Problem and Context
For many of the radio-frequency ablations (RFA) performed in the liver, there is still a significant mismatch between expected and truly induced lesion size. This can lead to over-treatment with severe injuries (up to 9% major complications), or under-treatment with tumour recurrence (up to 40%).
There are several individual factors related to inner body structures (e.g. vessel flow, liver tissue properties, tumour type and perfusion, proximity to vessels and liver surface etc.) which substantially affect the size and configuration of the ablated lesion. All these factors are not taken into account in the manufacturer's ablation protocol. It is known that the learning curve for RFA treatment is relatively low and reliable results typically require a thorough training of the IRs in the order of two years. A validated planning and simulation tool could shorten this period but is currently not commercially available.
The proposed RFA planning and support tool is therefore unique as it offers a validated software environment, where IRs can interact during the RFA treatment with the virtual tumour ablation through the extensive use of simulation and visualisation technology.
One main objective of the project is the development of an integrated, accurate tool for predicting RFA-induced lesions in liver tissue.
Sophisticated multi-disciplinary development, comprising image processing, perfusion measurement, real-time simulation, and visualization, ensures best possible parameters for quick and accurate prediction of treatment.
The resulting application combines a multitude of high-performance, high-accuracy algorithms in a single, integrated environment with high attention to usability in the clinical practice.
For optimal usage conditions, the distinct workflows in four clinical sites have been analyzed and unified into a common procedure. The results of this analysis significantly contributed to the development of the RFA planning application.
Testing results suggest acceptable usability conditions for all medical partners, despite the fact that the unification of the workflows omits some local details. The distinction of several phases allows us to focus the effort of creating high-performance methods for stages where time is a critical aspect.
While the pre-interventional and post-interventional phases take place off-line, the interventional phase requires special attention, since we must ensure minimal computation time during in-situ simulation as the patient is under general anesthesia.
A clinical study is at the core of the project. The patient data set acquired in the study is used to validate the software prediction and to improve its quality and accuracy. The existing RFA model has to be adapted to the real-time requirements of the clinical environment and to be integrated into the clinical workflow. The developed and validated software model has to be fully integrated into the interventional work flow to allow IRs for real-time and patient specific planning and support of the RFA intervention.