[基于CUDA的放射治疗快速剂量计算]

[CUDA-based fast dose calculation in radiotherapy].

作者信息

Wang Xianliang, Liu Cao, Hou Qing

机构信息

Key Laboratory of Radiation Physics and Technology, Ministry of Nuclear Science and Technology, Sichuan University, Chengdu 610064, China.

出版信息

Sheng Wu Yi Xue Gong Cheng Xue Za Zhi. 2011 Oct;28(5):881-5.

PMID:22097248

Abstract

Dose calculation plays a key role in treatment planning of radiotherapy. Algorithms for dose calculation require high accuracy and computational efficiency. Finite size pencil beam (FSPB) algorithm is a method commonly adopted in the treatment planning system for radiotherapy. However, improvement on its computational efficiency is still desirable for such purpose as real time treatment planning. In this paper, we present an implementation of the FSPB, by which the most time-consuming parts in the algorithm are parallelized and ported on graphic processing unit (GPU). Compared with the FSPB completely running on central processing unit (CPU), the GPU-implemented FSPB can speed up the dose calculation for 25-35 times on a low price GPU (Geforce GT320) and for 55-100 times on a Tesla C1060, indicating that the GPU-implemented FSPB can provide fast enough dose calculations for real-time treatment planning.

摘要

剂量计算在放射治疗的治疗计划中起着关键作用。剂量计算算法需要高精度和计算效率。有限尺寸笔形束（FSPB）算法是放射治疗治疗计划系统中常用的一种方法。然而，为了实现实时治疗计划等目的，仍需要提高其计算效率。在本文中，我们提出了一种FSPB的实现方法，通过该方法将算法中最耗时的部分进行并行化处理，并移植到图形处理单元（GPU）上。与完全在中央处理器（CPU）上运行的FSPB相比，在低价GPU（Geforce GT320）上，GPU实现的FSPB可以将剂量计算速度提高25至35倍，在Tesla C1060上则可提高55至100倍，这表明GPU实现的FSPB能够为实时治疗计划提供足够快的剂量计算。