Published studies have shown that methane yield (g CH4/kg dry matter) from sheep is positively correlated with the size (volume and surface area) of the reticulo-rumen (RR) and the weight of its contents. However, the relationship between CH4 yield and RR shape has not been investigated. In this work, shape analysis has been performed on a data set of computerised tomography (CT) scans of the RR from sheep having high and low CH4 yields (n=20 and n=17, respectively). The three-dimensional geometries of the RRs were reconstructed from segmented scan data and split into three anatomical regions. An iterative fitting technique combining radial basis functions and principal component (PC) fitting was used to create a set of consistent landmarks which were then used as variables in a PC analysis to identify shape variation within the data. Significant size differences were detected for regions corresponding to the dorsal and ventral compartments between sheep with high and low CH4 yields. When the analysis was repeated after scaling the geometries to remove the effect of size, there was no significant shape variation correlating with CH4 yield. The results have demonstrated the feasibility of CT-based computational shape determination for studying the morphological characteristics of the RR and indicate that size, but not shape correlates with CH4 yield in sheep.
Keywords: Reticulo-rumen; methane yield; principal component analysis; shape analysis; statistical shape modelling.