Volume rendering and connectivity algorithms for MR angiography

Abstract

Several display algorithms for three-dimensional angiographic data are evaluated. The mathematical analysis assumes additive Gaussian noise to predict the background distribution function for maximum intensity projection, sum projection, and connectivity display methods. In the maximum intensity projection method the mean noise level increases with the number of voxels in the ray, while in the sum projection the noise distribution width increases with the projection thickness, but the mean level remains constant. Comparisons of maximum intensity projection, sum projection, and connectivity algorithms applied to an MR angiogram of the circle of Willis are made. Measurements of the noise distribution are in agreement with the analysis. Algorithms combining connectivity with maximum intensity and sum projection are also evaluated. In these methods, a projection image is created using only the voxels marked by connectivity, typically with a 6% threshold of the data. Fine vessels are resolved and background noise is reduced in agreement with the analysis.