Accuracy of digital angiography for quantitation of normal coronary luminal segments in excised, perfused hearts.

The accuracy of coronary artery diameter determination by digital angiography was assessed by imaging 39 coronary segments of excised dog and human hearts and comparing these segments with pathologic sections 0.76 to 3.31 mm in luminal diameter. Digital images were obtained postmortem both during contrast injection using a fixed-pressure coronary perfusion system and after fixation when the coronary size was maintained by injection of a barium-gelatin casting mixture. Digital data were analyzed with commercially available, automated edge-detection software using a coronary catheter as the calibration standard. Coronary diameter measured during contrast injection was not significantly different from that measured after casting and fixation. Digital data from both methods correlated well with diameters from pathologic sections (injected, r = 0.85; fixed, r = 0.91). Linear regression parameters comparing pathologic diameters with the contrast injection method were slope = 0.82, intercept = 0.42 mm, and standard error of the estimate = 0.27 mm. Parameters for the comparison of pathology with casted coronary data were slope = 0.95, intercept = 0.16 mm, and standard error of the estimate = 0.23 mm. Intra- and interobserver variability were 3% (0.05 mm) and 4% (0.07 mm), respectively. These data indicate that when a coronary catheter as a calibration standard is used, coronary artery dimensions can be accurately measured by automated digital angiography techniques.