Abstract
Percutaneous thermal ablation is increasingly popular but still suffers from a complex preoperative planning, especially regarding the prediction of the ablation zone. We propose HEAT (High-Efficiency simulation for thermal Ablation Therapy), a novel GPU-accelerated simulation framework for thermal ablation that enables real-time planning.
This work was published in the International Journal of Computer Assisted Radiology and Surgery (IJCARS) in 2025.
Motivation
Thermal ablation is a minimally invasive cancer treatment that uses heat to destroy tumors. However, accurately predicting the ablation zone (the area that will be destroyed) remains challenging. Traditional simulation methods are too slow for clinical use, taking minutes or even hours to compute.
Approach
HEAT leverages GPU acceleration to achieve real-time performance:
- Fast computation: Results in under 1 second for interactive planning
- GPU parallelization: Efficient handling of 3D volumetric data
- Physical accuracy: Implements realistic heat transfer models including:
- Pennes bioheat equation
- Perfusion effects
- Heat sink effects near blood vessels
Results
The system demonstrates:
- Real-time performance: Sub-second simulation times
- Clinical accuracy: Validated against experimental data
- Practical utility: Integrates with existing surgical planning workflows
Authors
- Jonas Mehtali (Lead Author)
- Juan Verde
- Caroline Essert
ICube Laboratory, University of Strasbourg
Related Work
This work builds on my research internship in 2024 at ICube/IHU Strasbourg, which focused on networked computing for thermal ablation simulation. The HEAT project represents a significant advancement in computational efficiency and clinical applicability.
See also: Cryotrack: Planning and Navigation for Computer Assisted Cryoablation (MICCAI 2024)