This paper presents a methodology for shape analysis of anatomical structures. A template shape is used as a unit, and a shape transformation mapping the template to a particular structure is used to quantify the shape of the structure with respect to the template. The geometric characteristics of the boundaries of the template and the individual structure are used to define point correspondences between the template and the structure. These correspondences are then used to determine the shape transformation, which is based on an elastic adaptation of the template. Regional inter-subject and inter-population shape differences are then identified by comparing the corresponding shape transformations point-wise. Validation using simulation experiments is also presented.
ASJC Scopus subject areas
- Signal Processing
- Computer Vision and Pattern Recognition