Attention: Restrictions on use of AUA, AUAER, and UCF content in third party applications, including artificial intelligence technologies, such as large language models and generative AI.
You are prohibited from using or uploading content you accessed through this website into external applications, bots, software, or websites, including those using artificial intelligence technologies and infrastructure, including deep learning, machine learning and large language models and generative AI.
JU INSIGHT CT-Based Stone Volume Determination: Development of an Automated Artificial Intelligence Algorithm
By: Andrei D. Cumpanas, MD, University of California, Irvine; Chanon Chantaduly, BS, Center for Artificial Intelligence in Diagnostic Medicine, University of California, Irvine; Kalon L. Morgan, MD, University of California, Irvine; Wei Shao, MS, Center for Artificial Intelligence in Diagnostic Medicine, University of California, Irvine; Antonio R. H. Gorgen, MD, University of California, Irvine; Candices Minh Tran, BS, University of California, Irvine; Yi Xi Wu, BS, PhD, University of California, Irvine; Amanda McCormac, BS, University of California, Irvine; Zachary E. Tano, MD, University of California, Irvine; Roshan M. Patel, MD, University of California, Irvine; Peter Chang, MD, Center for Artificial Intelligence in Diagnostic Medicine, University of California, Irvine; Jaime Landman, MD, University of California, Irvine; Ralph V. Clayman, MD, University of California, Irvine | Posted on: 21 Feb 2024
Cumpanas AD, Chantaduly C, Morgan KL, et al. Efficient and accurate computed tomography–based stone volume determination: development of an automated artificial intelligence algorithm. J Urol. 2024;211(2):256-265.
Study Need and Importance
Renal stones are complex, 3D, irregularly shaped concretions. As such, a 1D or 2D assessment of the stone does not accurately represent the true stone burden with which the urologist is faced. Accordingly, we sought to investigate the diagnostic accuracy and precision of a University of California, Irvine—developed artificial intelligence (AI) algorithm for stone volume determination.
What We Found
The University of California, Irvine AI algorithm was accurate (R = 0.98) and precise (Dice score = 0.90) for determining stone volume. The AI outperformed all 3 of the recommended ellipsoid volume formulas (Figure). Of note, the algorithm’s accuracy and precision improved when measuring larger stones, which trended toward more irregular shapes. Even the “best-fit” ellipsoid formulas overestimated the true stone volume by 27% to 89%; this was most notable for larger stones.
Limitations
While one might argue that multiple radiologists should assess the “ground truth” stone volume, the physician research fellow performing the measurements was validated as a reliable reviewer of stone volume by comparing his 3D stone segmentations with that of a board-certified radiologist, a fellowship-trained endourologist, and an endourology fellow (interclass correlation coefficient = 0.99).
Interpretation for Patient Care
We are starting to incorporate stone burden volumetric assessment into our clinical practice. By employing regular 3D stone volumetric assessment, we can better determine treatment outcomes as they relate to 2 apparently similar 10-mm stones, such as one that is 10 × 3 × 4 mm (120 cc) vs another 10-mm stone measuring 10 × 8 × 8 mm (640 cc); these 2 stones differ 5-fold by volume. Stone volume determination also has the potential to improve medical management of stone disease, as the urologist will be able to determine if a 2-mm change in stone size is due to an actual increase in stone burden rather than a simple change in the position of the stone.
advertisement
advertisement