Accuracy of semiautomated analysis of 3D contrast-enhanced magnetic resonance angiography for detection and quantification of aortoiliac stenoses

We sought to compare reproducibility and accuracy of semiautomated stenosis detection and quantification in 3D contrast-enhanced magnetic resonance angiography (CE-MRA) images with conventional evaluation of 3D CE-MRA in patients with peripheral arterial disease (PAD) by using intra-arterial digital subtraction angiography (IA-DSA) as standard of reference. Twenty-five patients with PAD underwent CE-MRA and IA-DSA. Three blinded observers independently evaluated CE-MRA datasets for stenoses in the aortoiliac region using semiautomated computer analysis. Semiautomated measurements were compared with conventional measurements of stenosis on CE-MRA datasets, as measured by 3 other independent observers blinded to all other measurements. Interobserver agreement was quantified using kappa (kappa) and intraclass correlation coefficients (ICCs). Sensitivity and specificity were determined for both semiautomated and conventional measurements. Semiautomated measurements were successful in 124 of the 125 arterial segments. The sensitivity of semiautomated measurements was 89% for all observers; specificity varied between 87% and 89%. For conventional measurements of CE-MRA, sensitivity varied between 79% and 86%; specificity was between 86% and 96%. There was good interobserver agreement between all readers for semiautomated measurements (combined kappa for all 3 observers together = 0.78; ICC = 0.82), as well as for conventional measurements (combined kappa = 0.70; and ICC = 0.83). Differences between ICCs, combined kappa values, and accuracy of both measurements were not significant (all P > 0.05). Semiautomated analysis of aortoiliac 3D CE-MRA has the same high accuracy for detection and quantification of stenoses as conventional readings of CE-MRA