Objective: We sought to compare various reconstruction methods for CT angiographic images in evaluating living renal donors.
Materials and methods: In 76 patients who underwent donor nephrectomy, vascular phase CT data were obtained using an MDCT scanner (detector array, 1.25 mm x 4; beam pitch, 1.5). Two radiologists independently reconstructed CT angiographic images using thick-slab volume rendering, thick-slab maximum intensity projection (MIP), sliding thin-slab volume rendering, and sliding thin-slab MIP. The radiologists counted the number of renal arteries, early branching arteries, and renal veins. We compared the accuracy rates for the detection of vessels achieved with the four types of reconstructed images, using the surgical findings as the gold standard. Agreement between the two observers and between the surgical and CT angiographic findings was evaluated.
Results: The sensitivity for detecting the supernumerary artery was significantly greater with sliding thin-slab volume rendering and sliding thin-slab MIP (97%) than with thick-slab volume rendering (59%) (p = 0.039). No significant difference between the other comparison pairs of reconstruction methods was found. The interobserver agreement for detecting supernumerary and early branching arteries with sliding thin-slab volume rendering and MIP was excellent (kappa = 0.820-0.859) and good for renal veins (kappa = 0.698-0.724), whereas the agreement of thick-slab volume rendering and MIP was good for arteries (kappa = 0.706-0.791) and moderate for veins (kappa = 0.443-0.579). The agreement between CT angiographic reconstructed images and surgical findings for detection of vessels was better with sliding thin-slab volume rendering and MIP (kappa = 0.793-1.000) than in thick-slab volume rendering and MIP (kappa = 0.306-0.613).
Conclusion: For CT angiographic evaluation of living renal donors, sliding thin-slab reconstruction is superior to thick-slab reconstruction.