Statistical atlases of human anatomy and computer assisted diagnostic system

In this paper, we report our current progress results on computer assisted diagnostic (CAD) system, which consists of three units: database unit (statistical atlas of human anatomy), image processing unit (image enhancement, image segmentation, image registration), and visualization unit (volume rendering). In the database unit, we proposed a new method called generalized N-dimensional principal component analysis (GND-PCA) for statistical volume atlas. Not only the shape variations but also the intensity (CT value) variations can be modeled by the use of GND-PCA. In the processing unit, we proposed a new liver segmentation method from low contrast CT volumes and developed an interactive visualization system for liver and its vascular structure in the visualization unit.