Morphological analysis of pancreatic cystic masses

The aim of this study was to analyze the morphology of pancreatic cystic masses detected on multi-detector row computed tomography (MDCT) to determine whether single-dimension measurements of these masses are accurate reflections of their volumes.

Twenty-five pancreatic cystic masses detected on MDCT in 25 patients were evaluated. Pancreatic cysts were segmented on MDCT using commercially available software. All measurements were obtained twice by two independent investigators, and the means of values for segmented cyst volume (Vs) (milliliters), maximum transaxial diameter (millimeters), and elongation value (defined as 1 - [width/length], where 1 = ellipsoid and 0 = spherical) were reported for each cystic lesion. The volume of each cyst was also calculated (Vc) using the maximum transaxial diameter, with the hypothesis that the cyst was spherical. Student's t test was used to analyze the differences between values of Vs and Vc. Bland-Altman plots and Lin's concordance correlation were used to assess agreement between different measurement techniques. A P value < .05 denoted statistical significance. Interobserver variability was also determined using the Bland-Altman method.

There was a significant difference between Vs and Vc (P < .0001). The elongation values ranged from 0.28 to 0.83 (mean, 0.57 +/- 0.12; median, 0.56). Mean interobserver variability was 1.7% (95% confidence interval, -4.89% to 8.30%).

The results suggest that pancreatic cystic masses are not spherical. Therefore, a cyst's single largest transaxial dimension is not an accurate surrogate of its actual volume.