Colonic polyps: effect of attenuation of tagged fluid and viewing window on conspicuity and measurement--in vitro experiment with porcine colonic specimen.

Purpose: To investigate effect of attenuation of tagged fluid and viewing window on polyp conspicuity and measurement with porcine colonic specimen.
Materials and Methods: Eleven (3-10-mm-diameter) polyps were created in porcine colon and the specimen submerged in saline. Four-detector row CT was performed after gas distension and after filling with six barium sulfate suspensions (attenuation, 100-1000 HU). Two readers independently measured maximal two-dimensional polyp diameter on each data set with the following four viewing windows and window levels and window widths, respectively: colon (-150 HU, 1500 HU), lung (-500 HU, 1500 HU), bone (500 HU, 2500 HU), and abdomen (40 HU, 400 HU). In consensus, polyp conspicuity (compared with air data set) was assigned a grade of 1-4 for each viewing window (grade 1, not seen or barely visible; grade 4, optimally seen). For statistical analysis, conspicuity grades were collapsed to a two-point scale. Data were analyzed with Mann-Whitney, Kruskal-Wallis, and chi2 tests.
Results: Accuracy of polyp measurement was independent of viewing window for attenuation of tagged fluid of 100-300 HU but differed significantly for 500-1000 HU (P < .001); that for colonic and bone viewing windows was superior (median size difference, 1.0 mm; interquartile range, 0.5-1.5). Conspicuity differed significantly according to viewing window at all attenuation values (P < .001). For 100-300 HU with abdominal viewing window, 83% (24 of 29) of observations were assigned grade 3 or 4 (best). For 500-1000 HU with bone viewing window, 94% (30 of 32) of observations were assigned grade 3 or 4 (superior). Overall conspicuity was best with bone viewing windows at 700 HU.
Conclusion: Polyp conspicuity and measurement in tagged data sets were optimized at 700 HU with bone viewing windows. At less than 300 HU, conspicuity improved with abdominal viewing windows.
(RSNA, 2006)