查看文章

Medical diagnosis can be supported by segmenting structures of interest inside a large data volume. The Corkscrew Segmentation Algorithm (CSA) is known to work well on elongated structures [1]. Additionally it is a simple and easy to extend algorithm, thus providing a useful basis for further optimizations. Although it was designed for segmenting vessels, it has not been evaluated yet with the Rotterdam Coronary Artery Algorithm Evaluation Framework1 [2]. This is due to the fact that coronary arteries are indeed elongated structures but additionally characterized by bifurcations that the CSA can not handle. This paper discusses an extension to the CSA to find these bifurcations. First evaluations of the extended CSA (eCSA) showed reasonable results on segmenting the coronary tree using data sets of the Coronary Artery Algorithm Evaluation.