Polyp volume versus linear size measurements at CT colonography: Implications for noninvasive surveillance of unresected colorectal lesions

OBJECTIVE. Proposed surveillance of unresected medium-sized (6.0-9.9 mm) polyps will require reliable detection of small incremental changes. The purpose of this study was to assess the reproducibility of linear versus volume measurements of polyp size at CT colonography (CTC) and to correlate results with a hemispheric model. MATERIALS AND METHODS. The study group consisted of 30 polyps on supine and prone CTC data sets. Measurements were performed separately by two experienced radiologists using the same CTC system (Vitrea 2), resulting in 120 linear and volume measurements. Linear size was defined as the longest dimension among the 2D multiplanar reconstruction views. Semiautomated volume determination required manual tracing of polyp boundaries on 2D multiplanar reconstruction views. The relative error between reviewers 1 and 2 was defined as 100 x (|D₁ - D₂|) / D_ave for linear size (D) and as 100 x (|V₁ - V₂|) / V_ave for volume (V), where ave is average. RESULTS. The mean relative error for linear size and volume measurement was 10.4% ± 10.7% and 16.9% ± 13.2%, respectively. Median linear size and volume of polyps were 9.4 mm and 270 mm³, respectively. CTC-derived volumes for medium-sized polyps closely approximated hemispheric volume (V = (π / 12) x D^b, where b = 3.13, r² = 0.90). Small incremental changes in hemispheric size result in a 3:1 relative change in volume versus diameter, such that a 1-mm diameter change in a medium-sized hemispheric polyp results in a relative change in linear size and volume ranging from approximately 11-18% and 31-53%, respectively. CONCLUSION. Because changes in polyp volume are amplified compared with linear dimension, volume measurement rather than diameter measurement will better allow detection of small incremental changes in polyp size using CTC.