Three-Dimensional Endoluminal CT Colonography (Virtual Colonoscopy): Comparison of Three Commercially Available Systems

OBJECTIVE: The purposes of this study were to directly compare 3D endoluminal volume rendering and navigational capabilities of three different CT colonography systems and to assess feasibility of 3D evaluation for primary polyp detection. MATERIALS AND METHODS: Closely matched endoluminal images from three CT colonography software systems (Navigator, Vitrea 2, and V3D-Colon) and optical colonoscopy were obtained of eight pathologically proven colorectal polyps. All images were then reviewed by 25 physicians (12 radiologists and 13 gastroenterologists) who were not familiar with the three systems. For each polyp, the images yielded by the three systems were rated according to polyp conspicuity, 3D effect (depth), subjective quality, and likeness to optical colonoscopy. For comparison of endoluminal navigation capability, automated or semiautomated flight from rectum to cecum and cecum to rectum was attempted in 10 cases (20 flights) in which a continuous colonic air column could be identified on the 2D images. Additional 3D features were also compared. RESULTS: For polyp conspicuity, 3D effect, and likeness to optical colonoscopy, the V3D-Colon system was favored in 92.0% (184/200), 92.5% (185/200), and 92.5% (185/200) of responses, respectively (p < 0.001). For the same categories, the Navigator system ranked second in 73.0%, 74.0%, and 75.0% of cases, and the Vitrea 2 system ranked last in 79.0%, 77.5%, and 76.0% of cases, respectively. Automated or semiautomated navigation was successful in eight (40%) of 20 flights with Vitrea 2, in nine (45%) of 20 flights with Navigator, and in 20 (100%) of 20 flights with V3D-Colon (p < 0.001). The V3D-Colon system also had more navigational features than the other two systems. CONCLUSION: Pronounced subjective and objective differences in 3D endoluminal rendering and navigational capabilities exist among the systems evaluated. Of the three, effective time-efficient primary 3D evaluation appears to be feasible only with the V3D-Colon system.