3D MRA Image Segmentation Using Capillary Active Contour

Precise segmentation of three-dimensional (3D) magnetic resonance angiography (MRA) images can be a very useful computer aided diagnosis (CAD) tool for clinical routines. In this project, we present a segmentation scheme for accurately extracting vasculature from MRA images. Our proposed algorithm models capillary action and derives a capillary active contour for segmentation of thin vessels. The algorithm is implemented using the level set method and has been applied successfully on real 3D MRA images.

Challenges

• The vasculature has complex structure.
• The vessels are thin tube structures.
• Noise

Sample slices of 3D MRA image.

In order to deal with the thin vessels and extract the complex vasculature, we propose a novel algorithm called Capillary Active Contour (CAC). In our approach, front propagation is driven by capillary action to get the boundary of vessels, which is modeled as an energy minimization process. The algorithm is implemented using level set.

Capillary Action

With capillary action, liquid climbs up capillary tube without external pressure.
Cohesion tends to minimize the area of free surface.
Adhesion tends to pull the surface to the boundaries.

Results

(Click the image to download video)

(Click the image to download video)

Associated publications:

• MRA image segmentation with capillary active contour, (download PDF)
  Int. Conf. Medical Image Computing and Computer Assisted Intervention (MICCAI),
  vol. 1, pp. 51-58, Palm Springs, CA, USA, October, 2005.
  (Received MICCAI 2005 Student Award)
• MRA image segmentation with capillary active contours,
  Medical Image Analysis, to appear, 2006.