Probabilistic finite element method for large tumor radiofrequency ablation simulation and planning

Abstract

A challenging problem of radiofrequency ablation (RFA) in liver surgery is to accurately estimate the shapes and sizes of RFA lesions whose formation depends on intrinsic variations of the thermal-electrical properties of soft tissue. Large tumors, which can be as long as 10 cm or more, further complicate the problem. In this paper, a probabilistic bio-heating finite element (FE) model is proposed and developed to predict RFA lesions. Uncertainties of RFA lesions are caused by the probabilistic nature of five thermal-electrical liver properties: thermal conductivity, liver tissue density, specific heat, blood perfusion rate and electrical conductivity. Confidence levels of shapes and sizes of lesions are generated by the FE model incorporated with the mean-value first-order second-moment (MVFOSM) method. Based on the probabilistic FE method, a workflow of RFA planning is introduced to enable clinicians to preoperatively view the predicted RFA lesions in three-dimension (3D) within a hepatic environment. Accurate planning of the RFA needle placements can then be achieved based on the interactive simulation and confidence level selection.

Keywords: Mean-value first-order second-moment method; Probabilistic finite element simulation; Radiofrequency ablation; Radiofrequency ablation planning.