Dual-energy vs conventional computed tomography in determining stone composition.

ABSTRACT

OBJECTIVE: To compare the accuracy between conventional computed tomography (CT) and dual-energy CT (DECT) in predicting stone composition in a blinded, prospective fashion. METHODS: A total of 32 renal stones with known composition were scanned in vitro, first using standard CT techniques at 120 kilovolt peak (kV[p]) and then using fast-switched kilovolt DECT at 80 and 140 kilovolt peak (kV[p]). For the DECT scan, a spectral curve was created demonstrating the change of Hounsfield units (HU) across the kiloelectron volt spectrum. The composition of each stone was estimated by comparing each sample curve with curves of known materials. To attempt stone determination using single-energy CT, the HU of each stone was compared with ranges reported in previous studies. The accuracy of each method was compared. RESULTS: Included were 27 stones large enough to allow analysis. Single-energy measurements accurately identified 14 of 27 stones of all composition (52%), whereas the DECT spectral curves correctly identified 20 (74%). When analyzed by stone type, single-energy vs DECT correctly identified 12 vs 12 of the 12 uric acid stones, 2 vs 3 of the 6 struvite stones, 0 vs 3 of the 5 cystine stones, and 0 vs 2 of the 4 calcium oxalate stones, respectively. When simply attempting to differentiate uric acid vs nonuric acid stones, single-energy CT could accurately differentiate only 6 of 15 stones as nonuric acid (40%) compared with 14 of 15 stones (93%) for DECT. CONCLUSION: DECT appears to be superior to conventional CT in differentiating stone composition and is particularly accurate in differentiating nonuric acid from uric acid stones.