Assessing Change in Stone Burden on Baseline and Follow-Up CT: Radiologist and Radiomics Evaluations.

Area of Science:

• Medical imaging
• Artificial intelligence in medicine
• Nephrology

Background:

• Kidney stone burden assessment on follow-up CT is crucial for patient management.
• Radiologist qualitative assessments and radiomics have limitations in accurately evaluating stone burden changes.
• An automated artificial intelligence (AI)-based volumetric approach offers a potential improvement.

Purpose of the Study:

• To compare the diagnostic accuracy of an AI-based volumetric approach against radiologist assessments and radiomics for evaluating kidney stone burden changes on follow-up CT.
• To determine if automated volumetric quantification is superior to traditional methods.

Main Methods:

• Retrospective study including 157 patients with baseline and follow-up non-contrast abdomen-pelvis CT scans.
• Index test: AI-based prototype (Frontier, Siemens Healthineers) for automated kidney and stone segmentation, providing volumetric data.
• Reference standard: Threshold-defined volumetric assessment of stone burden change.
• Radiologist performance assessed via clinical reports and independent review.

Main Results:

• AI automated volumetric assessment identified stable, increased, or decreased stone burden.
• Qualitative radiologist assessments showed weak diagnostic performance (AUC 0.55-0.72).
• AI threshold-based volumetric quantification demonstrated higher diagnostic accuracy than qualitative assessments and radiomics.

Conclusions:

• Automated threshold-based volumetric quantification of kidney stone burdens is more accurate than qualitative radiologist assessments and radiomics.
• AI-driven volumetric analysis provides a reliable method for evaluating kidney stone burden changes on follow-up CT.
• This AI approach enhances diagnostic accuracy in kidney stone monitoring.