Fully automatic segmentation of coronary vessel structures in poor quality X-ray angiogram images

Kose C.

STRUCTURAL, SYNTACTIC, AND STATISTICAL PATTERN RECOGNITION, PROCEEDINGS, vol.4109, pp.74-82, 2006 (Peer-Reviewed Journal) identifier

  • Publication Type: Article / Article
  • Volume: 4109
  • Publication Date: 2006
  • Journal Indexes: Science Citation Index Expanded, Scopus, Compendex, EMBASE, MathSciNet, Philosopher's Index, zbMATH
  • Page Numbers: pp.74-82


In this paper a fully automatic method is presented for extracting blood vessel structures in poor quality coronary angiograms. The method extracts blood vessels by exploiting the spatial coherence in the image. Accurate sampling of a blood vessel requires a background elimination technique. A circular sampling technique is employed to exploit the coherence. This circular sampling technique is also applied to determine the distribution of intersection lengths between the circles and blood vessels at various threshold depths. After this sampling process, disconnected parts to the centered object are eliminated, and then the distribution of the intersection length is examined to make the decision about whether the point is on the blood vessel. To produce the final segmented image, mis-segmented noisy parts and discontinuous parts are eliminated by using angle couples and circular filtering techniques. The performance of the method is examined on various poor quality X-ray angiogram images.