QUANTIFICATION AND VISUALIZATION OF LOCALIZED AND INTUITIVE SHAPE VARIABILITY USING A NOVEL MEDIAL-BASED SHAPE REPRESENTATION

摘要

Quantification and visualization of anatomical shape variability in different populations is essential for diagnosis and tracking progression of diseases. We present a new 3D medial-based shape representation method capable of analysis and visualization of 3D anatomy and demonstrate its ability to quantify and highlight shape variability in an intuitive manner. 3D shapes are represented via orientations and elongations of one or more medial sheets, along with thickness values encoding the distances to the shape surface. Two parameters traverse each medial sheet and are mapped to orientation, elongation, and thickness values; we call this map a medial patch. Shape variability is decomposed intuitively into bend, stretch, or bulge deformations, via operators acting on the components of the medial patch. In a simple manner, the location, extent, type, and amplitude of the deformation operators can be specified to capture local and global intuitive shape variability. We demonstrate the capabilities and intuitiveness of this approach through synthetic 3D shape deformations, as well as deformations that capture the 3D shape of an anatomical structure. We demonstrate the ability to highlight regions containing specific types of intuitive changes in anatomy.