Surface modeling of blood vessels based on 3D medical images

This paper presents a method for constructing a 3D surface model of the cerebral vasculature from 3D medical images using active contours. The representation of the active contour utilizes Generalized Cylinders. In order to ensure that the active contour reaches a global minimum, dynamic programming is used for energy minimization. The method has been tested on both simulated data and on a computed tomography angiography (CTA) volume. Furthermore, the method has been tested on a magnetic resonance angiography (MRA) and 10 magnetic resonance image (MRI) volume in order to demonstrate its ability to perform on different imaging modalities. The tests show that the method has a mean radius and localization error less than 0.1 mm. The method performs well on all three imaging modalities using only slightly varying parameter settings.