Automatic construction of 2D shape models

A procedure for automated 2D shape model design is presented. The modeling system is given a set of training example shapes defined by the coordinates of their contour points. The shapes are automatically aligned using Procru stes analysis and clustered to obtain cluster prototypes (typical objects) and statistical information about intracluster shape variation. One differe nce from previously reported methods is that the training set is first auto matically clustered and those shapes considered to be outliers are discarde d. In this way, the cluster prototypes are not distorted by outlier shapes. A second difference is in the manner in which registered sets of points ar e extracted from each shape contour. We propose a flexible point matching t echnique that takes into account both pose/scale differences as well as non linear shape differences between a pair of objects. The matching method is independent of the initial relative position/scale of the two objects and d oes not require any manually tuned parameters. Our shape model design metho d was used to learn 11 different shapes from contours that were manually tr aced in MR brain images. The resulting model was then employed to segment s everal MR brain images that were not included in the shape-training set. A quantitative analysis of our shape registration approach, within the main c luster of each structure, demonstrated results that compare very well to th ose achieved by manual registration; achieving an average registration erro r of about 1 pixel. Our approach can serve as a fully automated substitute to the tedious and time-consuming manual 2D shape registration and analysis .