Texture Analysis using Random Field Models exemplified on Ultrasonic Images of the Liver

With the aid of a medical example the practical use of models for image analysis will be demonstrated. For that purpose a random field model, the so-called autoregressive periodic random field model, for texture analysis is briefly described. The use of a generative model offers the analysis by synthesis approach. By fitting the model to a given texture pattern the estimated model parameters are suitable textural features. Its successful application to the quantification of the echo pattern of ultrasonic images of the liver can be shown.