Related Experiment Videos
Accelerating Monte Carlo simulations of radiation therapy dose distributions using wavelet threshold de-noising
Joseph O Deasy1, M Victor Wickerhauser, Mathieu Picard
1Department of Radiation Oncology, Mallinckrodt Institute of Radiology, Washington University School of Medicine, St. Louis, Missouri 63110, USA. deasy@radonc.wustl.edu
Medical Physics
|November 1, 2002
Summary
Wavelet de-noising accelerates Monte Carlo dose calculations by post-processing noise. This method significantly reduces simulation time while maintaining dose accuracy, making it a promising technique for radiation therapy planning.
Area of Science:
- Medical Physics
- Computational Science
- Image Processing
Background:
- Monte Carlo (MC) methods are crucial for accurate radiation dose calculations.
- MC simulations require a large number of particles for convergence, leading to long computation times.
- Noise in MC dose distributions necessitates post-processing for accurate analysis.
Purpose of the Study:
- To introduce wavelet threshold de-noising as a post-processing technique to accelerate MC dose calculations.
- To evaluate the effectiveness of wavelet de-noising in suppressing noise and improving convergence speed.
- To assess the impact of de-noising on dose accuracy and image smoothness.
Main Methods:
- Implementation of wavelet hard-threshold de-noising using 9,7-biorthogonal filters in C.
- Averaging transform results over origin selections to minimize artifacts.
- Development of a method for optimal threshold value selection.
- Application to 2D dose distributions from electron beams in water and lung heterogeneity phantoms using the Integrated Tiger Series MC code.
Main Results:
- Wavelet de-noising effectively suppressed voxel-to-voxel noise with minimal bias introduction.
- The roughness of de-noised dose distributions showed near-independence from the number of simulated electrons.
- Dose image accuracy improved with an increasing number of source particles.
- The algorithm requires approximately 336 floating-point operations per dose grid point.
Conclusions:
- Wavelet shrinkage de-noising is a promising method for accelerating Monte Carlo dose calculations.
- The technique can achieve acceleration factors of 2 or more.
- This approach offers a computationally efficient way to improve the speed and quality of MC dose simulations.