Automated Characterization of Pyelocalyceal Anatomy Using CT Urograms to Aid in Management of Kidney Stones

Abstract

Nephrolithiasis is a costly and prevalent disease that is associated with significant morbidity including pain, infection, and kidney injury. While surgical treatment of kidney stones is generally based on the size and quality of the stones, studies have suggested that specific characteristics of the pyelocalyceal anatomy (i.e. urinary drainage system), such as the infundibulopelvic angle (IPA), can influence the success rate of various treatment modalities. However, the traditional methods of quantifying such anatomic features have typically relied on manual measurements using 2-dimensional (2D) images of a 3-dimensional (3D) system, which can be cumbersome and potentially inaccurate. In this paper, we propose a novel algorithm that automatically identifies and isolates the 3D volume and central frame of the urinary drainage system from computerized tomography (CT) Urograms, which then allows for 3D characterization of the pyelocalyceal anatomy. First, the kidney and pyelocalyceal system were segmented from adjacent soft tissues using an automated algorithm. A centerline tree structure was then generated from the segmented pyelocalyceal anatomy. Finally, the IPA was measured using the derived reconstructions and tree structure. 8 of 11 pyelocalyceal systems were successfully segmented and used to measure the IPA, suggesting that it is technically feasible to use our algorithm to automatically segment the pyelocalyceal anatomy from target images and determine its 3D central frame for anatomic characterization. To the best of our knowledge, this is the first method that allows for an automated characterization of the isolated 3D pyelocalyceal structure from CT images.