Key result
CTA-based conduction system annotation achieves high precision within ~2-5 mm error margins.
Why the study?
Does CTA-based anatomical landmark annotation of the cardiac conduction system by trained readers provide millimeter precision and reliability within clinically accepted margins compared to an expert reference standard?
Observational (n=50)
Does CTA-based anatomical landmark annotation of the cardiac conduction system by trained readers provide millimeter precision and reliability within clinically accepted margins compared to an expert reference standard?
Answers are grounded in this paper and cite their sources.
CTA-based annotation of the cardiac conduction system achieves millimeter precision and good-to-excellent reliability within clinically accepted error margins when transferred from an expert to trained readers.
BACKGROUND: CTA-based estimation of the conduction system from anatomical landmarks underpins a visualization platform (CARA) used in structural heart interventions and conduction system pacing. OBJECTIVE: To quantify the precision and reliability of the annotation method and the bias of trained readers relative to an expert reference standard. METHODS: The expert who defined the method and 6 trained readers each annotated 8 landmarks representing the sinus and atrioventricular nodes, His bundle course (Point B), left bundle branch origin (Point C), and course of Bachmann's bundle (B2-B5), in addition to the three aortic leaflet nadirs on 50 CTA studies in 3 sessions. Repeatability and reproducibility were assessed, referenced to clinically accepted error margins (2-5 mm). Intraclass correlations (ICC) were computed on heights above the aortic annular plane. 3D bias relative to the expert was tested for equivalence within half the margin. RESULTS: Repeatability SD was 1.67-3.98 mm and reproducibility SD 2.03-5.60 mm; all mean deviations were within margin, with 88.6-97.4% of repeat reads and 71.4-94.9% of reader means within margin. Inter-observer ICC on annular heights was 0.76-0.82 for the atrioventricular and 0.79-0.98 for the atrial points. Shared bias relative to the expert was 0.14-3.24 mm, within margin for all landmarks and equivalent within half the margin for sinus node, Point B, B2 and B3; at B3 and B2, 95% and 83% of the apparent separation was repeatability noise. CONCLUSION: The method meets its clinically accepted margins with millimeter precision and good-to-excellent reliability, and transfers from expert to trained readers with bias within margin.
No takes yet. Share an insight, caveat, or question.
Tretter et al. (2026) conducted an observational in Cardiac conduction system visualization on cardiac computed tomography angiography (CTA) (n=50). CTA-based anatomical landmark annotation of the cardiac conduction system by 6 trained readers vs. Expert reference standard annotation and clinically accepted error margins (2-5 mm) was evaluated on Repeatability, reproducibility, intraclass correlation (ICC), and 3D bias relative to expert reference standard within clinically accepted error margins (2-5 mm) (Repeatability SD 1.67-3.98 mm; Reproducibility SD 2.03-5.60 mm; Bias 0.14-3.24 mm; ICC 0.76-0.98). CTA-based conduction system landmark annotation showed high precision and reader bias within 2-5 mm error margins.
Synapse has enriched 10 related papers. Consider them for comparative context: