Ultrasound Image Segmentation using a Point Distribution Model in a Bayesian Framework

An empirical Bayes framework is used to incorporate point distribution models (PDMs) within Bayesian image analysis, thus synthesising two approaches to the tting of stochastic templates. Manually-segmented images are used to identify both a PDM and a likelihood function, leading to a posterior distribution from which inferences can be drawn. The methodology is explored and illustrated using 144 ultrasound images of sheep. Estimates of sheep fat and muscle depths are shown to be comparable in accuracy with manual interpretation of images.