Olstad K, Gangsei LE, Kongsro J. A method for labelling lesions for machine learning and some new observations on osteochondrosis in computed tomographic scans of four pig joints. BMC Vet Res. 2022;18:328.
This study presents a method for labeling osteochondrosis manifesta and cyst-like lesions in CT scans of four pig joints to aid the development of machine learning algorithms for automatic diagnosis. A total of 6,250 lesions were labeled in 201 pigs, with the highest prevalence of lesions found in the stifle joint (100%) and the lowest in the hock joint (64.7%). An inverse relationship was observed between lesion number and size, with pigs having fewer but larger lesions being more likely to develop clinically significant conditions. The findings emphasize the importance of both lesion number and volume in breeding selection efforts.
Comments