Coronary artery calcium severity grading on non-ECG-gated low-dose chest computed tomography: a multiple-observer study in a nationwide lung cancer screening registry

Abstract

Objectives: To identify the optimal method for evaluation of coronary artery calcium (CAC) severity on non-electrocardiogram-gated low-dose chest computed tomography (LDCT) in a nationwide lung cancer screening registry.

Methods: A total of 256 subjects were retrospectively enrolled from participants of the Korean Lung Cancer Screening (K-LUCAS) project (an LDCT lung cancer screening registry for high-risk individuals). Four board-certified cardiothoracic radiologists independently assessed CAC severity using four different scoring methods (visual assessment, artery-based grading, segment-involvement grading, and segment-based grading) and classified severity for each case using all four methods as none, mild, moderate, or severe. Agreements between the four observers for CAC category classification and between the four different scoring methods for the same observer were assessed by Fleiss kappa statistics. Evaluation time for CAC grading was compared between observers and between grading methods.

Results: Interobserver agreement was moderate for visual assessment (Fleiss kappa 0.451) and substantial for the other three methods (Fleiss kappa 0.673-0.704). Agreement between the four grading methods for the same observer was substantial for three observers (Fleiss kappa 0.610-0.705) and moderate for one (Fleiss kappa 0.578). Mean evaluation time differed significantly between methods (visual assessment, 14.3 ± 11.8 s; artery-based grading, 17.6 ± 22.3 s, segment-involvement grading, 19.2 ± 6.8 s; segment-based grading, 34.2 ± 37.4 s; p < 0.01).

Conclusion: Artery-based grading could be appropriate with substantial interobserver agreement and an acceptable mean evaluation time.

Key points: • CAC severity grading methods on LDCT show moderate to substantial agreements between grading methods and observers. • Artery-based grading could be appropriate with substantial interobserver agreement and a mean evaluation time of 17.6 s. • Visual assessment is disadvantaged by high interobserver variability despite having the shortest evaluation time.