Quantification of lumen stenoses with known dimensions by conventional angiography and computed tomography: implications of using conventional angiography as gold standard

Background Quantitative coronary angiography (QCA) has inherent limitations for displaying complex vascular anatomy, yet it remains the gold standard for stenosis quantification. Objective To investigate the accuracy of stenosis assessment by multi-detector computed tomography (MDCT) and QCA compared to known dimensions. Methods Nineteen acrylic coronary vessel phantoms with precisely drilled stenoses of mild (25%), moderate (50%) and severe (75%) grade were studied with 64-slice MDCT and digital flat panel angiography. Fifty-seven stenoses of circular and non-circular shape were imaged with simulated cardiac motion (60 bpm). Image acquisition was optimised for both imaging modalities, and stenoses were quantified by blinded expert readers using electronic callipers (for MDCT) or lumen contour detection software (for QCA). Results Average difference between true and measured per cent diameter stenosis for QCA was similar compared to MDCT: 7 (±6)% vs 7 (±5)% (p=0.78). While QCA performed better than MDCT in stenoses with circular lumen (mean error 4 (±3)% vs 7 (±6)%, p<0.01), MDCT was superior to QCA for evaluating stenoses with non-circular geometry (mean error 10 (±7)% vs 7 (±5)%, p<0.05). In such lesions, QCA underestimated the true diameter stenosis by >20% in 9 of 27 (33%) vs 1 of 29 (3%) in lumen with circular geometry. Conclusions QCA often underestimates diameter stenoses in lumen with non-circular geometry. Compared to QCA, MDCT yields mildly greater measurement errors in perfectly circular lumen but performs better in non-circular lesions. These findings have implications for using QCA as the gold standard for stenosis quantification by MDCT.