Kidney stone classification by speckle x-ray imaging

<i>Objective.</i>Urinary stone-related diseases affect approximately 6% of the global population, with nearly half of the patients experiencing recurrence. The diagnosis and management of the disease depend on the stone type and composition. Yet, current clinical imaging modalities (ultrasound, computed tomography, and radiography) lack the sensitivity and specificity required for accurate classification. Speckle-based dark-field x-ray imaging offers a potential non-invasive method for classifying urinary stones with the required hardware simplicity and robustness for potential<i>in vivo</i>, clinical applicability. However, the influence of diffuser masks and state-of-the-art speckle x-ray image retrievals on classification remains underexplored.<i>Approach.</i>This<i>ex vivo</i>study systematically compared the efficacy of custom diffuser masks and state-of-the-art speckle x-ray retrieval algorithms, using both grid and custom speckle masks, for single-shot dark-field imaging in urinary stone classification at high x-ray energy (80 kVp).<i>Main Results.</i>Among the various types of urinary stones examined in this study, canine ammonium urate showed the most distinct visibility contrast difference, deviating by 32%from the average x-ray transmission value. Overall, the results indicate a potential to differentiate between three main groups of urinary stones based on their attenuation-to-scattering coefficients: ammonium urate, calcium-based stones, and a third group comprising cystine and struvite. Regarding the different diffuser masks, the periodic grid mask is found to be the most suitable candidate for application to urinary stones. The different dark-field visibility reduction retrieval algorithms yielded nearly identical classification trends for the urinary stone samples. We recommend nonetheless, the Fokker-Planck-based approach due to its strong physical basis and favorable image quality characteristics, especially if the noise level can be kept low.<i>Significance.</i>The findings in this study establish a technical foundation for advancing speckle-based dark-field x-ray imaging toward clinical translation for non-invasive urinary stone classification.