A Modular Workflow Architecture for Coronary Centerline Extraction in Computed Tomography Angiography Data

• Published in: Computer Vision and Graphics pp. 155 - 162
• Main Authors: Correa-Agudelo, Esteban, Flórez-Valencia, Leonardo, Orkisz, Maciej, Mouton, Claire, Dávila Serrano, Eduardo E., Hernández Hoyos, Marcela
• Format: Book Chapter
• Language: English
• Published: Cham Springer International Publishing
• Series: Lecture Notes in Computer Science
• Subjects: centerline extraction; computed tomography; coronary arteries; minimum cost-path algorithms; vesselness
• DOI: 10.1007/978-3-319-11331-9_19
• ISBN: 3319113305; 9783319113302
• ISSN: 0302-9743

Efficient and reliable extraction of coronary artery centerline from computed tomography angiography data is a prerequisite for a variety of medical imaging applications. Many authors have combined minimum-cost path algorithms and vesselness measures to extract coronary centerlines. We propose a modular decomposition of this extraction process, in order to facilitate the implementation and comparison of different minimum-cost path strategies allowing users (radiologists and developers) to focus on subsequent image analysis tasks. Evaluation results show a good overlap (> 84%), and small distances with regard to reference centerlines (on average, not larger than the voxel size) in multi-vendor datasets, for two combinations of algorithms that follow this framework. Therefore, it can serve as a starting point to subsequent image analysis stages that require coronary centerlines.