Interobserver reliability in the assessment of coronary stenoses by multidetector computed tomography.

OBJECTIVE: To determine the interobserver reliability for grading coronary stenosis severity with coronary computed tomographic angiography (CCTA). METHODS: Five readers independently reviewed 40 CCTA studies, first the axial images alone, then in combination with multiplanar reconstructions. The stenosis severity in each segment was scored on a 5-point scale. Intraclass correlation (ICC) analysis was performed to assess interobserver reliability on a segmental basis. RESULTS: The reliability was good to moderate in the right coronary artery, left main artery, left anterior descending artery and branches, and the proximal circumflex (ICC: 0.44-0.75) but fair to poor for the posterior descending artery, the posterolateral branch, the obtuse marginal branches, and the distal circumflex (ICC: 0.15-0.39). The ICC correlated with the reference diameter. Although there was no significant difference in the ICC between the scanner types, there were more unevaluable segments in the 16-row scanner compared with the 64-row scanner (2.4 vs 1.4 segments/patient). Addition of multiplanar reconstruction to axial images led to fewer uninterpretable segments and reclassification of stenosis grade in 23% of the segments. CONCLUSIONS: Interobserver reliability for stenosis severity by CCTA varies between segments and correlates with the reference diameter.