Development of a NUCTECH prototype DEXA scanner for determination of lamb carcase composition

Summary

This pilot study evaluated the ability of a prototype NUCTECH DEXA system to differentiate fat, lean muscle and bone tissue using tissue calibration blocks of known composition. Additionally, this study assessed the ability of this DEXA prototype to predict the CT fat, lean muscle and bone % of lamb carcases.
The NUCTECH engineering team successfully developed a prototype DEXA system following a series of iterative hardware and software refinements. The prototype DEXA scanner was developed around a conveyor to facilitate the scanning of tissue calibration blocks, while a steel frame was custom built to allow entire lamb carcases to be scanned through the prototype DEXA device. A medical-grade CT scanner was sourced at a Beijing hospital to scan lamb carcases and assess the ability of the prototype DEXA to predict carcase composition (Milestones 1 & 2).
In Experiment 1, tissue calibration blocks were constructed from dissected lamb fat, lean and bone tissue, in defined mixtures and thicknesses. The calibration blocks were transported to Beijing for scanning with the prototype DEXA scanner. The prototype DEXA system was able to identify the thickness, chemical fat %, and bone % of the scanned tissue blocks, showing good capacity to differentiate between carcase tissue types. From these scans, a series of tables were constructed enabling the matching of DEXA pixel R values to tissue types and thicknesses. The NUCTECH engineering team reviewed the development of these tables and the subsequent equations used to predict lamb carcase composition from DEXA images with researchers at Murdoch University (Milestone 3).  
In Experiment 2, seven lamb carcases locally sourced in Beijing were scanned using both the NUCTECH prototype DEXA system, and the medical CT scanner. A thresholding method was applied to images separating soft-tissue pixels from bone-containing pixels, before the equations derived in Experiment 1 were applied to the images. The NUCTECH DEXA system was able to determine CT lean% and bone% with good precision, however the prediction of CT fat % was imprecise, suggesting that further development is needed in this area.
The next phase of work will be to install a commercial prototype DEXA that can operate at chain speed in an Australian abattoir. This will enable calibration against a larger population of lambs slaughtered under standard Australian conditions that reflect industry genotypes (Milestone 4).