3D MR angiography of renal arteries: comparison of volume rendering and maximum intensity projection algorithms

• Published in: Radiology Vol. 223; no. 2; p. 509
• Main Authors: Mallouhi, Ammar, Schocke, Michael, Judmaier, Werner, Wolf, Christian, Dessl, Andreas, Czermak, Benedikt V, Waldenberger, Peter, Jaschke, Werner R
• Format: Journal Article
• Language: English
• Published: United States 01-05-2002
• Subjects: Adult; Aged; Algorithms; Angiography, Digital Subtraction; Contrast Media; Female; Humans; Image Processing, Computer-Assisted; Imaging, Three-Dimensional; Linear Models; Magnetic Resonance Angiography; Male; Middle Aged; Renal Artery Obstruction - diagnosis; Retrospective Studies; Sensitivity and Specificity; Statistics, Nonparametric
• ISSN: 0033-8419
• DOI: 10.1148/radiol.2231010858

To compare volume rendering (VR) and maximum intensity projection (MIP) as postprocessing techniques of magnetic resonance (MR) angiography for detection and quantification of renal artery stenosis.
Twenty-seven patients underwent three-dimensional contrast material-enhanced MR angiography of the renal arteries with a 1.5-T imager. For each renal artery, targeted MIP and VR images were reconstructed in oblique coronal and transverse orientations. For each modality, image generation and evaluation were performed interactively by two independent radiologists blinded to angiographic results. In comparison with digital subtraction angiography (DSA) findings, stenosis quantification and detection by using MIP and VR were evaluated with the use of 50% and 70% cutoff points by using linear regression analysis and 2 x 2 tables. Overall image quality and vascular delineation on MIP and VR images were also compared.
All main and accessory renal arteries depicted at DSA were also demonstrated on MIP and VR images. VR performed slightly better than MIP for quantification of stenoses greater than 50% (VR: r(2) = 0.84, P <.001; MIP: r(2) = 0.38, P =.001) and significantly better for severe stenoses (VR: r(2) = 0.83, P <.001; MIP: r(2) = 0.21, P =.1). For detection of stenosis, VR yielded a substantial improvement in positive predictive value (VR: 95% and 90%; MIP: 86% and 68% for stenoses greater than 50% and 70%, respectively). Image quality obtained with VR was not significantly better than that with MIP; however, vascular delineation on VR images was significantly better.
The VR technique of renal MR angiography enabled more accurate detection and quantification of renal artery stenosis than did MIP, with significantly improved vascular delineation.