Diagnostic accuracy of translucency rendering to differentiate polyps from pseudopolyps at 3D endoluminal CT colonography: a feasibility study
by A. Guerrisi, D. Marin, A. Laghi et al.
| Full Text | |
Abstract
Purpose . The aim of this study was to assess the accuracy of translucency rendering (TR) in computed tomographic (CT) colonography without cathartic preparation using primary 3D reading.
Materials and methods
. From 350 patients with 482 endoscopically verified polyps, 50 pathologically proven polyps
and 50 pseudopolyps were retrospectively examined. For faecal tagging, all patients ingested 140 ml
of orally administered iodinated contrast agent (diatrizoate meglumine and diatrizoate sodium) at
meals 48 h prior to CT colonography examination and two h prior to scanning. CT colonography was
performed using a 64-section CT scanner. Colonoscopy with segmental unblinding was performed within
2 weeks after CT. Three independent radiologists retrospectively evaluated TRCT clonographic images
using a dedicated software package (V3D-Colon System). To enable size-dependent statistical
analysis, lesions were stratified into the following size categories: small (≤5 mm), intermediate
(6–9 mm), and large
(≥10 mm).
Results
. Overall average TR sensitivity for polyp characterisation was 96.6%, and overall average
specificity for pseudopolyp characterisation was 91.3%. Overall average diagnostic accuracy (area
under the curve) of TR for characterising colonic lesions was 0.97.
