Automated aortic calcium scoring on low-dose chest computed tomography

Area of Science:

• Radiology and Medical Imaging
• Cardiovascular Disease Assessment
• Artificial Intelligence in Healthcare

Background:

• Thoracic computed tomography (CT) scans are used for cardiovascular risk assessment.
• Non-ECG-synchronized CT scans limit precise coronary calcification quantification.
• Aortic calcium scoring is a viable alternative due to its lower sensitivity to cardiac motion.

Purpose of the Study:

• To develop and evaluate a computer-aided system for automatic detection and quantification of aortic calcifications.
• To assess the feasibility of using automated aortic calcium scoring in low-dose, non-ECG-synchronized chest CT scans.
• To establish aortic calcification as a reliable indicator of cardiovascular risk.

Main Methods:

• A computer-aided system was developed using a training set of 340 low-dose, non-ECG-synchronized chest CT scans.
• A multiatlas-based segmentation delineated the aorta, followed by 3D connected component labeling to identify calcifications.
• A two-stage classification using k-nearest neighbors analyzed 63 features of candidate calcifications.

Main Results:

• The system achieved 97.9% detection of calcified plaque volume (945 mm3 out of 965 mm3) with minimal false positives (64 mm3 per scan).
• Spearman rank correlation coefficients demonstrated strong agreement: p = 0.960 between the system and the first observer, and p = 0.961 between two observers.
• Total calcium volume score was determined for each subject based on detected aortic calcifications.

Conclusions:

• Automatic aortic calcium scoring is feasible using the developed computer-aided system.
• The system demonstrates good correlation with manual scoring, suggesting its potential for clinical use.
• Automated aortic calcification quantification can serve as an effective cardiovascular risk indicator.