Identification of Small Hepatocellular Carcinoma and Tumor-feeding Branches with Cone-beam CT Guidance Technology during Transcatheter Arterial Chemoembolization

Abstract

PurposeTo evaluate the performance of transcatheter arterial chemoembolization guidance software that uses cone-beam computed tomography (CT) technology in identifying small hepatocellular carcinoma (HCC) tumors and feeding branches. Materials and MethodsCone-beam CT and manual feeder vessel detection (MFD) software were used in chemoembolization of 68 HCCs 30 mm or smaller (mean±standard deviation, 15.3 mm±5.2). Detectability of tumors and tumor-feeding sub-subsegmental arteries was compared versus that of nonselective digital subtraction angiography (DSA). Technical success of chemoembolization was divided into three grades according to 1-week CT findings: entire tumor embolized with at least a 5-mm-wide margin (ie, complete), tumor embolized without a margin in parts (ie, adequate), or entire tumor not embolized (ie, incomplete). All cone-beam CT data were also reanalyzed with automatic feeder vessel detection (AFD) software that was developed later. ResultsCone-beam CT could depict all tumors, including eight that were first discovered during chemoembolization. Sixty-one tumors (89.7%) were detected on CT during arterial portography and during hepatic arteriography, and seven (10.3%) were detected with one or the other. Nonselective DSA depicted 49 tumors (72.1%). Among 100 tumor-feeding vessels, 81 were identified with MFD and 38 with nonselective DSA. Detectability of tumors with CT and tumor-feeding branches with MFD was significantly better than with nonselective DSA (both P<.001). Fifty-nine tumors (86.8%) were completely embolized and nine (13.2%) were adequately embolized. AFD identified 96 feeder vessels; 88 (88%) represented true-positive findings. ConclusionsTranscatheter arterial chemoembolization guidance software with cone-beam CT technology has a sufficient performance level to detect small HCCs and their feeding branches.