In vitro CT comparisons of dissected phleboliths and retrieved ureteral calculi

Purpose: This project was designed to assess the accuracy of radiologists in distinguishing between pelvic phleboliths and ureteral calculi by CT scanning in vitro. Materials and methods: Twenty-five phleboliths were dissected from autopsy specimens, and 23 ureteral calculi were retrieved intact and all calcifications made available for this study. Calcifications were suspended in ordinary butter and subjected to CT scanning using both bone and soft tissue windows. A panel of three radiologists blinded to the correct diagnoses independently assessed the images. The questions asked were: was the calcification a phlebolith or a stone? was it round or irregular? did it have a central “hole” or not? Confidence levels for each observer and each question were rated on a semicontinuous scale from 0 to 100 and receiver-operating characteristic (ROC) curves computed. Results: ROC curves for distinguishing between ureteral stones and calculi ranged from .78 to .99. As a subset, ROC curves for roundness alone, assuming phleboliths tended to be round and calculi not, ranged from .85 to .98. These ROC curves for roundness alone correlated closely with overall performance in distinguishing ureteral stones from phleboliths. ROC curves for determining central radiolucencies, however, although the latter were assumed to be hypothetically present in phleboliths, were degenerate for two observers, and only .57 for the third. Conclusions: Under optimal in vitro conditions, pelvic phleboliths and ureteral calculi do possess characteristics, especially shape, that many permit discrimination by CT scanning.