Coronary Artery Calcium Score Predicts Major Adverse Cardiovascular Events in Stable Chest Pain.

Background Coronary artery calcium (CAC) has prognostic value for major adverse cardiovascular events (MACE) in asymptomatic individuals, whereas its role in symptomatic patients is less clear. Purpose To assess the prognostic value of CAC scoring for MACE in participants with stable chest pain initially referred for invasive coronary angiography (ICA). Materials and Methods This prespecified subgroup analysis from the Diagnostic Imaging Strategies for Patients With Stable Chest Pain and Intermediate Risk of Coronary Artery Disease (DISCHARGE) trial, conducted between October 2015 and April 2019 across 26 centers in 16 countries, focused on adult patients with stable chest pain referred for ICA. Participants were randomly assigned to undergo either ICA or coronary CT. CAC scores from noncontrast CT scans were categorized into low, intermediate, and high groups based on scores of 0, 1-399, and 400 or higher, respectively. The end point of the study was the occurrence of MACE (myocardial infarction, stroke, and cardiovascular death) over a median 3.5-year follow-up, analyzed using Cox proportional hazard regression tests. Results The study involved 1749 participants (mean age, 60 years ± 10 [SD]; 992 female). The prevalence of obstructive coronary artery disease (CAD) at CT angiography rose from 4.1% (95% CI: 2.8, 5.8) in the CAC score 0 group to 76.1% (95% CI: 70.3, 81.2) in the CAC score 400 or higher group. Revascularization rates increased from 1.7% to 46.2% across the same groups (P < .001). The CAC score 0 group had a lower MACE risk (0.5%; HR, 0.08 [95% CI: 0.02, 0.30]; P < .001), as did the 1-399 CAC score group (1.9%; HR, 0.27 [95% CI: 0.13, 0.59]; P = .001), compared with the 400 or higher CAC score group (6.8%). No significant difference in MACE between sexes was observed (P = .68). Conclusion In participants with stable chest pain initially referred for ICA, a CAC score of 0 showed very low risk of MACE, and higher CAC scores showed increasing risk of obstructive CAD, revascularization, and MACE at follow-up. Clinical trial registration no. NCT02400229 © RSNA, 2024 Supplemental material is available for this article. See also the editorial by Hanneman and Gulsin in this issue.

Active multitask learning with uncertainty-weighted loss for coronary calcium scoring.

PURPOSE: The coronary artery calcification (CAC) score is an independent marker for the risk of cardiovascular events. Automatic methods for quantifying CAC could reduce workload and assist radiologists in clinical decision-making. However, large annotated datasets are needed for training to achieve very good model performance, which is an expensive process and requires expert knowledge. The number of training data required can be reduced in an active learning scenario, which requires only the most informative samples to be labeled. Multitask learning techniques can improve model performance by joint learning of multiple related tasks and extraction of shared informative features. METHODS: We propose an uncertainty-weighted multitask learning model for coronary calcium scoring in electrocardiogram-gated (ECG-gated), noncontrast-enhanced cardiac calcium scoring CT. The model was trained to solve the two tasks of coronary artery region segmentation (weak labels) and coronary artery calcification segmentation (strong labels) simultaneously in an active learning scenario to improve model performance and reduce the number of samples needed for training. We compared our model with a single-task U-Net and a sequential-task model as well as other state-of-the-art methods. The model was evaluated on 1275 individual patients in three different datasets (DISCHARGE, CADMAN, orCaScore), and the relationship between model performance and various influencing factors (image noise, metal artifacts, motion artifacts, image quality) was analyzed. RESULTS: Joint learning of multiclass coronary artery region segmentation and binary coronary calcium segmentation improved calcium scoring performance. Since shared information can be learned from both tasks for complementary purposes, the model reached optimal performance with only 12% of the training data and one-third of the labeling time in an active learning scenario. We identified image noise as one of the most important factors influencing model performance along with anatomical abnormalities and metal artifacts. CONCLUSIONS: Our multitask learning approach with uncertainty-weighted loss improves calcium scoring performance by joint learning of shared features and reduces labeling costs when trained in an active learning scenario.