AI algorithm boasts accuracy in stone volume determination

An automated artificial intelligence (AI) algorithm demonstrates accuracy and efficiency in determining stone volume using computed tomography (CT) scans, as shown in a study.

Researchers obtained 322 noncontrast CT scans from patients diagnosed with urolithiasis and designated the largest stone in each scan as the “index stone.” A validated reviewer determined the 3D volume of the index stone using 3D Slicer technology, and this was deemed the “ground truth” volume.

The AI-calculated index stone volume was then compared with the ground truth volume and with the estimated volumes using scalene, prolate, and oblate ellipsoid formulas.

A near-perfect association was noted between the AI-determined volume and the ground truth (R, 0.98). The AI algorithm showed efficiency in determining the stone volume for all sizes, and its accuracy further improved with larger stone size. The AI stone volume also generated an excellent 3D pixel overlap with the ground truth (Dice score, 0.90).

On the other hand, the ellipsoid formula-based volumes (R range, 0.79‒0.82) did not perform as precisely as the AI algorithm. Notably, the accuracy of the ellipsoid formulas decreased as the stone size increased (mean overestimation, 27 percent to 89 percent).

In addition, the maximum linear stone measurement for all stone sizes exhibited the poorest association with the ground truth (R range, 0.41‒0.82).

“The University of California, Irvine AI algorithm is an accurate, precise, and time-efficient tool for determining stone volume,” the researchers said.

“Expanding the clinical availability of this program could enable urologists to establish better guidelines for both the metabolic and surgical management of their urolithiasis patients,” they added.