In vivo determination of urinary stone composition using dual energy computerized tomography with advanced post-acquisition processing

Purpose We assessed whether dual energy computerized tomography with advanced post-image processing can accurately differentiate urinary calculi composition in vivo. Materials and Methods A total of 25 patients scheduled to undergo ureteroscopic/percutaneous nephrolithotomy were prospectively identified. Dual energy computerized tomography was performed using 64-slice multidetector computerized tomography. Novel post-processing (DECT _Slope) used pixel by pixel analyses to generate data sets grayscale encoding ratios of relative differences in attenuation of low (DECT _{80 kVp}) and high energy (DECT_{140 kVp}) series. Surgical extraction and Fourier spectroscopy resulted in 82 calculi. Of these stones 51 showed minor admixtures (uric acid, ammonium urate, struvite, calcium oxalate monohydrate and brushite) and 31 were polycrystalline (mixtures of calcium oxalate monohydrate/dihydrate and calcium phosphate). Analyses identified stone clusters of equal composition and distinct attenuation descriptors on DECT _{140 kVp}, DECT_{80 kVp} and DECT_Slope. Iterative cross-validation of the 3 dual energy computerized tomography data sets was used to identify characteristic attenuation limits for each stone type. Results Attenuatio profiles showed substantial overlap among various stones on DECT _{140 kVp} (uric acid 427.3 ± 168.1 HU, ammonium urate 429.9 ± 99.7 HU, struvite 480.2 ± 123.5 HU, calcium oxalate monohydrate 852.4 ± 301.4 HU, brushite 863.7 ± 180.1 HU and polycrystalline 858.1 ± 210.5 HU) and on DECT_{80 kVp} (uric acid 493.6 ± 182.8 HU, ammonium urate 591.5 ± 157.9 HU, struvite 712.4 ± 173.9 HU, calcium oxalate monohydrate 1,240.5 ± 494.7 HU, brushite 1,532.1 ± 273.1 HU and polycrystalline 1,358.7 ± 316.8 HU). Statistically spectral separation was not sufficient to characterize stones unambiguously based on DECT_{140 kVp}/DECT_{80 kVp} attenuation. Analysis of attenuation showed sufficient spectral separation on DECT_Slope (uric acid 14.9 ± 10.9 U, ammonium urate 56.1 ± 1.8 U, struvite 42.7 ± 1.4 U, calcium oxalate monohydrate 62.8 ± 1.8 U and brushite 113.2 ± 5.3 U). Polycrystalline stones (51.8 ± 3.7 U) overlapped with struvite and ammonium urate stones. This overlap was resolved as all struvite/ammonium urate stones measured 900 HU or less and all polycrystalline stones measured more than 900 HU on DECT_{80 kVp}. Conclusions Dual energy computerized tomography with novel post-processing allows accurate discrimination among main subtypes of urinary calculi in vivo and, thus, may have implications in determining the optimum clinical treatment of urinary calculi from a noninvasive, preoperative radiological assessment.