This processing device acquires anatomical features of a region of interest of a target object from images of a target object imaged in two different positions, generates, on the basis of the anatomical features, normalization transformation information for transforming the shape of the region of interest of the target object in the two different positions into a shape that will constitute a prescribed reference, and generates, on the basis of the normalization transformation information, a statistical transformation model indicating the statistical characteristics of the transformation between the images of the target object imaged in the two positions.