Abstract 6109: Comparison of Quantitative Coronary Lumen Assessment by Multi-detector Computed Tomography with Conventional Coronary Angiography Using Semi-Automatic Contour Detection Algorithms: The Core64 Study
Background: Multi-detector computed tomography angiography (MDCTA) and conventional coronary angiography (CCA) are used clinically to assess coronary artery stenoses in patients with coronary artery disease (CAD). While MDCTA is a tomographic method, CCA is based on projection imaging and a detailed comparison of their relative ability to measure coronary luminal dimensions has not been established.
Methods: We compared 2531 measurements of maximum lumen diameters obtained by 64-slice CT and quantitative coronary angiography (QCA) in 39 patients who were selected from the CORE-64 Multi-Center Trial data file based on image quality and presence of atherosclerosis. Of 40 vessels analyzed, 30 had a least one luminal stenosis of 50% or greater. Using semi-automated contour detection algorithms, MDCTA (VITAL Lesion Tool®) and QCA (CAAS II®) luminal diameters were obtained at 500 micron increments along the axial length of the segment by independent core laboratories. Images were evaluated in end diastolic phases. Cross-sectional lumen contours were initially traced by the software but frequently corrected manually for MDCTA to match visually inspected lumen borders. MDCTA and QCA maximum luminal diameters generated by the respective algorithms were compared at each axial position for agreement. Fiduciary landmarks were used to assure matching measurement sites within the artery after MDCTA and QCA vessels length were corrected for foreshortening. Statistical analysis was performed using Bland-Altman comparative analysis.
Results: Results are shown in table⇓.
Conclusions: When compared to QCA, MDCTA underestimates coronary luminal dimensions in arterial segments ≥3.0 mm and overestimates lumen diameter in coronary segments smaller than 2.0 mm. MDCTA shows good agreement with QCA in arterial segments of lumen size between 2–3 mm. The discrepancies observed may be explained by the inherent differences in image acquisition between MDCT and QCA.