Uncertainty-aware visual analysis of radiofrequency ablation simulations

Abstract

Radiofrequency (RF) ablation is a medical procedure for treating tumors by generating heat with RF current using needle-like probes that are inserted into the tumor. During RF ablation the entire tumor shall be destroyed to avoid recurrence while keeping the destruction of surrounding healthy tissue minimal. As the ablation area is affected by surrounding structures such as vessels, developing biomedical simulations that support treatment planning is an active research field. The simulations assess the area based on patient-specific imaging data to lower the risk of under- or over-ablation. However, there are several simulation parameters whose precise values are unknown. We capture and propagate these uncertainties by deploying a stochastic simulation model and computing the uncertainties in the resulting ablation probabilities. We propose an uncertainty-aware interactive visual analysis system of the probabilities based on coordinated views of volume visualizations, a slice-based visualization of 2D uncertainty glyphs with a perceptually uniform color-coding, and parameter analysis plots. Using our interactive system, simulation experts can analyze the influence of their parameter choices on the ablation probabilities and their uncertainties. Moreover, our glyph encoding provides an inherent continuous level-of-detail approach during zooming interaction that makes it suitable for application in clinical settings. We apply our techniques to RF ablation simulations for liver tumor treatment and document their value within a controlled user study.