Introduction

Abstract

Since the discovery of X-rays in 1895, many different imaging techniques have been developed which gain visual access to the interior of a closed body without opening it. Nowadays, these techniques are widely used in health-care and biomedical research and constitute a substantial part of the clinical practice. In order to facilitate the interpretation of the generated body images, a multitude of medical image analysing methods has been realized which support the physicians in the fields of diagnostics, surgical planning and image guided surgery as well as medical research. With the progress of image acquisition techniques, the modeling of anatomical structures in 3D or even 4D has become an important component in medical image computing as these models offer an additional perspective for the surgeons and are used for model-based analysis, segmentation and classification problems. A popular approach for shape modeling is constituted by statistical methods which aim to represent an organ by statistical shape models. As opposed to a single 3D model or an atlas of an organ which are only (typical) shape examples, a statistical shape model represents a set containing segmented organs by a mean shape and a variability model. Hence, statistical shape models incorporate a priori shape knowledge drawn from many organ examples. Especially for segmentation problems, the application of statistical shape models has been proven to be very successful for a wide range of anatomical structures in CT, MR and ultrasound images.