Objectives: To evaluate the diagnostic performance of MR colonography using automated carbon dioxide (CO2) insufflation for colonic distension, with colonoscopy serving as the reference standard.
Methods: Ninety-eight symptomatic patients underwent MR colonography with faecal tagging and automated CO2 insufflation. Three readers (one expert (reader 1), and two less experienced (readers 2 and 3)) evaluated the images for the presence of colorectal lesions. Bowel distension was evaluated on a 4-point scale. Results were verified with colonoscopy and histopathological analysis.
Results: Per-patient sensitivity for lesions ≥10mm was 91.7% (11 of 12) (reader 1), 75.0% (9 of 12) (reader 2), and 75% (9 of 12) (reader 3). Specificity was 96.5% (82 of 85) (reader 1), 97.7% (83 of 85) (reader 2), 95.3% (81 of 85) (reader 3). Per-patient sensitivity for lesions ≥6mm was 85.7% (18 of 21) (reader 1), 57.1% (12 of 21) (reader 2), and 57.1% (12 of 21) (reader 3). Specificity was 86.8% (66 of 76), 98.7% (75 of 76), 90.8% (69 of 76), respectively. Per-patient sensitivity for advanced neoplasia of ≥10mm and ≥6mm was 88.9% (8 of 9) for all readers. Specificity for ≥10mm and ≥6mm was 98.9% (87 of 88) (reader 1), 97.7% (86 of 88) (reader 2), 96.6% (85 of 88) (reader 3). 94.4% of the colon segments were adequate to optimal distended with dual positioning.
Conclusion: MR colonography can accurately detect lesions ≥10mm, and advanced neoplasia ≥6mm. Sufficient distension was achieved using automated CO2 insufflation for colonic distension in MR colonography.
Keywords: Carbon dioxide (CO(2)); Colorectal neoplasm; Insufflation; MRI; Virtual colonoscopy.
Copyright © 2014 Elsevier Ireland Ltd. All rights reserved.