Dataset Information

Fast patient-specific Monte Carlo brachytherapy dose calculations via the correlated sampling variance reduction technique.

ABSTRACT: To demonstrate potential of correlated sampling Monte Carlo (CMC) simulation to improve the calculation efficiency for permanent seed brachytherapy (PSB) implants without loss of accuracy.CMC was implemented within an in-house MC code family (PTRAN) and used to compute 3D dose distributions for two patient cases: a clinical PSB postimplant prostate CT imaging study and a simulated post lumpectomy breast PSB implant planned on a screening dedicated breast cone-beam CT patient exam. CMC tallies the dose difference, ?D, between highly correlated histories in homogeneous and heterogeneous geometries. The heterogeneous geometry histories were derived from photon collisions sampled in a geometrically identical but purely homogeneous medium geometry, by altering their particle weights to correct for bias. The prostate case consisted of 78 Model-6711 (125)I seeds. The breast case consisted of 87 Model-200 (103)Pd seeds embedded around a simulated lumpectomy cavity. Systematic and random errors in CMC were unfolded using low-uncertainty uncorrelated MC (UMC) as the benchmark. CMC efficiency gains, relative to UMC, were computed for all voxels, and the mean was classified in regions that received minimum doses greater than 20%, 50%, and 90% of D(90), as well as for various anatomical regions.Systematic errors in CMC relative to UMC were less than 0.6% for 99% of the voxels and 0.04% for 100% of the voxels for the prostate and breast cases, respectively. For a 1?×?1?×?1 mm(3) dose grid, efficiency gains were realized in all structures with 38.1- and 59.8-fold average gains within the prostate and breast clinical target volumes (CTVs), respectively. Greater than 99% of the voxels within the prostate and breast CTVs experienced an efficiency gain. Additionally, it was shown that efficiency losses were confined to low dose regions while the largest gains were located where little difference exists between the homogeneous and heterogeneous doses. On an AMD 1090T processor, computing times of 38 and 21 sec were required to achieve an average statistical uncertainty of 2% within the prostate (1?×?1?×?1 mm(3)) and breast (0.67?×?0.67?×?0.8 mm(3)) CTVs, respectively.CMC supports an additional average 38-60 fold improvement in average efficiency relative to conventional uncorrelated MC techniques, although some voxels experience no gain or even efficiency losses. However, for the two investigated case studies, the maximum variance within clinically significant structures was always reduced (on average by a factor of 6) in the therapeutic dose range generally. CMC takes only seconds to produce an accurate, high-resolution, low-uncertainly dose distribution for the low-energy PSB implants investigated in this study.

SUBMITTER: Sampson A

PROVIDER: S-EPMC3281974 | biostudies-literature | 2012 Feb

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Fast patient-specific Monte Carlo brachytherapy dose calculations via the correlated sampling variance reduction technique.

Sampson Andrew A Le Yi Y Williamson Jeffrey F JF

Medical physics 20120201 2

<h4>Purpose</h4>To demonstrate potential of correlated sampling Monte Carlo (CMC) simulation to improve the calculation efficiency for permanent seed brachytherapy (PSB) implants without loss of accuracy.<h4>Methods</h4>CMC was implemented within an in-house MC code family (PTRAN) and used to compute 3D dose distributions for two patient cases: a clinical PSB postimplant prostate CT imaging study and a simulated post lumpectomy breast PSB implant planned on a screening dedicated breast cone-beam ...[more]

PMID: 22320816

Similar Datasets

Project description:In diffusion MRI analysis, advances in biophysical multi-compartment modeling have gained popularity over the conventional Diffusion Tensor Imaging (DTI), because they can obtain a greater specificity in relating the dMRI signal to underlying cellular microstructure. Biophysical multi-compartment models require a parameter estimation, typically performed using either the Maximum Likelihood Estimation (MLE) or the Markov Chain Monte Carlo (MCMC) sampling. Whereas, the MLE provides only a point estimate of the fitted model parameters, the MCMC recovers the entire posterior distribution of the model parameters given in the data, providing additional information such as parameter uncertainty and correlations. MCMC sampling is currently not routinely applied in dMRI microstructure modeling, as it requires adjustment and tuning, specific to each model, particularly in the choice of proposal distributions, burn-in length, thinning, and the number of samples to store. In addition, sampling often takes at least an order of magnitude, more time than non-linear optimization. Here we investigate the performance of the MCMC algorithm variations over multiple popular diffusion microstructure models, to examine whether a single, well performing variation could be applied efficiently and robustly to many models. Using an efficient GPU-based implementation, we showed that run times can be removed as a prohibitive constraint for the sampling of diffusion multi-compartment models. Using this implementation, we investigated the effectiveness of different adaptive MCMC algorithms, burn-in, initialization, and thinning. Finally we applied the theory of the Effective Sample Size, to the diffusion multi-compartment models, as a way of determining a relatively general target for the number of samples needed to characterize parameter distributions for different models and data sets. We conclude that adaptive Metropolis methods increase MCMC performance and select the Adaptive Metropolis-Within-Gibbs (AMWG) algorithm as the primary method. We furthermore advise to initialize the sampling with an MLE point estimate, in which case 100 to 200 samples are sufficient as a burn-in. Finally, we advise against thinning in most use-cases and as a relatively general target for the number of samples, we recommend a multivariate Effective Sample Size of 2,200.

Project description:A new design of 125I (Model IR-Seed2) brachytherapy source has been manufactured recently at the Applied Radiation Research School, Nuclear Science and Technology Research Institute in Iran. The source consists of six resin beads (0.5 mm diameter) that are sealed in a cylindrical titanium capsule of 0.7 mm internal and 0.8 mm external diameters. This work aims to evaluate the dosimetric parameters of the newly designed 125I source using experimental measurements and Monte Carlo (MC) simulations. Dosimetric characteristics (dose rate constant, radial dose function, and 2D and 1D anisotropy functions) of the IR-Seed2 were determined using experimental measurements and MC simulations following the recommendations by the Task Group 43 (TG-43U1) report of the American Association of Physicists in Medicine (AAPM). MC simulations were performed using the MCNP5 code in water and Plexiglas, and experimental measurements were carried out using thermoluminescent dosimeters (TLD-GR207A) in Plexiglas phantoms. The measured dose to water in Plexiglas data were used for verification of the accuracy of the source and phantom geometry in the Monte Carlo simulations. The final MC simulated data to water in water were recommended for clinical applications. The MC calculated dose rate constant (?) of the IR-Seed2 125I seed in water was found to be 0.992 ± 0.025 cGy h-1U-1. Additionally, its radial dose function by line and point source approximations, gL(r) and gp(r), calculated for distances from 0.1 cm to 7 cm. The values of gL(r) at radial distances from 0.5 cm to 5 cm were measured in a Plexiglas phantom to be between 1.212 and 0.413. The calculated and measured of values for 2D anisotropy function, F(r, ?), were obtained for the radial distances ranging from 1.5 cm to 5 cm and angular range of 0°-90° in a Plexiglas phantom. Also, the 2D anisotropy function was calculated in water for the clinical application. The results of these investigations show that the uncertainty of the experimental data is within ± 7% between the measured and simulated data in Plexiglas. Based on these results, the MC-simulated dosimetric parameters of the new 125I source model in water are presented for its clinical applications in brachytherapy treatments.

Project description:To present the implementation and validation of a geometrical based variance reduction technique for the calculation of phase space data for proton therapy dose calculation.The treatment heads at the Francis H Burr Proton Therapy Center were modeled with a new Monte Carlo tool (TOPAS based on Geant4). For variance reduction purposes, two particle-splitting planes were implemented. First, the particles were split upstream of the second scatterer or at the second ionization chamber. Then, particles reaching another plane immediately upstream of the field specific aperture were split again. In each case, particles were split by a factor of 8. At the second ionization chamber and at the latter plane, the cylindrical symmetry of the proton beam was exploited to position the split particles at randomly spaced locations rotated around the beam axis. Phase space data in IAEA format were recorded at the treatment head exit and the computational efficiency was calculated. Depth-dose curves and beam profiles were analyzed. Dose distributions were compared for a voxelized water phantom for different treatment fields for both the reference and optimized simulations. In addition, dose in two patients was simulated with and without particle splitting to compare the efficiency and accuracy of the technique.A normalized computational efficiency gain of a factor of 10-20.3 was reached for phase space calculations for the different treatment head options simulated. Depth-dose curves and beam profiles were in reasonable agreement with the simulation done without splitting: within 1% for depth-dose with an average difference of (0.2 ± 0.4)%, 1 standard deviation, and a 0.3% statistical uncertainty of the simulations in the high dose region; 1.6% for planar fluence with an average difference of (0.4 ± 0.5)% and a statistical uncertainty of 0.3% in the high fluence region. The percentage differences between dose distributions in water for simulations done with and without particle splitting were within the accepted clinical tolerance of 2%, with a 0.4% statistical uncertainty. For the two patient geometries considered, head and prostate, the efficiency gain was 20.9 and 14.7, respectively, with the percentages of voxels with gamma indices lower than unity 98.9% and 99.7%, respectively, using 2% and 2 mm criteria.The authors have implemented an efficient variance reduction technique with significant speed improvements for proton Monte Carlo simulations. The method can be transferred to other codes and other treatment heads.

Dataset Information

Fast patient-specific Monte Carlo brachytherapy dose calculations via the correlated sampling variance reduction technique.

Publications

Fast patient-specific Monte Carlo brachytherapy dose calculations via the correlated sampling variance reduction technique.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets