Nicole Dion, Danielle Pettigrew and Jean-Paul Daigle
Centre de Développement du Porc du Québec inc. (CDPQ)


At the present time, Canada's swine genetic improvement is based on three main components : adjusted backfat at 100 kg, age at 100 kg and total number piglets born. However, to stay more competitive and to adjust the selection to the market and consumers demands, new traits must be included into the Canadian genetic evaluation program. Meat quality and quantity are some of tomorrow's battle fields. One can use different parameters to measure meat quality and quantity. Some of them are the lean meat yield, loin eye area and marbling.

Lean meat yield can be evaluated from carcass measurements at the slaughterhouse or can be obtained indirectly from the measure of the backfat on live animals (Sather et al., 1987), while loin eye area and marbling are mainly obtained on carcasses.

When you wish to make selection on a specific trait, it is always more efficient when the measure is taken on the animal itself and not sibs. Furthermore, the selection is more accurate when you measure the trait itself and not another trait (except if the correlation between the two is very high). Since the pigs are leaner than before, it should be considered that the correlation between fat thickness and lean meat yield for these leaner pigs may be not the same as for fatter ones. Thus, it is important to look at other ways of predicting lean meat yield.

Even though ultrasound has been used for evaluating live animals for many years (Stouffer et al., 1961) its main use in hogs is still to measure the backfat of live animals. However, new technologies and more knowledge will increase the use of ultrasound equipment to measure other traits on live pigs and will improve the prediction of lean meat yield.


At the CDPQ test station, 324 hogs were probed prior to slaughter at an average weight of 105 kg. The first ultrasound equipment used was a B-mode type. It consisted of an Aloka model 500 with a 126 mm-3.5MHz transducer. Two scans have been taken on the left side of each hog : one cross sectional between the third and fourth last ribs (CROSS) and one longitudinal over the last four or five ribs at 5 cm off the back midline (LONG).

The images were stored on a computer and a VCR for subsequent analysis. The longitudinal images have been analysed for lean meat yield and loin eye area predictions from fat and loin depth measurements with AUSKey AUTOD (LIN, 1994). The cross sectional images have been manually analysed to determine the lean meat yield and loin eye area. The AUSKey AUTOQ has been used to determine the loin marbling from cross sectional and longitudinal images.

The second ultrasound equipment used was An A-mode type. It consisted of a Krautkramer USM2. Three different sites have been probed : behind the last rib, 5 cm off the back midline, average of the fat of both sides of the animal (ANT) ; 3.5 cm in front of the ischion, 5 cm off the back midline, average of the fat of both sides of the animal (POST) ; between the third and fourth last ribs, 5 cm off the back midline, average of the fat of both sides of the animal (FAT34) and average of the loin depths of both sides of the animal (MUSC34). The ANT and POST are the usual fat measures that are currently used in the Canadian genetic evaluation program. They have been pooled together to have only one fat measurement (FAT).

All the animals have been slaughtered and the loin and fat depth measured with a Hennessy probe on the carcass have been recorded and the lean meat yield has been calculated according to the equation currently used in Quebec plants using a Hennessy probe :

Lean meat yield (%) = 67.2327 - 0.7877 fat + 0.1086 loin depth + 0.0087 fat2 - 0.0004 loin depth2 - 0.0002 fat*loin depth

A loin eye trace was done at the plant to calculate the loin eye area and visual marbling score has been determined for each animal on a cut of the loin between the third and fourth last ribs. Scoring of the marbling was done using the Agriculture Canada standard on a 5 point scale with 1 = trace, 2 = slight, 3 = small, 4 = moderate and 5 = abundant.

Forty-eight animals had a full cutout to determine their precise contents in lean, fat, bone and skin. A cutout lean meat yield has been determined using the following equation (Jones et al., 1994A):

% lean yield = (boneless loin + tenderloin + boneless and defated ham + boneless and defated shoulder + skinned belly + side ribs) / weight of the half carcass x 100

The data have been analysed with procedures CORR and GLM from SAS (SAS Institute, 1988). The effects used in the analysis model were sex (gilts and barrows) and crossbred types. Animal weight and prober were not considered since all animals have been probed at about the same weight by the same technician. Probing date was non-significant.


1. Descriptive statistics

Descriptive statistics of the measurements of the different traits are shown in table 1. As expected, Hennessy fat was higher than fat measured on live animals. This has already been reported by other authors such as Sather et al. (1982). Fat measured on the live animal with the longitudinal B-mode was higher than measured with the A-mode or the cross sectional B-mode (non-significant). While loin depth measured on the live animal with the longitudinal B-mode was lower than measured with the A-mode (p<0.05).

An overestimation of fat and underestimation of loin depth was expected with the B-mode equipment compared to the A-mode. Longitudinal measure is the average of the depth over 4 to 5 ribs. Furthermore, the measures taken with ultrasound equipment depend on the velocity of the ultrasound in the measured materials, and calibration of the ultrasound equipment must be made to take into consideration the velocity differences.

Ultrasound velocity in fat is 1 480 m/s and is 1 620 m/s in the muscle. The B-mode equipment used was calibrated to 1 550 m/s. This calibration causes underestimation of the loin depth and overestimation of fat. So if these measures are to be used directly, this fact should be considered and the measurements corrected with a constant factor.

The A-mode equipment used could be manually calibrated to the chosen velocity. We have calibrated it to 1 480 m/s to measure the fat with accuracy. When we have measured the loin depth, calibration was set up to 1 480 m/s also, because when the measure of loin depth is done, it includes fat and muscle depth. The fat depth was subtracted from the total and a correction factor (10 %) was added to give the loin depth. The technique used to measure the loin depth was adapted from previous trials done at CDPQ.

2. Lean meat yield and loin eye area predictions

Prediction of the lean yield from the A-mode, B-mode, slaughterhouse and cutout measures have been compared to each others (table 2). Live measures were more accurate than the Hennessy probe to predict lean yield as obtained from the cutout.

As already reported (Jones et al., 1994b ; Stouffer, 1996), mode B ultrasound equipment usually predicts more accurately the lean yield than the probes used in the slaughterhouses (Hennessy or Destron). The relation between the prediction from the Hennessy probe and the cutout was the same as in the 1992 national cutout (R2=0.64) while A-mode, B-mode (LONG) and B-mode (CROSS) had accuracy of 0.70, 0.65 and 0.55 respectively.

Thus, from this trial, the best measurements to predict lean meat yield were the measures taken with A-mode ultrasound equipment between the 3rd and 4th last ribs.

For the loin eye area, the two machines gave almost the same accuracy at around 0.59 and 0.60.

3. Marbling prediction

Marbling has been predicted from the B-mode cross sectional and longitudinal scans. It has been compared with visual evaluation on the loin of the carcass as previously explained (table 3). The accuracy of prediction is essentially zero.

However, when the distribution of the frequency of the number of observations is plotted with the marbling score, we can note that CROSS and LONG scores follow about the same distributions (figure 1). However, these measures overestimate the number of animals in score 1 and underestimate the number of animals in score 2 as determined by the visual observation at the slaughterhouse.

The marbling cannot be predicted by live scans, but even if the correlations are quite low, it seems possible, with more work, to be able to determine it.


It is now possible to measure with accuracy the loin depth of the live animal. This should permit direct selection of the loin muscle instead of using backfat alone. Loin eye area should be another selection trait because lean yield is no longer the only researched trait by the markets. Specificly loin eye area. distribution of the meat in the different parts of economical interest of the animal (loin, ham, etc.) and meat quality traits such as marbling are more and more in demand.

Selection must be done considering the needs of all the swine industry and consumers needs. A better link is required between all aspects of the production. This is the tomorrow's biggest challenge !

Table 1 - Descriptive statistics of ultrasonic,
carcass and cutout measurements
Measured trait
Number of animals
Standard deviation
Aloka (B-mode ultrasounds)
CROSS fat mm
CROSS loin eye area cm2
LONG fat mm
LONG loin depth mm
LONG loin eye area cm2
Krautkramer (A-mode ultrasounds)
FAT mm
FAT34 mm
MUSC34 mm
Hennessy fat mm
Hennessy muscle mm
Loin eye area cm2

Table 2 - Accuracy (R-square) of the prediction
of the lean meat yield from the different types of measures
Hennessy lean yield (%)
Cutout lean yield (%)
Loin eye area (cm2)
Aloka (B-mode ultrasounds)
Krautkramer (A-mode ultrasounds)
FAT34 and MUSC34
All prediction significant at p<0.05.
n.a. : not available.

Table 3 - Accuracy (R-square) of the prediction
of the marbling from live measurements
Aloka (B-mode ultrasounds)
Prediction of the visual marbling
*Significant at p<0.05.

Figure 1 - Marbling distribution of the different evaluation methods


Jones, S.D.M., A.K.W. Tong, W.M. Robertson and P. Skoczylas. 1994a. National pork carcass cutout project (1992) part 1 : a comparison of the 1978 and 1992 cut outs. Canadian Pork Council, Ottawa, Canada. Pp. 1-20.

Jones, S.D.M., A.K.W. Tong, A.F. Fortin, C. Campbell and W.M. Robertson. 1994b. National pork carcass cutout project (1992) part 3 : Evaluation of new technology for grading pork carcasses. Canadian Pork Council, Ottawa, Canada. Pp. 39-46.

Lin, Y. 1994. Userís Manual for AUSkey system, version 2.0. Animal Ultrasound Services, inc. Ithaca, New York. 151 p.

SAS Institute. 1988. SAS/STAT User's guide. Release 6.03 Edition. Cary, North Carolina. 1028 pp.

Sather, A.P., H.T. Fredeen and A.H. Martin. 1982. Live animal evaluation of two ultrasonic probes as estimators of subcutaneous backfat and carcass composition in pigs. Can. J. Anim. Sci., 62 :943-949.

Sather, A.P., A.K.W. Tong and D.S. Harbison. 1987. A study of ultrasonic probing techniques for swine. II. Prediction of carcass yield from the live pig. Can. J. Anim. Sci., 67 :381-389.

Stouffer, J.R. 1996. Ultrasound grading of carcasses. In New technologies for carcass grade and quality assessment, Proceedings of the 11th Annual Symposium of the Canadian Meat Science Association. February 7, 1996, Toronto, Ontario. Pp. 1-8.

Stouffer, J.R., M.V. Wallentine, G.H. Wellington and A. Diekmann. 1961. Development and application of ultrasonic methods for measuring fat thickness and rib-eye area in cattle and hogs. J. Anim. Sci., 20 :759-767.