DeepBeam: a machine learning framework for tuning the primary electron beam of the PRIMO Monte Carlo software
Authors:
- Zbisław Tabor,
- Damian Kabat,
- Michael P.R. Waligórski
Abstract
Background: Any Monte Carlo simulation of dose delivery using medical accelerator-generated megavolt photon beams begins by simulating electrons of the primary electron beam interacting with a target. Because the electron beam characteristics of any single accelerator are unique and generally unknown, an appropriate model of an electron beam must be assumed before MC simulations can be run. The purpose of the present study is to develop a fexible framework with suitable regression models for estimating parameters of the model of primary electron beam in simulators of medical linear accelerators using real reference dose profles measured in a water phantom. Methods: All simulations were run using PRIMO MC simulator. Two regression models for estimating the parameters of the simulated primary electron beam, both based on machine learning, were developed. The frst model applies Principal Component Analysis to measured dose profles in order to extract principal features of the shapes of the these profles. The PCA-obtained features are then used by Support Vector Regressors to estimate the parameters of the model of the electron beam. The second model, based on deep learning, consists of a set of encoders processing measured dose profles, followed by a sequence of fully connected layers acting together, which solve the regression problem of estimating values of the electron beam parameters directly from the measured dose profles. Results of the regression are then used to reconstruct the dose profles based on the PCA model. Agreement between the measured and reconstructed profles can be further improved by an optimization procedure resulting in the fnal estimates of the parameters of the model of the primary electron beam. These fnal estimates are then used to determine dose profles in MC simulations. Results: Analysed were a set of actually measured (real) dose profles of 6 MV beams from a real Varian 2300 C/D accelerator, a set of simulated training profles, and a separate set of simulated testing profles, both generated for a range of parameters of the primary electron beam of the Varian 2300 C/D PRIMO simulator. Application of the two-stage procedure based on regression followed by reconstruction-based minimization of the diference between measured (real) and reconstructed profles resulted in achieving consistent estimates of electron beam parameters and in a very good agreement between the measured and simulated photon beam profles. Conclusions: The proposed framework is a readily applicable and customizable tool which may be applied in tuning virtual primary electron beams of Monte Carlo simulators of linear accelerators. The codes, training and test data, together with readout procedures, are freely available at the site: https://github.com/taborzbislaw/DeepBeam.
- Record ID
- CUT46f7ddb1702843b7b67b67c0bd0ad107
- Publication categories
- ;
- Author
- Journal series
- Radiation Oncology, ISSN 1748-717X
- Issue year
- 2021
- Vol
- 16
- Pages
- [1-14]
- Article number
- 124
- Other elements of collation
- schem.; tab.; wykr.; Bibliografia (na s.) - 13-14; Bibliografia (liczba pozycji) - 23; Oznaczenie streszczenia - Abstr.; Numeracja w czasopiśmie - Vol. 16
- Substantive notes
- Collection: The Monte Carlo radiotherapy system PRIMO: Dosimetry and treatment planning research problems
- The online version contains supplementary material available at: https://doi. org/10.1186/s13014-021-01847-w
- Keywords in English
- machine learning, deep learning, Monte Carlo, beam simulation, Quality Assurance (QA), Quality Control (QC), Principal Component Analysis (PCA), support vector regression
- DOI
- DOI:10.1186/s13014-021-01847-w Opening in a new tab
- URL
- https://ro-journal.biomedcentral.com/articles/10.1186/s13014-021-01847-w Opening in a new tab
- Related project
- Rekonfigurowalny detektor do pomiaru przestrzennego rozkładu dawki promieniowania dla zastosowań w przygotowaniu indywidualnych planów leczenia pacjentów. . Project leader at PK: , ,
- Language
- eng (en) English
- License
- Score (nominal)
- 100
- Additional fields
- Indeksowana w: Web of Science, Scopus
- Uniform Resource Identifier
- https://cris.pk.edu.pl/info/article/CUT46f7ddb1702843b7b67b67c0bd0ad107/
- URN
urn:pkr-prod:CUT46f7ddb1702843b7b67b67c0bd0ad107
* presented citation count is obtained through Internet information analysis, and it is close to the number calculated by the Publish or PerishOpening in a new tab system.