Detection and characterization of intracranial aneurysms with MR angiography: comparison of volume-rendering and maximum-intensity-projection algorithms.

OBJECTIVE. The purpose of this study was to compare volume rendering and maximum intensity projection as postprocessing techniques of MR angiography in the detection and characterization of intracranial aneurysms. MATERIALS AND METHODS. Three-dimensional time-of-flight MR angiography studies performed in 82 patients were retrospectively evaluated by two independent reviewers who were unaware of digital subtraction angiography findings, the standard of reference. Panoramic maximum-intensity-projection and volume-rendered angiograms were produced from each data set to investigate the presence of underlying aneurysms. Each detected aneurysm was then interactively evaluated with subvolume maximum-intensity-projection and targeted volume-rendering algorithms to evaluate aneurysm morphology and size. Aneurysm detection and characterization were evaluated by means of the receiver operating characteristic analysis, and aneurysm size was evaluated using the limits-of-agreement method. Image quality, aneurysm neck depiction, and vascular delineation were also compared between maximum-intensity-projection and volume-rendered images. The time required for the generation and interpretation of maximurn-intensity-projection and volume-rendered images was assessed. RESULTS. Volume rendering tended to improve the diagnostic confidence (A(Z) [area under the receiver operating characteristic curve] = 0.95 vs A(Z) = 0.90 for maximum intensity projection) and yielded a considerable improvement in sensitivity (89% vs 71% for maximum intensity projection), particularly in the detection of small cerebral aneurysms. Regarding aneurysm morphology, volume rendering performed significantly better than maximum intensity projection in lobulation detection (p < 0.001) and slightly better in neck categorization (p > 0.238). Limits-of-agreement analysis showed a trend toward improved assessment of the aneurysm size by volume rendering (-0.31 +/- 1.62 mm vs -1.27 +/- 2.84 nim by maximum intensity projection). Overall image quality and vascular delineation of involved vessels on volume-rendered images were rated better than that obtained by maximum intensity projections (p less than or equal to 0.007 and p less than or equal to 0.001, respectively). Evaluation of time-of-flight MR angiography data sets was significantly facilitated with volume rendering (p < 0.001). CONCLUSION. The volume-rendering technique facilitates the evaluation of cerebral time-of-flight MR angiography data sets and allows better detection and more reliable characterization of intracranial aneurysms than does maximum intensity projection.