DosePatch: physics-inspired cropping layout for patch-based Monte Carlo simulations to provide fast and accurate...

Area of Science:

• Medical Physics
• Computational Biology
• Radiotherapy

Background:

• Personalized dosimetry-based therapies, like liver selective internal radiation therapy (SIRT), offer clinical benefits.
• Widespread adoption is hindered by the extensive computation time required for Monte Carlo (MC) simulations, the gold standard for dosimetry.
• Addressing computation time is crucial for advancing dosimetry in clinical practice.

Purpose of the Study:

• To investigate a patch-based approach for MC simulations to accelerate internal dosimetry.
• To introduce and evaluate a novel physics-inspired cropping layout for patch-based MC dosimetry.
• To compare the accuracy and efficiency of this new method against existing techniques and whole-body MC simulations.

Main Methods:

• Developed a patch-based MC simulation technique with a physics-inspired cropping layout for improved parallelization.
• Compared dosimetry results from the patch-based MC approach to organ-S-values, dose kernels, and whole-body MC simulations.
• Evaluated the method in five patients undergoing Yttrium-90 liver SIRT.

Main Results:

• The patch-based MC approach closely approximated the ground truth from whole-body simulations, proving a viable alternative.
• The physics-inspired cropping layout achieved a voxel-wise error of less than 2% in soft tissue compared to whole-body MC.
• This method required only 10% of the computation time of conventional approaches.

Conclusions:

• Patch-based MC simulations with physics-inspired cropping layouts are feasible and accurate for internal dosimetry.
• This approach offers a significant reduction in computation time, paving the way for faster personalized dosimetry.
• The findings support the integration of accelerated MC simulations into clinical workflows for radiation therapy.