GPU-based 3D iceball modeling for fast cryoablation simulation and planning.

Area of Science:

• Medical Imaging and Simulation
• Computational Physics
• Oncology

Background:

• Percutaneous cryoablation is a minimally invasive technique for abdominal tumor destruction.
• Accurate planning of cryoablation requires precise prediction of iceball formation.
• Challenges in planning include determining cryoprobe placement and accounting for heat sources like blood vessels.

Purpose of the Study:

• To develop a fast GPU-based method for modeling iceball formation during cryoablation.
• To accurately predict iceball shape by simulating thermal propagation and accounting for heat sources.

Main Methods:

• A GPU-based algorithm was developed to solve the heat equation around cryoprobe tips.
• The model simulates thermal propagation within a defined tissue volume.
• It incorporates the influence of heating sources, such as major blood vessels and warm saline.

Main Results:

• The method was validated using ex vivo experiments and simulations on five kidney tumor cryoablation cases.
• Accurate iceball shape predictions were achieved within seconds.
• The algorithm demonstrated effectiveness across varying complexities of vascular structures and saline protection.

Conclusions:

• The developed GPU-based method provides accurate and rapid iceball shape predictions.
• This technique is suitable for enhancing the surgical planning of percutaneous cryoablation procedures.
• The speed and accuracy show potential for real-time application in clinical settings.