Automated Segmentation Of Coronary Arteries

This is the challenge design document for the "Automated Segmentation Of Coronary Arteries" Challenge, accepted for MICCAI 2020.

Cardiovascular disease is a major cause of death. Medical imaging such as Computed Coronary Tomography
Angiography (CCTA) is often used to evaluate the severity of coronary artery disease. Due to the non-invasive
nature of CCTA, it is also commonly used for evaluating and reconstructing heart and coronary vessel structures.
Coronary artery tree models have a wide array of applications in anatomy, physiology and pathophysiology for
educational, training and research purposes; Including study of anatomy of coronary vessels and effects of disease
on anatomy, developing machine learning models for disease risk prediction, education and training of medical
professionals, In-silico testing of medical devices and 3D printing models for testing and education.
Due to the small size of coronary arteries, possible disease and image artefacts, segmentation of coronary arteries
has been focused on semi-automatic methods where a human expert guides the algorithm and corrects errors.
This severely limits large scale processing of medical images and possibility of integration with clinical systems.
Previous challenges have been focused on specific tasks such as centreline extraction, stenosis quantification and
segmentation of specific artery segments. However, to our knowledge this is the first challenge focused on
developing fully automatic segmentation methods of the full coronary artery tree. This challenge aims to establish
a large standardized annotated dataset of healthy and diseased coronary vessels and utilize this dataset to develop
fully automated segmentation algorithms. Automated segmentation would allow processing the large number of
CCTAs available to create larger datasets for use cases mentioned above.