Background and purpose: Cystoscopy and ureteroscopy have limitations in the evaluation for urothelial tumors, and both are invasive. We studied the utility of three-dimensional (3D) CT virtual endoscopy in phantom models.
Materials and methods: A phantom pelvis was constructed of Plexiglas, porcine pelvic bones, and processed animal fat and scanned at various table speeds in a four detector-row CT machine for ability to detect "tumors" of Solidwater plastic polymer. Images were reconstructed at slice thicknesses of 2.5 to 5.0 mm and reconstructed in 3D for evaluation by two radiologists with no knowledge of the scanning parameters or tumor location. Similar studies were performed with a ureter model.
Results: With 5-mm slices, the sensitivity for bladder tumors ranged from 67% for 2-mm tumors to 100% for 4-mm tumors, with 12 false-positive findings. The overall sensitivity was 86% with 3.75-mm slices with one false positive, and with 2.5-mm slices, the sensitivity was 93%, again with one false positive. For the ureteral tumors, the overall sensitivities and numbers of false positives were 88.9% and eight with 5.0-mm collimation, 88.9% and four with 3.75-mm collimation, and 100% and three with 2.5-mm collimation. The effective radiation dose for all studies was equivalent to that of a standard abdomen/pelvis scan.
Conclusions: Although virtual endoscopy traditionally has had difficulty detecting tumors <5 mm, the multidetector-row CT protocols used in this study could detect most lesions smaller than this. The scan also depicts the other tissues of the pelvis, which is valuable for staging. The 3D images were produced using data from the CT urogram parameters standard at our institution.