Background: The aim of this study was to investigate reproducibility and accuracy of computer-assisted coronary plaque measurements by multislice computed tomography coronary angiography (QMSCT-CA).
Methods and results: Forty-eight patients undergoing MSCT-CA and coronary arteriography for symptomatic coronary artery disease and quantitative intravascular ultrasound (IVUS, QCU) were examined. Two investigators performed the QMSCT-CA twice and a third investigator performed the QCU, all blinded for each other's results. There was no difference found for the matched region of interest (ROI) lengths (QCU 29.4 +/- 13 mm vs. QMSCT-CA 29.6 +/- 13 mm, P = 0.6; total length = 1,400 mm). The comparison of volumetric measurements showed (lumen QCU 267 +/- 139 mm(3) vs. mean QMSCT-CA 177 +/- 91 mm(3), P < 0.001; vessel 454 +/- 194 mm(3) vs. 398 +/- 187 mm(3), P <<0.001; and plaque 189 +/- 93 mm(3) vs. 222 +/- 121 mm(3); investigator 1, P = 0.02; and investigator 2, P = 0.07) significant differences. Automated lumen detection was also applied for QMSCT-CA (218 +/- 112 mm(3), P < 0.001 vs. QCU). The interinvestigator variability measurements for QMSCT-CA showed no significant differences.
Conclusion: QMSCT-CA systematically underestimates absolute coronary lumen- and vessel dimensions when compared with QCU. However, repeated measurements of coronary plaque by QMSCT-CA showed no statistically significant differences, although, the outcome showed a scattered result. Automated lumen detection for QMSCT-CA showed improved results when compared with those of human investigators.
(c) 2007 Wiley-Liss, Inc.