Deep learning for vessel-specific coronary artery calcium scoring: validation on a multi-centre dataset.

AIMS: To present and validate a fully automated, deep learning (DL)-based branch-wise coronary artery calcium (CAC) scoring algorithm on a multi-centre dataset. METHODS AND RESULTS: We retrospectively included 1171 patients referred for a CAC computed tomography examination. Total CAC scores for each case were manually evaluated by a human reader. Next, each dataset was fully automatically evaluated by the DL-based software solution with output of the total CAC score and sub-scores per coronary artery (CA) branch [right coronary artery (RCA), left main (LM), left anterior descending (LAD), and circumflex (CX)]. Three readers independently manually scored the CAC for all CA branches for 300 cases from a single centre and formed the consensus using a majority vote rule, serving as the reference standard. Established CAC cut-offs for the total Agatston score were used for risk group assignments. The performance of the algorithm was evaluated using metrics for risk class assignment based on total Agatston score, and unweighted Cohen's Kappa for branch label assignment. The DL-based software solution yielded a class accuracy of 93% (1085/1171) with a sensitivity, specificity, and accuracy of detecting non-zero coronary calcium being 97%, 93%, and 95%. The overall accuracy of the algorithm for branch label classification was 94% (LM: 89%, LAD: 91%, CX: 93%, RCA: 100%) with a Cohen's kappa of k = 0.91. CONCLUSION: Our results demonstrate that fully automated total and vessel-specific CAC scoring is feasible using a DL-based algorithm. There was a high agreement with the manually assessed total CAC from a multi-centre dataset and the vessel-specific scoring demonstrated consistent and reproducible results.

Ischemic burden and clinical outcome: is one 'culprit' ischemic segment by dobutamine stress magnetic resonance predictive?

AIMS: We sought to evaluate the impact of ischemic burden for the prediction of hard cardiac events (cardiac death or nonfatal myocardial infarction) in patients with known or suspected CAD who undergo dobutamine stress cardiac magnetic resonance imaging (DCMR). METHODS: We included 3166 patients (pts.), mean age 63 ± 12 years, 27% female, who underwent DCMR in 3 tertiary cardiac centres (University Hospital Heildelberg, German Heart Institute and Kings College London). Pts. were separated in groups based on the number of ischemic segments by wall motion abnormalities (WMA) as follows: 1. no ischemic segment, 2. one ischemic segment, 3. two ischemic segments and 4. ≥ three ischemic segments. Cardiac death and nonfatal myocardial infarction were registered as hard cardiac events. Pts. with an "early" revascularization procedure (in the first three months after DCMR) were not included in the final survival analysis. RESULTS: Pts. were followed for a median of 3.1 years (iqr 2-4.5 years). 187 (5.9%) pts. experienced hard cardiac events. 2349 (74.2%) had no inducible ischemia, 189 (6%) had ischemia in 1 segment, 292 (9.2%) in 2 segments and 336 (10.6%) ≥ 3 segments. Patients with only 1 ischemic segment showed a high rate of hard cardiac events of ∼ 6% annually, which was 10-fold higher compared to those without ischemia (0.6% annually, p < 0.001) but similar to those with 2 and ≥ 3 ischemic segments (∼ 5.5% and ∼ 7%, p = NS). CONCLUSIONS: The presence of inducible ischemia even in a single 'culprit' myocardial segment during DCMR is enough to predict hard cardiac events in patients with known or suspected CAD.