Open Access System for Information Sharing

Login Library

 

Thesis
Cited 0 time in webofscience Cited 0 time in scopus
Metadata Downloads

Development of a Dose Estimation Code for BNCT with GPU-accelerated Monte Carlo and Collapsed Cone Convolution Method

Title
Development of a Dose Estimation Code for BNCT with GPU-accelerated Monte Carlo and Collapsed Cone Convolution Method
Authors
이창민
Date Issued
2021
Publisher
포항공과대학교
Abstract
The conventional Monte Carlo based treatment planning system of Boron Neutron Capture Therapy (BNCT) is time consuming to calculate the distribution of the treatment dose. In pre-study, Monte Carlo and diffusion hybrid method was developed and tested to improve the calculation speed. However, this method was not effective to reduce the calculation time to an appropriate level. Therefore, a new method of dose calculation algorithm, called GPU-accelerated Monte Carlo and collapsed cone Convolution (GMCC) was developed to improve the calculation speed to a practical level. The GMCC consists of a GPU-accelerated Monte Carlo method and a Collapsed Cone Convolution method. The former is used to calculate the neutron flux over a whole energy range, fast to thermal, and the latter is used to calculate the gamma dose. Other dose components, caused by alpha particles and protons, are calculated by the multiplication of the neutron flux, reaction rate and Q-value. In this paper, the mathematical principle and the algorithm architecture of the GMCC are introduced. The accuracy and performance of the GMCC are verified by comparing with the FLUKA results. A water phantom and a head CT voxel model were simulated by the GMCC. The comparison of the neutron flux and the absorbed dose obtained by the GMCC and the FLUKA calculations were consistent. In the case of head CT voxel model, the comparison showed the mean absolute percentage error for the neutron flux and the absorbed dose as 3.98% and 3.91%, respectively. In terms of performance, absorbed dose calculation of the GMCC was 56 times faster than the FLUKA code. It was found out that the GMCC could be a good candidate tool instead of the Monte Carlo method in the BNCT dose calculations.
현재 전 세계의 모든 기관에서 중성자포획치료(BNCT)의 선량 계산은 몬테카를로 코드를 사용하고 있다. 몬테카를로 방법은 정확도가 높지만, 계산의 통계적 오차를 충분히 줄이기 위해서는 매우 오랜 시간의 연산 시간을 요구한다는 문제점이 있다. 따라서 본 논문에서는 몬테카를로 방법의 정확성을 유지하면서 연산 속도를 향상시키기 위해 새로운 알고리즘을 시도했다. 선행 연구에서는 확산방정식-몬테카를로 혼합 방식을 시도했으나 만족할 만한 속도향상 효과를 보이지 못했다. 이후 CUDA C++ 코드를 이용한 GPU 기반 몬테카를로와 기존 Brachytherapy Collapsed Cone Convolution 알고리즘을 연계한 새로운 알고리즘(GMCC)을 개발하여 BNCT 선량 계산을 시도했으며, 계산 가속 효율이 뛰어남을 확인했다. GPU 기반 몬테카를로 방법은 기존 CPU 기반 몬테카를로 방법의 장점을 그대로 가지고 있으면서, 연산 속도가 수십 배 빠른 장점을 가지고 있다. 따라서 GMCC의 Convolution 알고리즘을 GPU 기반 몬테카를로 코드로 모두 대체하여 Convolution 연산에서 발생 가능한 문제점을 모두 해결하며 빠른 연산 속도를 유지하는 새로운 알고리즘을 검토 중이며, 더 나아가 치료 분야 이외의 모든 분야에 사용 가능한 GPU 기반 범용 몬테카를로 코드 개발을 계획하고 있다.
URI
http://postech.dcollection.net/common/orgView/200000508136
https://oasis.postech.ac.kr/handle/2014.oak/114192
Article Type
Thesis
Files in This Item:
There are no files associated with this item.

qr_code

  • mendeley

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

Views & Downloads

Browse