Combining front propagation with shape knowledge for accurate curvilinear modeling

摘要

A method of constructing a curvilinear model of a tubular structure from a 3­dimensional data set, the method comprising the steps of: (a) forming a substantially minimal cost path for the tubular structure by the steps of: (al) constructing a distance field; (a2) utilising the distance field to construct a propagation map; (a3) utilising the propagation map to construct the substantially minimal cost path; (b) for each point on the substantially minimal cost path, providing the steps of: (bl) sampling the 3­dimensional data set in a substantially locally perpendicular axis to the minimal cost path; (b2) constructing a deformable tube around the substantially minimal cost path determined by alterations in pixel values around the minimal cost path; (b3) extracting an axial measure of the deformable tube as the curvilinear model of the tubular structure. The method can further include the step (b4) of transforming the curvilinear model back into the original data set.