Evaluating Region of Interest Measurement Strategies to Characterize Upper Urinary Tract Stones on Computerized Tomography.

Purpose: Computerized tomography imaging is regularly used to assess stone HU values as a surrogate for stone composition and fragility. Techniques for measuring HU values are unstandardized, leading to high variability. We investigated several region of interest measurement strategies to quantify this variability. Materials and Methods: Patients from an institutional database who underwent preoperative computerized tomography, surgical stone extraction and stone composition analysis were identified. HU measurements were made of each patient stone using transverse/coronal slices in the abdominal/bone windows with 4 region of interest techniques, including 1) the maximum diameter region of interest, 2) the maximum diameter region of interest at all stone inclusive slices, 3) 2 equal-sized, nonoverlapping circular regions of interest and 4) 3 to 5 smaller nonoverlapping regions of interest randomly placed on the stone. Stones that were 80% or greater pure by composition were separately analyzed. Results: A total of 172 patients were included in study. Mean +/- SD stone size was 19.3 +/- 15.6 mm. On subtype analysis 51 stones were calcium oxalate monohydrate, 9 were calcium oxalate dihydrate, 7 were calcium phosphate hydroxyapatite/ brushite and 16 were uric acid. Mean HU values in the abdominal window for all stones identified by region of interest techniques 1 to 4 were 457 +/- 253, 351 +/- 210, 581 +/- 363 and 587 +/- 329, respectively. The distribution of means significantly differed across region of interest techniques, planes and windows when considering all stones together (p < 0.0001), stones with greater than 80% calcium oxalate dihydrate (p = 0.0113) and greater than 80% calcium oxalate monohydrate (p < 0.0001), and uric acid stones (p < 0.0001). Conclusions: HU values obtained to assess stone density vary depending on window, plane and region of interest technique. We recommend that clinicians select a single region of interest measurement technique and use it consistently to minimize interinstitutional variability.