Efficacy of Automated Supply Artery Tracking Software Using Preoperative Computed Tomography for Renal Carcinoma

Abstract

Purpose: To evaluate the ability of automated supply artery tracking software to detect feeding vessels for renal tumors using preoperative dynamic contrast-enhanced computed tomography.

Material and Methods: For 10 sessions in 10 patients in which transarterial embolization was performed before percutaneous ablation therapy for a single renal cell carcinoma, data that had been obtained from dynamic contrast-enhanced computed tomography in the arterial phase were examined. Automated supply artery tracking software was retrospectively applied with arterial phase images of preoperative contrast-enhanced computed tomography, and the extracted feeding vessels were identified by two observers: a radiologist and a radiological technologist. Real supply arteries were determined by arteriography during transarterial embolization. Extracted feeding vessel and real supply arteries were compared. The concordance rate of extracted feeding vessel between observers was examined. Sensitivity and positive predictive value of automated supply artery tracking software and changes in sensitivity and positive predictive value under conversion of the distance recognized as extracted feeding vessel between the tumor and vessels from the preset distance (20 mm) to the cut-off value using receiver operating characteristic curve analysis were investigated.

Results: Twenty real supply arteries were identified among 10 cases. Number of extracted feeding vessel was 32 and 34 by the observers. The concordance rate of extracted feeding vessel was 80% (8/10 cases). Sensitivity of automated supply artery tracking software was 70% (14/20) by both observers and positive predictive value was 43.8% (14/32) and 41.2% (14/34) by each observer. When the cut-off value (12.1 mm) replaced distance, positive predictive value was elevated from 43.8% to 73.7% and from 41.2% to 68.4%.

Conclusions: Ability of automated supply artery tracking software based on transvenous contrast-enhanced computed tomography was acceptable for identifying feeding vessels of a renal tumor preoperatively.