Virtual Source Monte Carlo Dose Calculation With Reduced I/O Burden
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Monte Carlo simulations for dose calculations in computed tomography are hindered by the computational resources and time required, particularly due to the large size of phase space files and the I/O burden, which limits their clinical application.
Innovation Solution
A method and apparatus utilizing a virtual source model (VSM) for fast Monte Carlo dose calculation, involving the processing of 3D images and radiotherapy plans to build inverse cumulative density function tables, simulating and transporting beams through virtual treatment machines, and post-processing dose images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If phase space files are used for Monte Carlo simulations, then accurate dose calculation is achieved, but computational time and I/O burden increase significantly
Solution Approach 1:
The patent pre-calculates and stores particle parameters (position, direction, energy) at a reference surface before actual dose calculation. This preliminary action creates phase space files that can be reused across multiple simulations, eliminating the need to re-simulate particle transport through the treatment head for each new dose calculation scenario.
Solution Approach 2:
The patent creates a virtual copy of the particle distribution at the reference surface, storing essential parameters in a compressed format. This copy contains sufficient information to reconstruct particle behavior without requiring the full computational resources of a complete treatment head simulation, enabling faster subsequent calculations.
2Measurement precision
If phase space files are used for Monte Carlo simulations, then accurate dose calculation is achieved, but disk space consumption increases significantly
Solution Approach 1:
The patent extracts only the essential particle parameters (position, direction cosines, energy) at the reference surface, discarding redundant information about particle generation and transport through the treatment head. This extraction creates a compact phase space representation that maintains accuracy while reducing storage requirements.
Solution Approach 2:
The patent changes the representation parameters by storing particles in a normalized coordinate system at a standardized reference surface, using binned histograms for directional distributions. This parameter transformation reduces the data volume while preserving the physical information needed for accurate dose calculation.
3Productivity
If multi-core processing is used for Monte Carlo simulations, then computational speed is improved, but I/O bandwidth contention increases
Solution Approach 1:
The patent segments the particle data into independent, randomly accessible units that can be loaded into memory once and then processed by multiple cores simultaneously without further I/O operations. This segmentation eliminates I/O bandwidth contention between cores while maintaining parallel processing speedup.
Data Source
AI summary
The present disclosure relates to a method and apparatus for fast Monte Carlo (MC) dose calculation using a virtual source model (VSM). The method includes: receiving three-dimensional (3D) CT images obtained by a CT system; receiving 3D planned dose images, 3D organ segmentation contour images, and radiotherapy plans generated by a treatment planning system (TPS); processing 3D CT images, 3D planned dose images, 3D organ segmentation contour images to have the same spatial resolution and matrix size; further processing 3D CT images to convert image intensity to 3D density maps; processing the radiotherapy plans to generate instructions on how to simulate plan delivery; building VSM using inverse cumulative density function (CDF) tables for the simulation of radiotherapy plans, wherein the step of building VSM comprises: receiving output data files containing phase-space information for the radiation output of a specific medical linear accelerator treatment head; calculating the probability of the inplane and crossplane positions of the radiation particles reverse transported from the phase-space surface back to the treatment head; calculating the Gaussian means and standard deviations of the radiation particles' positions at the treatment head; calculating the probabilities for the source of each radiation particle; calculating the probabilities for the medical linear accelerator treatment head to produce different radiation particle species; binning the inplane position probability information of radiation particles into a single histogram for each source and radiation particle species; binning the crossplane position probability information of radiation particles into histograms for each bin of the inplane position histogram for each source and radiation particle species; binning the inplane direction cosine probability information of radiation particles into histograms for each bin of the inplane position histogram for each source and radiation particle species; binning the crossplane direction cosine probability information of radiation particles into histograms for each bin of the crossplane position histogram for each source and radiation particle species; binning the kinetic energy probability information of radiation particles into radially binned histograms for each source and radiation particle species; converting probability densities for inplane and crossplane positions, inplane and crossplane direction cosines, and kinetic energies histograms into cumulative probability densities for each source and radiation particle species; and inverting cumulative probability densities and converting into probability binned inverse CDF tables; simulating and transporting external beams using VSM through virtual treatment machines to the 3D density maps according to radiotherapy plans to produce 3D simulated dose images; and post-processing the 3D planned dose images, 3D organ segmentation contour images, and the 3D simulated dose images to obtain a final report comparing planned versus simulated dose images.