Maxime Ronot MD, Presenter: Nothing to Disclose

Mohamed Abdel-Rehim MD, Abstract Co-Author: Nothing to Disclose

Viseth Kuoch MD, Abstract Co-Author: Nothing to Disclose

Antoine Hakime MD, Abstract Co-Author: Nothing to Disclose

Marion Roux, Abstract Co-Author: Nothing to Disclose

Melanie Chiaradia MD, Abstract Co-Author: Nothing to Disclose

Valerie Vilgrain MD, Abstract Co-Author: Nothing to Disclose

Thierry J. De Baere MD, Abstract Co-Author: Consultant, Terumo Corporation Speaker, Covidien AG Speaker, Terumo Corporation Speaker, General Electric Company Consultant, General Electric Company Consultant, Guerbet SA Speaker, Guerbet SA

Frederic Deschamps, Abstract Co-Author: Nothing to Disclose

To compare the ability of a dedicated software and conventional cone beam computed tomography (CBCT) analysis to identify tumoral feeders in a series of malignant liver tumors treated with transarterial chemoembolization (TACE).

Between January 2011 and January 2012, 66 hypervascular malignant liver tumors from patients who underwent TACE with contrast-enhanced CBCT at the arterial phase were included (51 HCC, 13 NET and 2 adrenal cancer metastases). Data were analyzed by 6 interventional radiologists blinded to each other analyses (2 junior and 4 experienced readers). Readers were asked to identify tumor feeders by performing 1) a conventional analysis using post-processing tools such as maximum intensity projection, multiplanar reconstruction, volume rendering, 2/ a computer-aided analysis using FlightPlan for liver (referred to as raw-FPFL), and 3) a review of this computer aided analysis for which reader were asked to validate or invalidate each feeder detected by the software (referred to as reviewed-FPFL). Analyses were compared to a “Reference Reading” established by two study supervisors in consensus. Sensitivities, positive predictive values (PPV), and false positive ratios (FPR) were compared using Mac-Nemar, Chi-square and exact Fisher tests. Analysis durations were compared using a Mann-Whitney U test. Inter-readers agreements were assessed by mean of percentage of agreement. 

A total of 179 feeding vessels were identified in the ‘Reference Reading’. The sensitivity of raw-FPFL was significantly higher than that of both reviewed-FPFL and conventional analyses (90.9% vs. 83.2% and 82.1%, p<0.0001), with lower PPV (82.9% vs. 91.2% and 90.6%, respectively (p<0.0001), higher FPR (17.1% vs. 9.4% and 8.8%, respectively (p<0.0001), and higher inter-reader agreement (92% vs. 80 and 79%, respectively, p<0.0001). The conventional analysis was significantly longer than that of both raw- and reviewed-FPFL (<0.0001).

Contrast-enhanced CBCT with software analysis enabled accurate and sensitive detection of tumor feeders of malignant liver tumors before TACE. The review of the software analysis was responsible for a significant decrease in the number of identified feeders.

Dedicated software analysis of contrast-enhanced CBCT images should be used when performing transarterial chemoembolization of liver tumors. 

Ronot, M, Abdel-Rehim, M, Kuoch, V, Hakime, A, Roux, M, Chiaradia, M, Vilgrain, V, De Baere, T, Deschamps, F, Cone-Beam Computed Tomography Angiography for Depiction of Tumor-feeding Vessels during Chemoembolization of Malignant Liver Tumors: Comparison of Conventional and Dedicated-software Analysis.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14003708.html