Computer Tomography Software Program 'Osirix' Aids Prediction of Sheep Body Composition

Measurement of changes in amounts of muscle (lean), fat and bone of animals during feeding experiments is essential for the estimation of changes in body and carcass composition, and for prediction of carcass meat yield. Compositional changes can be determined by real-time ultrasound although the method is limited to the measurement of specific parts of the animal. Computer-aided tomography (CT) measurements of live animals can accurately quantify lean, fat and bone. In sheep, live animal lean, fat and bone compositional measurements are difficult to perform because of the large size of digestive tract, its contents, and difficulties in distinguishing these components from lean and fat tissue in CT images. However, techniques for CT image analysis are improving and newer software programs, such as Osirix (Rosset et al. 2004), have improved analytical capabilities that should increase the accuracy of measuring body composition of live animals. Eighty second-cross wether lambs (initial average liveweight 40.1 kg) were scanned live, without sedation using a Picker UltraZ 2000 CT scanner (Philips Medical Imaging Australia, Sydney NSW), at the beginning and end of a 49 day feeding experiment. Liveweights were recorded immediately prior to scanning then each animal had approximately 66 CT 5 mm cross-sectional images taken, starting at the cervical neck region and continuing through the torso and finishing approximately half way between the patella and tarsal region of the hind-leg. To determine live animal lean, fat and bone content, the virtual removal of internal organs and digestive tract was performed by manual editing of each CT image using Osirix. Once the images were modified, the weight of lean, fat and bone was calculated using Autocat, a software program developed by Thompson and Kinghorn (1998). The lambs were slaughtered at completion of the experiment and hot carcass weight was determined. To determine the accuracy of Osirix, aided by Autocat, the correlation between hot carcass weight and total CT-derived carcass weight was calculated. There was a strong relationship between hot carcass weight and CTderived carcass weight (r² = 0.98, Figure 1). The final CT-derived carcass weight included skin (wool was not computed as it is not a reflective x-ray object) which averaged 4.6 kg (s.e.m. 0.08 kg). The relationship between liveweight and CT estimates of liveweight determined from compositional data was assessed at the beginning and end of the experiment and had r² = 0.95 and 0.96, respectively. For this purpose, the weight of head and feet was estimated as 8 % of liveweight, and wool weight was not included. ... The results of this study show that the use of CT scanning to measure lean, fat and bone should be enhanced by a combination of Osirix image editing and Autocat calculations. However, this technique requires greater automation for more routine use in assessment of body composition.