Objectives: The aim of this study was to investigate the feasibility of kidney stone composition analysis using spectral detector computed tomography scanner (SDCT) with normal- and low-dose imaging protocols.
Methods: A total of 154 stones harvested from nephrolithotripsy or nephrolithotomy with a known monocrystalline composition as determined by infrared spectroscopy were examined in a nonanthropomorphic phantom on an SDCT (IQon, Philips, Best, the Netherlands). Imaging was performed with 120 kVp and (a) 40 mAs and (b) 200 mAs, resulting in a computed tomography dose index (CTDIvol) of 2 and 10 mGy, respectively. Besides conventional CT images (CIs), SDCT enables reconstruction of virtual monoenergetic images (40-200 keV). Spectral coefficient images were calculated by performing a voxel-by-voxel combination of 40 and 200 keV images (Matlab R2017b, Mathworks Inc). All stones were semiautomatically 3D-segmented on CI using a threshold-based algorithm implemented in an offline DICOM viewer. Statistical assessment was performed using Steel-Dwass method to adjust for multiple comparisons.
Results: Ca-phosphate (n = 22), Ca-oxalate (n = 82), cysteine (n = 20), struvite (n = 3), uric acid (n = 18), and xanthine stones (n = 9) were included in the analysis. Stone diameter ranged from 3.0 to 13.5 mm. On CI, attenuation differed significantly between calcific and noncalcific stones only (P ≤ 0.05), the spectral coefficient differed significantly between (//): Ca-oxalate//Ca-phosphate//cystine//struvite//uric acid//xanthine in 10 mGy protocol (all P ≤ 0.05). The same results were found for the 2 mGy-protocol, except that differentiation of Ca-oxalate and Ca-phosphate as well as uric acid and xanthine was not possible (P ≥ 0.05).
Conclusions: Spectral detector CT allows for differentiation of kidney stones using semi-automatic segmentation and advanced image post-processing, even in low-dose imaging protocols.