MR imaging: influence of imaging technique and postprocessing on measurement of internal carotid artery stenosis

Abstract

Background and purpose: MR angiography (MRA) is increasingly used as an alternative to digital subtraction angiography (DSA) to evaluate internal carotid artery (ICA) stenosis. Because MRA is not standardized in data acquisition and postprocessing, we sought to evaluate the effects of different acquisition techniques (time-of-flight MRA [TOF-MRA]) and contrast-enhanced MRA [CE-MRA]) and postprocessing methods (maximum intensity projection [MIP], multiplanar reformation [MPR], and volume-rendering on stenosis grading.

Materials and methods: Fifty patients (33 men, 17 women) with symptomatic ICA stenosis were examined at 1.5T. Two imaging techniques and 3 postprocessing methods resulted in 6 image datasets per patient. Two readers independently evaluated ICA stenosis according to the North American Symptomatic Carotid Endarterectomy Trial criteria. Interobserver variability was calculated with the Pearson correlation coefficient and simultaneous confidence intervals (CI). The relationship of the values of ICA stenosis between the techniques was assessed by means of simultaneous 95% Tukey CI.

Results: Interobserver agreement was high. Higher concordance was found for postprocessing techniques with TOF- than with CE-MRA; the mean difference between TOF-MPR and TOF-MIP was 0.4% (95% CI, -2.9%-3.8%). Stenosis values for CE-MPR differed significantly from those of CE volume-rendering (7.2%; 95% CI, 3.9%-10.6%).

Conclusion: Stenosis grading was found to be independent of the postprocessing technique except for comparison of CE-MPR with CE volume-rendering, with the volume-rendering technique resulting in higher stenosis values. MPR seems to be best-suited for measurement of ICA stenosis. Parameter setting is critical with volume-rendering, in which stenosis values were consistently higher compared with the other methods.

Fig 1.

Comparison of differences in stenosis values between the observers. All lines indicating the 95% CI cross the zero line. This shows that observers do not differ significantly in the measurement of ICA stenosis. obs indicates observer; VR, volume-rendering.

Fig 2.

Comparison between TOF-MPR and TOF-MIP shows the smallest deviation, and comparison between CE-MPR and CE volume-rendering (VR) shows the highest deviation between ICA stenosis values. Differences are displayed by means of scatterplots (left row) and Bland-Altman plots (right row). CIs of the difference in stenosis values are given as gray bands.

Fig 3.

Comparison of techniques by differences between stenosis values. Pairs of techniques that cross the zero line are not considered significantly different. ICA stenosis values differ significantly between CE-MPR and CE volume-rendering (VR).

Fig 4.

Stenosis of the right ICA. A, An overview with the CE-MIP method. Enlarged pictures with CE-MIP (B) and CE volume-rendering (C) show local signal-intensity loss with distal enhancement, but residual signal intensity is detectable with TOF-MRA (D, TOF-MIP; E, TOF volume-rendering) and CE-MPR (F, G) at the site of minimal lumen (arrow). With TOF-MRA (D, E), signal intensity is fading at the edge of the scan volume, resulting in artificial lumen reduction.