Pure uric acid stone prediction model using the variant coefficient of stone density measured by thresholding 3D...

Area of Science:

• Urology
• Radiology
• Medical Imaging Analysis

Background:

• Urinary stones are prevalent worldwide, with uric acid stones being treatable via oral chemolysis.
• Current management often involves invasive procedures, highlighting the need for non-surgical alternatives.

Purpose of the Study:

• To develop and validate a prediction model for pure uric acid (pUA) stones.
• To utilize a variant coefficient of stone density (VCSD) derived from noncontrast CT images.

Main Methods:

• A 3D segmentation-based method with a 130 HU threshold was used to analyze CT images.
• VCSD was calculated as the stone heterogeneity index divided by mean stone density.
• The study included training, internal validation, and external validation datasets.

Main Results:

• The prediction model identified VCSD as the most powerful independent predictor of pUA stones.
• The model demonstrated high discrimination, with AUCs of 0.960, 0.931, and 0.938 in the respective validation sets.
• Key predictors included VCSD, age, sex, radio-opacity, hypertension, diabetes, and urine pH.

Conclusions:

• VCSD derived from 3D segmentation is a significant predictor for identifying pure uric acid stones.
• The developed prediction model offers a noninvasive preoperative tool for pUA stone identification, potentially guiding treatment decisions.