Cardiovascular diseases are the global leading cause of death. The amount of calcified plaque deposited in the wall of the coronary arteries has been shown to be a strong and independent predictor of cardiovascular events. In clinical routine, CT based quantification of coronary artery calcium (calcium scoring) is performed and reported with increasing frequency. In current practice, coronary calcifications are manually identified, which is prohibitive in large scale clinical trials and impractical in daily practice. A number of algorithms describing automatic detection and quantification of calcified lesions in coronary arteries have been proposed. However, these algorithms have not been evaluated for robustness in large and diverse sets of images.The purpose of this challenge is to compare methods for automatic coronary artery calcium scoring in cardiac CT examinations originating from different CT scanners on a whole-heart and a per-vessel basis. This evaluation framework was launched at a MICCAI 2014 workshop in Boston, USA, where we organized the Challenge on Automatic Coronary Calcium Scoring.
The task in this challenge is to automatically identify coronary calcifications in non-contrast enhanced ECG-triggered cardiac CT examinations that are acquired as part of standard clinical routine. For this, both non-contrast and contrast-enhanced CT images are provided. Results are assessed on a lesion-level in sensitivity and positive predictive value. Additionally, teams can choose to identify the coronary artery in which each calcification is located. In that case, the performance of the algorithm is also evaluated per artery. Finally, the total amount of coronary calcium per patient and the assignment of patients to cardiovascular risk categories based on this amount is assessed.