Loss of urocystolith architectural clarity during in vivo radiographic simulation versus in vitro visualization

Urocystoliths of 9 mineral types from 434 canine patients submitted to the University of Minnesota Urolith Bank were imaged in a urinary bladder phantom. Imaging techniques simulated were survey radiography and double contrast cystography. Morphologic characteristics visually observed in vitro or by interpretation of high-resolution specimen radiographs were compared to those seen using the simulated in vivo imaging techniques. Shape characteristics that were accurately detected ≥25% of the time on simulated survey or double contrast radiography were faceted, irregular, jackstone, ovoid, and round. Surface characteristics that were accurately detected ≥25% of the time on simulated survey or double contrast radiography were rough, smooth, and smooth with blunt tips. Internal architecture characteristics that were accurately detected ≥25% of the time on simulated survey or double contrast radiography were lucent center, random-nonuniform, and uniform. Shapes such as bosselated, facetedovoid, and rosette; surfaces such as botryoidal, and knife-edged; and internal architecture characteristics such as dense center, dense shell, laminated, and fissures were of almost no value either due to poor detectability or poor accuracy of recognition. Based on optimized simulated survey and double contrast radiographic procedures, it appears that a number of shape, surface, and internal architecture characteristics may be of limited or no value in discriminating among urocystolith mineral types under clinical circumstances. Shapes and surfaces were more accurately characterized by the simulated double contrast technique, but for internal architecture, the simulated survey radiographic technique seemed slightly superior overall.