Real-time modelling and visualization of soft tissue thermomechanical behaviour for radiofrequency thermal ablation

A review of current literature indicates a limited knowledge and documentation of thermomechanical response of soft tissue during Minimally Invasive Surgical (MIS) hyperthermia procedures such as Radiofrequency Thermal Ablation (RFA). Furthermore, current models and simulations have not accounted for the temperature-dependence of the stress-strain behaviour of soft tissue. The only quantified data for temperature-dependent stress-strain relationships in literature is yielded from Xu and Lu (2009) and Xu, Seffen and Lu (2008a). As well as this, hardware-accelerated (by use of Graphics Processing Units (GPUs)) heat transfer simulations of RFA had not been documented prior to commencement of this project, and the first conference paper announcing this achievement was published in April of 2014 following research and implementation by NE Scientific LLC. A computational three-dimensional (3D) simulated virtual model of liver tissue is developed to establish temperature distributions resulting from single point temperature sources in emulation of the RFA heat treatment procedure. The temperature distribution in the virtual tissue domain is produced by Finite Element (FEM) spatial discretization and Finite Difference (FDM) temporal discretization of the Pennes bioheat transfer equation. The modelled temperature distribution is utilized to determine the degree of transient thermal damage to the virtual tissue based on the Arrhenius Burn Integration. Furthermore, the temperature distribution is used in conjunction with linear thermal expansion to model thermal strains and thermal stresses within the virtual tissue, resulting from the heat source. Novel work is undertaken in producing a thermal stress profile of virtual liver tissue under operational temperatures based on temperature-dependent material stress-strain. The simulation of tissue bioheat transfer and thermal damage is developed in C++ and the High-Level Shader Language (HLSL) with Microsoft’s Direct3D11 Application Programming Interface (API), where the numerical solution process is parallelized and accelerated in performance upon an NVIDIA GTX 770M GPU far beyond its performance upon a single thread/core of an Intel® Core™ i7-4700MQ Central Processing Unit (CPU). A maximum mesh resolution is determined for producing visual real-time post-processing output data based on the GPU accelerated simulation performance. A commercial FEM software package (LISA) is used for determining thermal strain and thermal stress distributions from the temperature distribution data. Examination of simulation results when comparing tissue thermomechanical response for temperature-independent and temperature-dependent stress-strain relationships indicates a dramatic difference in magnitude and distribution of the thermally-induced stresses within the soft tissue. The implications are that RFA simulations must account for this stress-strain temperature-dependence in order to produce remotely accurate stress and strain distributions (both thermomechanical and mechanical) due to the behavioural response of the collagenous biological soft tissue. Furthermore, GPU acceleration is highly recommended for RFA simulation, as the visual real-time maximum mesh resolution far surpasses that of real-time performed on a single modern CPU core