Biliary anatomy on 3D MRCP: Comparison of volume‐rendering and maximum‐intensity‐projection algorithms

Abstract

To compare volume-rendering (VR) and maximum-intensity-projection (MIP) of three-dimensional T2-weighted turbo spin-echo magnetic resonance cholangiopancreatography using a free-breathing navigator-triggered prospective acquisition correction (3D-TSE-PACE-MRCP) to define biliary anatomies. VR and MIP images of 3D-TSE-PACE-MRCP for 102 patients were retrospectively evaluated. Interpretation of cystic duct variation and biliary branching patterns of each image were recorded independently by two radiologists in a blinded fashion. Interpretation confidence on a five-point scale was compared using the Wilcoxon signed-rank test. The McNemar test was used to compare the accuracies of each reformation with the reference standard obtained by consensus interpretation of both the images and source images. The reference standard identified all biliary bifurcations and 95 of 102 cystic duct confluences (93.1%). VR findings agreed with the reference standard findings more often than MIP with regard to cystic duct variation (94 [92.2%] vs. 76 [74.5%], P<0.01) while there was no significant difference for biliary branching patterns (99 [97.1%] vs. 92 [90.2%], P=0.092). The mean confidence score was significantly higher with VR than MIP with regard to both cystic duct variation and biliary branching patterns (3.7 vs. 2.4; P<0.01; 4.1 vs. 3.3; P<0.01). VR reformation of 3D-TSE-PACE-MRCP defines biliary anatomies more accurately than MIP.