Login to Access Video or Poster Abstract: MP01-07
Sources of Funding: General Electric

Introduction

The current gold standard for imaging evaluation of urinary stones is non-contrast CT (NCCT), although in recent years the concerns regarding repeated radiation doses has called into question the need for this modality in all cases. For this reason, plain abdominal radiography (KUB) is used by many urologists in a diagnostic and follow-up setting. Digital tomosynthesis (DT) is a novel imaging technique that produces a number of coronal images from a single tomographic sweep, creating high quality images with less radiation than low-dose NCCT. Our aim was to evaluate the added utility of DT to KUB for the identification of urinary stones.

Methods

Seven fresh cadavers with an intact genitourinary system and no history of nephrolithiasis were implanted with kidney and ureteral stones of known size and composition using endoscopic methods or a small ureterotomy. After stone implantation was completed, the cadavers were imaged with KUB and DT. Three blinded readers (2 urologists and 1 radiologist with experience in evaluating KUB/KUB-DT for stone disease) evaluated all sets of radiographs. Readers initially evaluated only KUB for the presence and location of calculi and recorded their responses. Readers then were instructed to add in tomogram images to their evaluation and re-record the presence and location of calculi to assess the possible value added by tomograms without changing their initial responses based on KUB only. Reference standard was established by consensus reading with a board-certified urologist and board-certified radiologist with 7 years of experience, neither of which served as a blinded reader on this study. Accuracy of stone detection and assessment of stone location was performed using an exact and nearest neighbor match to account for potential movement of stones after implantation as well as perception differences between readers as to nomenclature of stone location.

Results

A total of 59 stones were identified in the seven cadavers as part of the gold standard interpretation. Using KUB only with an exact and nearest neighbor match, Reader 1 accurately identified 45.7% (27/59 stones), Reader 2 identified 47.4% (28/59) stones, and Reader 3 identified 35.6% (21/59) stones. Using KUB-DT with an exact and nearest neighbor match, there was a statistically significant improvement in accuracy of stone detection (p <0.01 for all readers) as Reader 1 accurately identified 72.9% (43/59 stones), Reader 2 identified 62.7% (37/59) stones, and Reader 3 identified 66.1% (39/59) stones. Overall this was calculated as a 57% relative increase in stone detection. Of note, the number of false positives (suspected calculi based on reader assessment that were not present on gold standard reading) on KUB and KUB-DT was similar across both reading sessions (11 and 16 respectively, for all readers combined).

Conclusions

Addition of digital tomosynthesis to KUB leads to significant improvement in the detection of urinary calculi when compared to KUB alone without a concomitant significant increase in false positives. Further studies will determine the true cost and radiation savings associated with the use of this technology, but it appears to be a promising imaging modality for urinary stones and a possible alternative to NCCT in some settings.

Funding

General Electric