Accuracy of a CBCT-based virtual injection software for vessel detection during hepatic arterial embolization

Area of Science:

• Medical imaging
• Interventional radiology
• Vascular imaging

Background:

• Nonselective cone-beam CT (CBCT) arteriography provides an overview of liver vasculature.
• Selective CT angiography (CTA) offers detailed visualization but requires precise catheter placement.
• Virtual injection (VI) software aims to simulate selective injections from nonselective CBCT data.

Purpose of the Study:

• To evaluate the accuracy, sensitivity, positive predictive value (PPV), and interobserver agreement of VI software.
• To assess the software's ability to simulate selective arterial injections from nonselective CBCT arteriography.
• To determine the reliability of VI for analyzing liver vasculature.

Main Methods:

• Retrospective analysis of 20 patients with both nonselective CBCT and selective CTA.
• Simulation of VI from the exact microcatheter tip position on CBCT.
• Comparison of VI with actual selective CTA by three interventional radiologists using a 6-point scale.
• Calculation of sensitivity, PPV, and Fleiss' kappa for interobserver agreement.

Main Results:

• VI accuracy was high: 96.8% for 1st-3rd order arteries and 83.4% for 4th-5th order arteries.
• Substantial interobserver agreement was observed (Fleiss' kappa = 0.79).
• Average sensitivity was 90.6% and average PPV was 92.7%, with a false negative rate of 9.4%.

Conclusions:

• CBCT-based VI software accurately simulates distal liver injections.
• The software demonstrates high sensitivity and substantial interobserver agreement.
• VI is a reliable tool for enhancing the analysis of liver vasculature in interventional radiology.