Control of radiofrequency ablation in ex vivo human liver tissue using 3D echo decorrelation imaging feedback

Limitations of radiofrequency ablation (RFA) treatment for nonresectable liver cancer could potentially be overcome by monitoring and controlling RFA to confirm complete treatment. Echo decorrelation imaging has been shown to predict local thermal ablation and to successfully control ablation in animal liver and tumor tissue, both ex vivo and in vivo. Toward translation of this monitoring and control approach to clinical liver tumor ablation, this study investigated control of RFA in ex vivo human liver tissue using 3D echo decorrelation imaging. Human liver tissue, including postmortem normal liver as well as explanted cirrhotic and fatty liver, was ablated using a clinical RFA system. During ablation, 3D echo decorrelation images were constructed from paired sequential volumes of acquired echo data. For controlled RFA trials, ablation was ceased automatically after cumulative echo decorrelation, spatially averaged within a spherical ROI matching the planned ablation zone, exceeded a predetermined threshold. This control algorithm was tested in N=6 controlled ablation trials, of which 5 attained the predetermined decorrelation threshold. 3D echo decorrelation maps were compared to segmented ablation zones, using receiver operating characteristic (ROC) curve analysis. To assess the effect of image-based feedback control on treatment reliability, ablation volumes and rates, as well as Dice coefficients for measured vs. targeted ablation zones, were statistically compared for controlled vs. uncontrolled trials. Results suggest potential for 3D echo decorrelation imaging as a monitoring and control approach for clinical ablation of liver tumors.