Computed tomography colonography: Automated diameter and volume measurement of colonic polyps compared with a manual technique - In vitro study

Objective: To investigate inter- and intraobserver agreement of automated measurement of polyp diameter in vitro. Methods: Two phantoms ("QRM" and "Whiting") containing simulated polyps of known diameter and volume were scanned using 16-detector row computed tomography. Two observers estimated polyp diameter using 3 methods: software calipers ("manual") , freehand boundary identification ("semiautomatic"), and automated software segmentation ("fully automatic"). Results: Intraobserver 95% limits of agreement for diameter were narrowest for the fully automatic method (QRM span: 0.39 mm, 0.48 mm; Whiting span: 0.24 mm, 0 mm). Manual estimates were approximately 10 times wider (QRM span: 3.57 mm, 3.21 mm; Whiting span: 3.2 mm, 2.02 mm). Volume estimates were narrowest for the fully automatic method (span: 24.2 mm³, 24.1 mm³ vs. 97.9 mm³, 102.9 mm ³ for semiautomatic measurement). Interobserver agreement for diameter was narrowest for the fully automatic method (QRM span: 0.12 mm, Whiting span: 0.16 mm), with the manual method approximately 18 times wider (QRM span: 2.87 mm, Whiting span: 2.18 mm). Conclusion: Fully automated measurement of polyp diameter and volume is technically feasible and results in superior inter- and intraobserver agreement.