 −0.15
LD12us 1.00 0.72*** 0.69*** 0.43*** 0.43*** 0.37*** 0.34*** 0.38*** 0.46*** 0.43***
LD3Tus 1.00 0.90*** 0.42*** 0.30** 0.34*** 0.30** 0.43*** 0.51*** 0.53***
LD3Lus 1.00 0.49*** 0.34*** 0.36*** 0.30** 0.42*** 0.48*** 0.52***
GR 1.00 0.78*** 0.78*** −0.02 −0.09 0.07 −0.03
FD12 1.00 0.82*** −0.13 −0.15 −0.01 −0.09
FD3 1.00 −0.08 −0.35*** −0.06 −0.22*
LD12 1.00 0.34*** 0.72*** 0.48***
LD3 1.00 0.55*** 0.79***
LMA12 1.00 0.73***
LMA3 1.00
1GRus = ultrasound GR—total tissue depth between the 11th and 12th ribs at 11 cm from the midline, longitudinal measure; FD12us = ultrasound fat depth between the 12th and 13th ribs,
transverse measure; FD3Tus = ultrasound fat depth between the 3rd and 4th lumbar vertebrae, transverse measure; FD3Lus = ultrasound fat depth between the 3rd and 4th lumbar vertebrae, longi-
tudinal measure; LD12us = ultrasound LM depth between the 12th and 13th ribs, transverse measure; LD3Tus = ultrasound LM depth between the 3rd and 4th lumbar vertebrae, transverse measure;
LD3Lus = ultrasound LM depth between the 3rd and 4th lumbar vertebrae, longitudinal measure; GR = total tissue depth over the 12th rib at 11 cm from the midline, Canadian carcass grading
site; FD12 = fat depth between the 12th and 13th ribs on digitized image; FD3 = fat depth between the 3rd and 4th lumbar vertebrae on digitized image; LD12 = LM depth between the 12th and
13th ribs on digitized image; LD3 = LM depth between the 3rd and 4th lumbar vertebrae on digitized image; LMA12 = LM area between the 12th and 13th ribs on digitized image; LMA3 = LM
area between the 3rd and 4th lumbar vertebrae on digitized image.
*P < 0.05; **P < 0.01; ***P < 0.001.
Thériault et al.
1806
at Bibliotheque De L'Univ. Laval Section Des Acquisitions on April 27, 2009. jas.fass.orgDownloaded from
our study, measurements taken on the digitized images
were used as a reference and considered as true values.
Although they represent an acceptable compromise in
terms of technical feasibility, carcass measurements are
not a perfect reference and also entail their own degree
of error that is included in the total measurement error.
We used the error decomposition method to study the
accuracy of ultrasound measurements in terms of preci-
sion and trueness and to understand the nature of the
disagreements between the measurements taken on the
live animal and the reference measurements (Table 3).
Total Tissue Depth. The GRus measurement ex-
hibited a significant bias as indicated by the ECT error,
which represents 87% of the MSPE error (Table 3). The
ER of the GRus measurement was low and represented
only 2% of the total error. However, carcasses with a
smaller GR were more overestimated by ultrasonogra-
phy than those with a larger GR, which is indicated by
a regression slope less than unity (P < 0.001; Figure
3). Finally, the ED was 4.91 and represents 8.5% of the
total error.
Because GR ultrasound measurement included the
skin, the GRus should be approximately 3.5 mm greater
than the carcass measurement. In addition, the GRus
was evaluated between the ribs, whereas the GR was
obtained with a ruler knife directly on the rib. In our
preliminary trials, repeatable assessment of total tissue
depth was obtained only between the 12th and 13th
ribs due to the presence of connective tissues, giving a
clear boundary with ultrasound device. These 2 vari-
ants could partly explicate the difference of 7.19 mm
observed between GRus and GR (ECT). Additionally,
the greater propensity of thicker tissue to be compressed
by the pressure on the ultrasound probe compared with
thinner tissue (Purchas and Beach, 1981) could explain
the error due to regression for GR measurement. Final-
ly, our observations showed that the movements of the
animal during the scanning process (movement of the
head, breathing, etc.) influenced the total tissue depth,
as it is generally admitted for all ultrasound measure-
ments (Stouffer, 2004). Animal movement, combined
with the differences specific to the measurements them-
selves (ultrasound vs. carcass), could explain the ED of
the measure.
Fat Depth. For ultrasound fat depths, the MSPE
and the trueness (ECT + ER) were similar between
the different scanning sites (Table 3). It was mainly in
the partitioning of the systematic errors (ECT and ER)
that differences between scanning sites were noted. In
the thoracic region (FD12us), most of the error was due
to central tendency (ECT/MSPE = 74%). If the skin
was the only source of discrepancy between the ultra-
sound and carcass fat depth measurements, it would
be plausible to obtain a difference of approximately
3.5 mm, or an ECT of 12.3 (3.52). However, the ER
Table 3. Accuracy of live ultrasound measurements (dependent variable) relative to carcass measurements (inde-
pendent variable) in lamb1 (n = 96)
Dependent variable r2RSD, mm CVe, % MSPE ECT ER ED SEP, mm
Total tissue depth 0.689 2.24 9.51 57.70 51.69 1.10 4.91 2.47
Fat depth
12th-13th rib 0.670 0.92 10.74 7.70 5.68 1.20 0.82 1.43
3rd-4th lumbar vertebrae trans. 0.615 1.06 12.08 7.53 0.85 5.58 1.10 2.60
3rd-4th lumbar vertebrae long. 0.667 0.97 10.87 7.52 1.18 5.41 0.93 2.53
LM depth
12th-13th rib 0.113 1.99 6.35 12.82 5.56 3.40 3.86 2.71
3rd-4th lumbar vertebrae trans. 0.185 1.78 5.65 12.66 6.43 3.14 3.10 2.51
3rd-4th lumbar vertebrae long. 0.179 2.06 6.70 16.94 10.02 2.74 4.17 2.64
Sum of depths
12th-13th rib 0.457 2.26 5.66 6.44 0.00 1.46 4.98 2.55
3rd-4th lumbar vertebrae trans. 0.694 1.61 4.01 6.82 2.60 1.67 2.54 2.06
3rd-4th lumbar vertebrae long. 0.697 1.78 4.47 8.47 4.32 1.05 3.10 2.05
1RSD = residual SD; CVe = CV of the residuals; MSPE = mean square prediction error, total measurement error; ECT = error of central
tendency; ER = error due to regression; ED = error due to disturbance; SEP = SE of prediction; trans. = transverse; and long. = longitudinal.
Figure 3. Relationship between live ultrasound total tissue depth
and carcass total tissue depth measurements in lamb (n = 96). Re-
gression line (– – –). Solid line represents perfect relationship between
ultrasound and carcass measurement, y = x.
Accuracy of ultrasound measurements in lamb 1807
at Bibliotheque De L'Univ. Laval Section Des Acquisitions on April 27, 2009. jas.fass.orgDownloaded from
also must be taken into account, which was low for the
FD12us (1.20), but high for the FD3Tus and FD3Lus
measurements (5.58 and 5.41, respectively). These val-
ues indicate that the slope of the regression line was
different from unity for the 3 fat depths (Figure 4) and
that ultrasound measurements tend to overestimate
the fat thickness in lean animals and to underestimate
these measurements in the fat ones. This observation
is reported by several authors for lambs (Fernandes,
2000), cattle (Greiner et al., 2003), and pigs (Moeller
and Christian, 1998). However, after correcting our fat
depth measurements to subtract the skin thickness, it
appears that the ultrasound fat depths were truer in
leaner lambs but underestimated carcass fat depth in
fatter lambs (data not shown). These results were in
agreement with those of Robinson et al. (1992) in cattle
and those of Purchas and Beach (1981) and Fernandes
(2000) in lambs. In the lumbar region, the underestima-
tion of fat depth by the ultrasound measurements was
even greater in fatter lambs. In fact, for these measure-
ments, the MSPE was mainly due to ER. For all fat
measurements, precision was good as indicated by their
low ED and RSD of around 1.0 mm (Table 3).
For fat depth measurements, the root squares of the
MSPE are comparable with the RMSE of 2.7 to 3.3 mm
obtained between the 12th and 13th ribs by Herring et
al. (1994) in cattle. In addition, the SEP of the ul-
trasound fat measurements calculated here (Table 3)
was equal to the 1.4 obtained by Leeds et al. (2008)
and compared favorably with those ranging from 1.8
to 3.2 mm reported by various authors in pigs (Moeller
and Christian, 1998; Schwab et al., 2003) and cattle
(Herring et al., 1994; Greiner et al., 2003). According to
Tait et al. (2005), an SEP less than 2.54 mm would be
an acceptable standard for fat measurements between
the 12th and 13th ribs in lambs.
Fat depths were greater in the carcass than in the
equivalent ultrasound measurements. This ECT ap-
pears to be mainly related to differences between ul-
trasound and reference measurements: inclusion of the
skin in the ultrasound measurements but not in the
carcass measurements, ultrasound measurements on
hot living tissue vs. reference measurements on chilled
dead tissue, etc. Pelt removal and carcass hanging can
generate ECT, causing, respectively, expansion of fat
layers (Robinson et al., 1992) and sliding of fat from
the posterior (fatter) region toward the anterior, both
modifying the fat depth measurement of hanged car-
casses compared with live animals in a standing po-
sition (Mersmann, 1982; Turlington, 1990; Robinson
et al., 1992). Moreover, because they have a greater
influence in fatter than in leaner animals or tissues,
these 2 phenomena together with pressure on the scan-
ning probe can explain the observed ER values at given
measurement sites and the difference of ER values be-
tween sites (Purchas and Beach, 1981; Robinson et al.,
1992).
The ER in the lumbar region can be explained by the
difference in tissue depth between sites (Table 1; 7.9 vs.
Figure 4. Relationship between live ultrasound fat depths and cor-
responding carcass measurements on digitized image in lamb (n = 96)
a) between the 12th and 13th ribs; b) between the 3rd and 4th lumbar
vertebrae, transverse; and c) between 3rd and 4th lumbar vertebrae,
longitudinal. Regression line (– – –). Solid line represents perfect rela-
tionship between ultrasound and carcass measurement, y = x.
Thériault et al.
1808
at Bibliotheque De L'Univ. Laval Section Des Acquisitions on April 27, 2009. jas.fass.orgDownloaded from
6.2 mm; P < 0.001, for FD3 vs. FD12). The presence of
a third fat layer between the 3rd and 4th lumbar ver-
tebrae was observed in some digitized images (Figure
5), but never between the 12th and 13th ribs. Appear-
ance of this additional fat layer has been documented
in hogs (Fortin, 1986) but not in lambs. Lambs show-
ing a third fat layer were fatter than those without
it (31.8 vs. 23.1% dissected fat, respectively; data not
shown). In addition, this fat layer seemed to increase
from the ventral end of the loin eye muscle toward the
backbone. We presumed that the third fat layer was an
image artifact because its appearance in the images was
inconsistent and because its definition at the loin ex-
tremity was usually poor. Ultrasound fat depth in fat-
ter lambs having a third fat layer was underestimated
by the exclusion of this unclear layer. Leaner lambs
had better agreement between carcass and ultrasound
measurements, probably as a result of the absence of
this third fat layer.
As previously explained, the low probe resolution
adds imprecision, increasing ED, in fat depth measure-
Figure 5. Digitized images between the 3rd and 4th lumbar ver-
tebrae in lamb carcass. Fat depth is designated by letter F. Third fat
layer (F3) is apparent only on the second image.
Figure 6. Relationship between live ultrasound LM depths and
corresponding carcass measurements on digitized image in lamb (n
= 96) a) between the 12th and the 13th ribs; b) between 3rd and
4th lumbar vertebrae, transverse; and c) between 3rd and 4th lumbar
vertebrae, longitudinal. Regression line (– – –). Solid line represents
perfect relationship between ultrasound and carcass measurement, y
= x.
Accuracy of ultrasound measurements in lamb 1809
at Bibliotheque De L'Univ. Laval Section Des Acquisitions on April 27, 2009. jas.fass.orgDownloaded from
ments. Even if it was easier to distinguish the probe/
skin interface than the skin/fat interface (Alliston,
1983), variation in skin thicknesses, as included in fat
measurements, also increases ED in ultrasound fat
depths. Pelt removal could also generate random er-
rors of variable magnitude because some fat may be
torn off the carcass (McLaren et al., 1991; Young and
Deaker, 1994). Carcass fabrication and the handling of
the pieces during image digitization could have caused
some deformation in fat layers and thus increased ED.
According to Pomar et al. (2001), it would be prefer-
able to freeze the cuts before digitizing them to allow
the fat to solidify and minimize fat layer deformation.
LM Depth. The magnitude of ultrasound LM mea-
surement errors was greater than those observed for fat
depths. Within scanning sites, the MSPE was greater
for LD3Lus than for LD12us and LD3Tus measurements
(Table 3). This difference was mainly caused by a larger
ECT of the LD3Lus measurement (Table 3; 10.02 vs.
5.56 and 6.43, for LD3Lus vs. LD12us and LD3Tus, re-
spectively). The ER was less for the LD3Lus measure-
ment, but the slopes of the lines between ultrasound
and carcass LM depths were significantly different from
unity in all cases (P < 0.001; Figure 6). Several authors
reported such slopes meaning that ultrasound measure-
ments overestimate the smaller LM and underestimate
the larger ones in hogs (Moeller and Christian, 1998),
lambs (Fernandes, 2000), and cattle (Greiner et al.,
2003). Precision of ultrasound LM depths (3.10 ≤ ED ≤
4.17) and their relationship with carcass measurements
(0.11 ≤ r2 ≤ 0.19) were low, even if the RSD and coeffi-
cient of variation of residuals (CVe) values appeared to
be acceptable (Table 3). Despite the fact that the stud-
ied population was composed of lambs of 2 breeds with
very different growth characteristics, the SD for carcass
muscle depth was 2.6 mm, and the majority of this
variation, nearly 2.0 mm (RSD), remained unexplained
by ultrasound measurements. Leeds et al. (2008) also
concluded that the low variability of the LM depth, in
comparison with LM area, reduced its usefulness in pre-
diction of carcass yield and value in lamb. With such
muscle depth variability and such random error, it ap-
pears that ultrasound measurement could not be used
to distinguish between the heavier- and lighter-muscled
lambs in the current study. Nevertheless, the SEP of
the ultrasound LM measurements (Table 3) is compa-
rable with the 2.6 mm reported by Leeds et al. (2008).
Comparison between species is limited because most of
the authors measured loin area rather than depth.
Difficulty of identifying the deepest part of the LM
due to its proximity to the ribs could explain the high
ED of the LD12us measurements. Young et al. (1992)
mentioned that the pressure on the probe deforms the
fat layer uniformly, but not the muscle, due to the pres-
ence of the ribs in the thoracic region. On the other
hand, the LD3Tus and LD3Lus random errors could be
attributed to the incorrect interpretation of the fat-
muscle boundaries. Underestimation of the fat depth
resulting from the unintentional exclusion of the third
Figure 7. Relationship between sum of fat + muscle depths taken
by ultrasound and sum of corresponding fat + muscle depths on car-
cass digitized image in lamb (n = 96) a) between the 12th and 13th
ribs; b) between the 3rd and 4th lumbar vertebrae, transverse; and c)
between 3rd and 4th lumbar vertebrae, longitudinal. Regression line
(– – –). Solid line represents perfect relationship between ultrasound
and carcass measurement, y = x.
Thériault et al.
1810
at Bibliotheque De L'Univ. Laval Section Des Acquisitions on April 27, 2009. jas.fass.orgDownloaded from
fat layer could cause overestimation of muscle depths.
To confirm these hypotheses, the sum of the ultra-
sound fat and muscle depths was compared with the
sum of the same values assessed on the digitized im-
ages (Table 3 and Figure 7). Because the skin bound-
ary (gel pad/skin interface) was easily identifiable, the
agreement of these ultrasound values with the carcass
measurements means that the lower muscle boundary
were clearly distinguished on the ultrasound images.
As expected, the sum of depth in the thoracic region
contained a high proportion of random error (Table
3; ED/MSPE = 77.35%) compared with the other to-
tal measurements. Difficulty of distinguishing the LM
end appears to explain the majority of the error for
ultrasound total depths at this site, and then, muscle
depth as speculated previously. Conversely, ultrasound
and carcass sums of depths between the 3rd and 4th
lumbar vertebrae were strongly related (R2 of 0.69 and
0.70 and RSD of 1.61 and 1.78 mm, for transverse and
longitudinal measures, respectively; Table 3 and Figure
7). Errors of these measures were, in fact, smaller than
those of the fat and muscle depths at this site. In the
lumbar region, the inner muscle boundary seems to be
clearly discerned on the ultrasound images. These re-
sults suggest that the difficulty in distinguishing the
fat/muscle boundary can reduce accuracy of muscle
and fat measurements in the lumbar region and that
image interpretation errors related to the presence of
the third fat layer were responsible for the inaccuracy
of these measurements between the 3rd and 4th lumbar
vertebrae.
In regard to carcass measurements, changes on LM
shape occurring during postslaughter chilling, hanging
and handling of the carcass could be implicated in the
lack of agreement between live ultrasound and carcass
measurements (Fortin and Shrestha, 1986; Turlington,
1990; Hopkins et al., 1993). However, it is difficult to
quantify how these phenomena affect muscle shape and
the type of errors that will occur. Freezing the meat
pieces before digitization could, as for fat, minimize
muscle deformation (Pomar et al., 2001). Binnie et al.
(1995) has performed measurements on frozen carcasses
and obtained better precision (RSD = 1.74 mm).
Finally, the ED and ECT for longitudinal measure-
ments could partly be explained by the variation in the
angle of wave penetration and, hence, by the angle of
measurement of LM depth. Depths will be larger when
the probe is perpendicular to the skin and smaller when
it is directed toward the backbone or toward the side of
the animal (Youssao et al., 2002).
Comparing Measurements
and Measuring Sites
As the accuracy of ultrasound measurement was es-
tablished, comparison of scanning sites can be realized
in regard to both precision (minimizing ED) and tech-
nical considerations.
Total Tissue Depth. Showing r-values comparable
with those observed for fat depths, the GR ultrasound
measurement appears to offer advantages in terms of
accuracy, despite the greater absolute value of its ran-
dom error (Table 3). In term of relative error variation
(Table 3; CVe), GR seems to be slightly more precise
than fat depths. Difficulty of distinguishing small dif-
ferences in depth together with image interpretation er-
rors would proportionally be less important when tissue
depth or variability increase (Thompson et al., 1977;
Simm, 1992; Young and Deaker, 1994), giving advan-
tage to ultrasound GR over fat depth measurements.
Fat Depth. In regard to precision, the ED was
smaller for fat depth between the 12th and 13th ribs
than between the 3rd and 4th lumbar vertebrae. These
findings agree with those of Fernández et al. (1998)
based on coefficient of correlation at both scanning
sites. However, Silva et al. (2006) obtained similar pre-
cisions for these 2 measurements sites (RSD ~0.8 vs.
0.9 mm).
In our study, the growth of a third layer of fat, be-
tween the 3rd and 4th lumbar vertebrae in the fattest
lambs, makes this scanning site less useful in terms of
precision of the fat measurements as well as in terms
of practical application. Image interpretation problems
encountered at this site demonstrate, as reported by
Starck et al. (2001), the importance of having a good
knowledge of the morphology of the studied tissues.
Recognition of the existence of this additional fat layer
will improve the accuracy of fat and muscle measure-
ments at the lumbar region.
LM Depth. From a practical standpoint, measure-
ments near the last ribs are the most used mainly due
to the presence of ribs, which is an easily identifiable
anatomical reference (Alliston, 1983; McLaren et al.,
1991). However, our results, like those of Young et al.
(1992), demonstrate that the proximity between the ribs
reduces the accuracy of the LM measurements. Given
the greater distance between the 3rd and 4th trans-
verse processes than between the ribs, the exact site—
maximum depth between the transverse processes—is
easier to locate. Despite the low precision of ultrasound
muscle depths in the present study, slight advantage
goes to transverse measurements in the lumbar region,
as confirmed by the greater r-values reported by Jensen
(1977) and Silva et al. (2006) in the lumbar compared
with the thoracic region. Results of Miles et al. (1972)
in cattle also indicated that the boundary of the LM is
more clearly defined in the lumbar than in the thoracic
region.
Transverse vs. Longitudinal Measurements.
This study is the first to compare the accuracy of trans-
verse and longitudinal ultrasound measurements in
lambs. Longitudinal measurements are very popular in
hogs, and the work of Cisneros et al. (1996) demonstrat-
ed that there was no difference in the precision of the 2
types of measurement in this species. However, our re-
sults showed that transverse measurements were more
precise than longitudinal ones for lumbar muscle depth
Accuracy of ultrasound measurements in lamb 1811
at Bibliotheque De L'Univ. Laval Section Des Acquisitions on April 27, 2009. jas.fass.orgDownloaded from
(ED = 4.17 vs. 3.10; Table 3) and sum of depths (ED
= 2.54 vs. 3.10), most likely because of the greater sta-
bility in the angle of wave penetration. For ultrasound
fat depth, the precision of longitudinal measurements
compared favorably with transverse measurements (ED
= 0.93 vs. 1.10; r2 = 0.67 vs. 0.62). Longitudinal mea-
surements were not performed between the 12th and
13th ribs because of the irregularity of the vertical line
due to the presence of the ribs.
Conclusions
The primary focus of this study was to understand
the discrepancies between ultrasound measurements
taken on the live animal and carcass measurements.
In vivo fat and GR depths in lambs were successfully
measured using real-time ultrasound. However, in the
population studied, variability in muscle depths was
minimal, and ultrasound muscle depth measurements
were neither precise nor correlated with carcass mea-
surements. Therefore, we were unable to use real-time
ultrasound measurements of muscle depth to rank the
lambs according to their degree of muscle development.
When only fat depth was assessed, the site between the
12th and 13th ribs seemed to be the most appropri-
ate for measurement, but the accuracy of muscle depth
measurement at this site was low. Transverse measure-
ments of fat and muscle depths between the 3rd and 4th
lumbar vertebrae were of an acceptable accuracy and
were more accurate than longitudinal measurements.
Based on the results of this study, we suggest that par-
ticular attention should be given to image interpreta-
tion to correctly identify and measure all fat layers.
Moreover, because the depth of tissue studied in lamb
was small, unlike in swine, ultrasound fat thickness in
lambs might be better evaluated with probe resolution
greater than 3.5 MHz. Ultrasound GR showed potential
to be included in genetic selection programs, especially
in countries where the GR measurement was used to
estimate carcass quality. Ultimately, further studies are
needed to establish the relationships between these dif-
ferent ultrasound measurements and carcass composi-
tion (fat, muscle, and bone) with the intention of pre-
dicting carcass quality.
LITERATURE CITED
Agriculture Canada. 1992. Livestock and poultry carcass grading
regulation (SOR/92–541). Canada Gazette 126(Part II):3848.
Alliston, J. C. 1983. Evaluation of carcass quality in the live animal.
Pages 75–95 in Sheep Production. W. Haresign, ed. Butter-
worths, London, UK.
Benchaar, C., J. Rivest, C. Pomar, and J. Chiquette. 1998. Pre-
diction of methane production from dairy cows using existing
mechanistic models and regression equations. J. Anim. Sci.
76:617–627.
Berg, E. P., M. K. Neary, J. C. Forrest, D. L. Thomas, and R. G.
Kauffman. 1996. Assessment of lamb carcass composition from
live animal measurement of bioelectrical impedance or ultra-
sonic tissue depths. J. Anim. Sci. 74:2672–2678.
Binnie, D. B., R. J. Farmer, and J. N. Clarke. 1995. Ultrasonic
scanning of lamb carcasses for non-destructive carcass quality
measurements. Proc. N.Z. Soc. Anim. Prod. 55:111–113.
Cameron, N. D., and J. Bracken. 1992. Selection for carcass lean
content in a terminal sire breed of sheep. Anim. Prod. 54:367–
377.
Canadian Council on Animal Care. 1993. Guide to the care and
use of experimental animals. D. Olfert, B. M. Cross, and A.
A. McWilliam, ed. Vol. 1. 2nd ed. CCAC, Ottawa, Ontario,
Canada.
Cisneros, F., M. Ellis, K. D. Miller, J. Novakofski, E. R. Wilson, and
F. K. McKeith. 1996. Comparison of transverse and longitudinal
real-time ultrasound scans for prediction of lean cut yields and
fat-free lean content in live pigs. J. Anim. Sci. 74:2566–2576.
Delfa, R., A. Teixeira, I. Blasco, and F. Colomer-Rocher. 1991. Ul-
trasonic estimates of fat thickness, C measurement and longis-
simus dorsi depth in Rasa Aragonesa ewes with same body
condition score. Options Méditerranéennes Série Séminaires.
13:25–30.
Fernandes, T. 2000. Genetic parameter estimates for ultrasound-
measured carcass traits in sheep. MSc Thesis. Univ. Guelph,
Ontario, Canada.
Fernández, C., A. García, H. Vergara, and L. Gallego. 1998. Us-
ing ultrasound to determine fat thickness and longissimus dorsi
area on Manchego lambs of different live weight. Small Rumin.
Res. 27:159–165.
Fortin, A. 1986. Development of backfat and individual fat layers
in the pig and its relationship with carcass lean. Meat Sci.
18:255–270.
Fortin, A., and J. N. B. Shrestha. 1986. In vivo estimation of carcass
meat by ultrasound in ram lambs slaughtered at an average live
weight of 37 kg. Anim. Prod. 43:469–475.
Ginther, O. J. 1986. Ultrasonic imaging and reproductive events in
the mare. Equiservices Publishing, Cross Plains, WI.
Gooden, J. M., A. D. Beach, and R. W. Purchas. 1980. Measurement
of subcutaneous backfat depth in live lambs with an ultrasonic
probe. N. Z. J. Agric. Res. 23:161–165.
Greiner, S. P., G. H. Rouse, D. E. Wilson, L. V. Cundiff, and T. L.
Wheeler. 2003. The relationship between ultrasound measure-
ments and carcass fat thickness and longissimus muscle area in
beef cattle. J. Anim. Sci. 81:676–682.
Herring, W. O., D. C. Miller, J. K. Bertrand, and L. L. Benyshek.
1994. Evaluation of machine, technician, and interpreter effects
on ultrasonic measures of backfat and longissimus muscle area
in beef cattle. J. Anim. Sci. 72:2216–2226.
Hopkins, D. L. 1994. Predicting the weight of lean meat in lamb car-
casses and suitability of this characteristic as a basis for valuing
carcasses. Meat Sci. 38:235–241.
Hopkins, D. L., D. G. Hall, and A. F. Luff. 1996. Lamb carcass
characteristics. 3. Describing changes in carcasses of growing
lambs using real-time ultrasound and the use of these measure-
ments for estimating the yield of saleable meat. Aust. J. Exp.
Agric. 36:37–43.
Hopkins, D. L., K. L. Pirlot, A. H. K. Roberts, and A. S. Beattie.
1993. Changes in fat depths and muscle dimensions in grow-
ing lambs as measured by real-time ultrasound. Aust. J. Exp.
Agric. 33:707–712.
Houghton, P. L., and L. M. Turlington. 1992. Application of ultra-
sound for feeding and finishing animals: A review. J. Anim.
Sci. 70:930–941.
ISO. 1993. Statistics—Vocabulary and Symbols Part 1—Probability
and general statistical terms (ISO 3534). Int. Org. Standardiza-
tion, Geneva, Switzerland.
Jensen, N. E. 1977. Preliminary results from Danscan ultrasonic
measurements of fat thickness and loin area of lamb. Pages
74–78 in Proc. 27th Annu. Meet. Eur. Assoc. Anim. Prod.,
Zurich, Switzerland.
Jones, S. D. M., W. M. Robertson, M. A. Price, and T. Coupland.
1996. The prediction of saleable meat yield in lamb carcasses.
Can. J. Anim. Sci. 76:49–53.
Thériault et al.
1812
at Bibliotheque De L'Univ. Laval Section Des Acquisitions on April 27, 2009. jas.fass.orgDownloaded from
Kirton, A. H., and D. L. Johnson. 1979. Interrelationships between
GR and other lamb carcass fatness measurements. Proc. N.Z.
Soc. Anim. Prod. 39:195–201.
Leeds, T. D., M. R. Mousel, D. R. Notter, H. N. Zerby, C. A. Moffet,
and G. S. Lewis. 2008. B-mode, real-time ultrasound for esti-
mating carcass measures in live sheep: Accuracy of ultrasound
measures and their relationships with carcass yield and value.
J. Anim. Sci. 86:3203–3214.
McEwan, J. C., J. N. Clarke, M. A. Knowler, and M. Wheeler. 1989.
Ultrasonic fat depths in Romney lambs and hoggets from lines
selected for different production traits. Proc. N.Z. Soc. Anim.
Prod. 49:113–119.
McLaren, D. G., J. Novakofski, D. F. Parrett, L. L. Lo, S. D. Singh,
K. R. Neumann, and F. K. McKeith. 1991. A study of opera-
tor effects on ultrasonic measures of fat depth and longissimus
muscle area in cattle, sheep and pigs. J. Anim. Sci. 69:54–66.
Mersmann, H. J. 1982. Ultrasonic determination of backfat depth
and loin area in swine. J. Anim. Sci. 54:268–275.
Miles, C. A., R. W. Pomeroy, and J. M. Harries. 1972. Some fac-
tors affecting reproductibility in ultrasonic scanning of animals.
Anim. Prod. 15:239–249.
Moeller, S. J. 2002. Evolution and use of ultrasonic technology in
swine industry. J. Anim. Sci. 80(E Suppl. 2):E19–E27.
Moeller, S. J., and L. L. Christian. 1998. Evaluation of the accuracy
of real-time ultrasonic measurements of backfat and loin muscle
area in swine using multiple statistical analysis procedures. J.
Anim. Sci. 76:2503–2514.
Pálsson, H. 1939. Meat qualities in the sheep with special reference
to Scottish breeds and crosses. I. J. Agric. Sci. 29:544–626.
Pomar, C., and M. Marcoux. 2005. The accuracy of measuring
backfat and loin muscle thicknesses on pork carcasses by the
Hennessy HGP2, Destron PG-100, CGM and ultrasound CVT
grading probes. Can. J. Anim. Sci. 85:481–492.
Pomar, C., J. Rivest, P. Jean dit Bailleul, and M. Marcoux. 2001.
Predicting loin-eye area from ultrasound and grading probe
measurements of fat and muscle depths in pork carcasses. Can.
J. Anim. Sci. 81:429–434.
Purchas, R. W., and A. D. Beach. 1981. Between-operator repeat-
ability of fat depth measurements made on live sheep and lambs
with an ultrasonic probe. N. Z. J. Exp. Agric. 9:213–220.
Ramsey, C. B., A. H. Kirton, B. Hogg, and J. L. Dobbie. 1991. Ultra-
sonic, needle, and carcass measurements for predicting chemical
composition of lamb carcasses. J. Anim. Sci. 69:3655–3664.
Robinson, D. L., C. A. McDonald, K. Hammond, and J. W. Turn-
er. 1992. Live animal measurement of carcass traits by ultra-
sound: Assessment and accuracy of sonographers. J. Anim. Sci.
70:1667–1676.
Schwab, C. R., T. J. Baas, S. J. Moeller, and D. W. Newcom. 2003.
Accuracy of prediction in the national swine improvement fed-
eration (NSIF) ultrasound certification program. J. Anim. Sci.
81(Suppl. 2):43.
Silva, S. R., J. J. Afonso, V. A. Santos, A. Monteiro, C. M. Guedes,
J. M. T. Azevedo, and A. Dias-da-Silva. 2006. In vivo estima-
tion of sheep carcass composition using real-time ultrasound
with two probes of 5 and 7.5 MHz and image analysis. J. Anim.
Sci. 84:3433–3439.
Simm, G. 1992. Selection for lean meat production in sheep. Pages
193–215 in Progress in Sheep and Goat Research. A. W. Speedy,
ed. CAB Int., Edinburg, UK.
Starck, J. M., M. W. Dietz, and T. Piersma. 2001. The assessment of
body composition and other parameters by ultrasound scanning.
Pages 188–210 in Body Composition Analysis of Animals—A
Handbook of Non-Destructive Methods. J. R. Speakman ed.
Cambridge Univ. Press, Cambridge, UK.
Stouffer, J. R. 2004. History of ultrasound in animal science. J.
Ultrasound Med. 23:577–584.
Tait, R., B. Kimm, and D. Morrical. 2005. Accuracy of ultrasound
measures relative to carcass measures of body composition in
sheep. A.S. Leaflet R2046, Dep. Anim. Sci., Iowa State Univ.,
Ames.
Theil, H. 1966. Applied Economic Forecasting. North-Holland Pub-
lishing Company, Amsterdam, the Netherlands.
Thompson, J. M., W. A. Pattie, and R. M. Butterfield. 1977. An
evaluation of the «Scanogram» as an ultrasonic aid in assessing
carcass composition of live sheep. Aust. J. Exp. Agric. Anim.
Husb. 17:251–255.
Turlington, L. M. 1990. Live animal evaluation of swine and sheep
using ultrasonics. MSc Thesis. Kansas State Univ., Manhat-
tan.
Williams, A. R. 2002. Ultrasound applications in beef cattle car-
cass research and management. J. Anim. Sci. 80(E. Suppl.
2):E183–E188.
Wilson, D. E. 1992. Application of ultrasound for genetic improve-
ment. J. Anim. Sci. 70:973–983.
Young, M. J., and J. M. Deaker. 1994. Ultrasound measurements
predict estimated adipose and muscle weights better than car-
cass measurements. Proc. N.Z. Soc. Anim. Prod. 54:215–217.
Young, M. J., J. M. Deaker, and C. M. Logan. 1992. Factors af-
fecting repeatability of tissue depth determination by real-time
ultrasound in sheep. Proc. N.Z. Soc. Anim. Prod. 52:37–39.
Youssao, A. K. I., V. Verleyen, and P. L. Leroy. 2002. Évaluation de
la composition de la carcasse et de la qualité de la viande par
ultrasonographie chez le porc. Ann. Med. Vet. 146:19–26.
Accuracy of ultrasound measurements in lamb 1813
at Bibliotheque De L'Univ. Laval Section Des Acquisitions on April 27, 2009. jas.fass.orgDownloaded from
