基于GPU的蒙特卡洛放疗剂量模拟

目的:蒙特卡洛模拟被认为是目前剂量计算方面最为精确的算法,但是因为其模拟时间过长,在临床应用上受到限制。EGSnrc作为目前在医学物理领域应用最为广泛的蒙特卡洛模拟软件,因为其过长的执行时间,其在临床方面的应用受到很大限制。为了克服这一障碍,我们开发了一个基于GPU的蒙特卡洛模拟程序,以期为放疗计量提供一个高效和低成本的蒙卡程序。方法:本文给出了一种基于GPU(Graphic Processing Unit)的蒙特卡洛模拟的新方法,开发语言是CUDA 5.0,将目前最为通用的蒙特卡洛程序EGSnrc移植到GPU平台,保留EGSnrc的核心物理过程以及输运过程的算法,这可以在最大限度保持原来EGSnrc模拟精度的前提下,极大地提高蒙特卡洛模拟的效率。GPU版本的蒙特卡洛模拟程序运行在一块英伟达Tesla C2050显卡上。GPU版本的EGSnrc精度的验证采用了纯水模体,同时,入射的射线我们选择为6 MV的光子。为了进一步检验GPU版本的EGSnrc的精度,我们进行了一个逐体素的检验,检验结果显示,GPU版本的EGSnrc和EGSnrc符合的很好。结果:最终实验结果表明,在模拟20亿个相空间事例的情况下,使用NVIDIA Tesla C2050显卡,新的基于GPU的蒙特卡洛程序的速度比在单核的Intel Xeon 2.0 GHz CPU上的模拟速度提高了43倍,且其精度与EGSnrc的精度相当。计算结果的方差在高剂量区域(D>Dmax)小于0.5%,计量误差经过Dmax归一化之后,其和EGSnrc的误差小于1%的比率在占整个区域的90%以上。结论:通过此新程序表明,基于GPU的蒙特卡洛算法可以极大地提高蒙特卡洛程序的运行效率,与此同时,GPU版本的EGSnrc在最大程度上保持了EGSnrc的模拟精度。考虑到GPU版本的EGSnrc程序的速度以及精度优势,其在未来的临床应用中有着巨大的前景。     

基于GPU的蒙特卡洛模拟在辐射剂量计算中的研究进展及应用

蒙特卡洛(MC)算法在辐射剂量计算中发挥着重要作用,但是计算速度限制了其在临床中的应用。随着图形处理器(GPU)技术的发展,GPU并行加速方法被越来越多地应用到MC计算中。本文主要介绍基于GPU的MC在光子、电子和质子输运模拟过程在辐射剂量计算方面的研究进展及其在医学物理上的应用。     

用CUDA实现放射治疗中剂量的快速计算

剂量计算是放射治疗计划系统的关键技术之一,它既要有较高的计算精度又要有较快的计算速度。有限笔束(FSPB)算法是目前放射治疗计划系统大多采用的光子线剂量计算算法,其计算速度尚不能达到实时治疗计划程度。本文采用图形处理器(GPU),对FSPB算法中最耗时的部分实现了基于GPU并行化计算,与基于中央处理器(CPU)的计算相比,在中低端GPU(Geforce GT320)上,剂量计算速度提高可达25~35倍,在较高端GPU(TeslaC1060)上计算速度提高可达55~100倍,计算效率完全可用于实时治疗计划中的剂量计算。     

基于GPU的蒙特卡洛放疗剂量并行计算

目的：蒙特卡洛模拟在放疗剂量计算领域被广泛视为最精确的计算方法,但对于日常的临床应用,其效率仍有较大提升需求和空间。方法：本文会呈现放疗剂量计算领域的最新成果－维持相同的粒子输运原理的同时,使用CUDA语言,利用显卡的GPU（GraphicProcessingUnit）并行处理蒙特卡洛计算中的主要过程,计算光子剂量沉积。这样既可以保证不失去蒙卡模拟的精度．又可以极大地提高运算速度。结果：实践表明在使用NVIDIAGTX4601GDDR5plusINTELi52300的硬件设备,在GPU上并行计算蒙特卡洛放疗剂量沉积时．计算100万个光子剂量沉积时加速因子达到116．6,处理1000万光子入射,加速因子可达127．5。结论：本文中利用显卡GPU运行CUDA语言对放疗剂量计算进行模拟,是一种可以大幅有效提高剂量计算效率方法。     

一种医学图像去噪程序的并行优化

目的：在采集、处理和传输过程中,医学图像会存在各种噪声,严重影响医学图像的质量和后续对图像的各种处理,因此医学图像去噪具有重要意义。同时医学图像数据量大,去噪处理算法复杂,在一般个人电脑上进行医学图像去噪仍是一个非常耗时的过程．很难满足实际应用中高实时性的要求．因此需要通过优化来提高去噪的处理速度。方法：本文利用CUDA（Compute Unified Device Architecture）并行编程对基于同质算法的三维医学图像去噪进行加速,CPU和GPU（Graphic ProcessorUnit）异构编程方式能发挥GPU高强度的计算能力,提高算法的执行速度。通过使用纹理存储器将图像数据与纹理绑定,优化存储器访问,提高数据访问速度。优化过程中,合理选择三维图像数据的分块方式和线程块维度。可以获得更快的加速。结果：与基于同质的matlab和CPU去噪程序相比,并行优化后GPU程序在保持去噪图像质量的前提卞可以达到几百倍的加速。结论：CUDA加速大大缩短了三维医学图像去噪的运行时间,解决了医学图像去噪的速度瓶颈问题．可以应用于对运行速度有要求的图像处理中。     

一种基于GPU的体积CT快速重建算法

为了解决体积CT图像重建时间较长问题,提出了一种适合于医学临床应用的快速重建算法。首先,提出了一种基于图形处理器(GPU)的体积CT图像重建方式,利用GPU强大的并行和浮点运算能力进行计算效能的提升。其次,将体积CT图像重建中的几何运算与像素运算分离,减少了重复运算,进一步提高了计算效率。最后,基于医学应用背景,算法中实现了体积CT扫描和重建的并行化的思想。结果表明,利用上述的快速算法,在普通计算机硬件平台上即可实现重建时间减少70倍以上。     

三维医学图像核回归算法的GPU加速研究

【摘 要】 核回归理论被广泛应用于医学图像处理和医学图像重建领域,并取得了十分显著的效果。它包括传统核回归方法（CKR）和控制核回归方法（SKR）。三维SKR算法比三维CKR算法具有更优的去噪效果和边缘保持效果,但三维SKR算法的计算量过于庞大且复杂,使其应用领域受到限制。目前,医学图像重建使用的是基于GPU的三维CKR算法,所以基于GPU的三维SKR算法的实现是一项有研究价值且具有挑战性的工作。本文首先优化三维SKR算法的计算过程,然后利用GPU进行CUDA编程实现三维SKR并行加速算法。实验表明,基于GPU的三维SKR算法与基于CPU单线程三维SKR算法相比能获得约244.9~246.3倍的加速比,与基于CPU多线程三维SKR算法相比能获得约123.0~137.4倍的加速比。     

三维医学图像的交互式图形处理器快速重建

目的:开发一种基于图形处理器(GPU)的医学三维图像交互式重建系统,用于临床辅助诊断、手术计划等领域。方法:在GPU重建算法基础上使用了八叉树空间结构和多边形辅助光线投射方法实现进一步的优化,分别用基础算法和优化后的算法对一组CT图像进行重建,验证优化效果。结果:本研究实现的优化算法在真实医学三维图像重建中得到了高质量的重建结果,并且比原有的基于GPU的重建算法快2～3倍。结论:本研究实现的三维重建系统能有效加快重建速度,实现交互式快速重建。     

基于GPU的医学图像快速体切割算法

目的:在体绘制过程中,有许多重要的、医生感兴趣的细节信息隐藏在数据场内部,在进行显示时,这部分信息很容易被其他组织或器官遮挡,无法显示在重建图像中,为了给医生提供全面、直观和准确的诊断信息,本文提出了一种基于GPU加速的体切割算法。方法:通过将切割算法和基于GPU的光线投射算法结合,实现体数据的快速切割。本文在基于GPU加速的医学图像快速体绘制的基础上,将剖面的空间信息传入着色器,然后通过比较体数据的空间坐标与剖面位置的关系来决定体数据的取舍。该方法不同于以往基于深度模板信息的体切割,在定义好切割平面后,可从任意角度对保留下来的有效体数据的重建结果进行观察。结果:该方法能够精确地按照用户定义的形状对体数据进行切割,并且由于使用了硬件的加速功能,该方法可以达到实时交互的速度。结论:该方法能够满足医学影像可视化的实时交互要求,在手术模拟等临床技术中有广泛应用。     

剂量分布比较中γ因子的快速计算方法

γ比较方法作为放射治疗剂量学验证中的一种手段,现在已经在科研和临床的剂量分布比较中得到广泛应用。但是,在比较三维剂量分布时,γ因子的计算量大,需要花费大量的时间。本文采用一种预先排序技术和基于图形处理器(GPU)的并行计算技术结合,实现了γ因子的快速计算。通过7对剂量分布的测试,基于GPU的γ因子的计算速度提高了几十倍,而且与CPU相比保持了相同的计算精度。实验结果表明,利用GPU的并行计算对γ比较方法进行加速是切实有效的。     

GPU-based fast Monte Carlo simulation for radiotherapy dose calculation

Monte Carlo (MC) simulation is commonly considered to be the most accurate dose calculation method in radiotherapy. However, its efficiency still requires improvement for many routine clinical applications. In this paper, we present our recent progress toward the development of a graphics processing unit (GPU)-based MC dose calculation package, gDPM v2.0. It utilizes the parallel computation ability of a GPU to achieve high efficiency, while maintaining the same particle transport physics as in the original dose planning method (DPM) code and hence the same level of simulation accuracy. In GPU computing, divergence of execution paths between threads can considerably reduce the efficiency. Since photons and electrons undergo different physics and hence attain different execution paths, we use a simulation scheme where photon transport and electron transport are separated to partially relieve the thread divergence issue. A high-performance random number generator and a hardware linear interpolation are also utilized. We have also developed various components to handle the fluence map and linac geometry, so that gDPM can be used to compute dose distributions for realistic IMRT or VMAT treatment plans. Our gDPM package is tested for its accuracy and efficiency in both phantoms and realistic patient cases. In all cases, the average relative uncertainties are less than 1%. A statistical t-test is performed and the dose difference between the CPU and the GPU results is not found to be statistically significant in over 96% of the high dose region and over 97% of the entire region. Speed-up factors of 69.1 ～ 87.2 have been observed using an NVIDIA Tesla C2050 GPU card against a 2.27 GHz Intel Xeon CPU processor. For realistic IMRT and VMAT plans, MC dose calculation can be completed with less than 1% standard deviation in 36.1 ～ 39.6 s using gDPM.     

Fast perspective volume ray casting method using GPU-based acceleration techniques for translucency rendering in 3D endoluminal CT colonography

Recent advances in graphics processing unit (GPU) have enabled direct volume rendering at interactive rates. However, although perspective volume rendering for opaque isosurface is rapidly performed using conventional GPU-based method, perspective volume rendering for non-opaque volume such as translucency rendering is still slow. In this paper, we propose an efficient GPU-based acceleration technique of fast perspective volume ray casting for translucency rendering in computed tomography (CT) colonography. The empty space searching step is separated from the shading and compositing steps, and they are divided into separate processing passes in the GPU. Using this multi-pass acceleration, empty space leaping is performed exactly at the voxel level rather than at the block level, so that the efficiency of empty space leaping is maximized for colon data set, which has many curved or narrow regions. In addition, the numbers of shading and compositing steps are fixed, and additional empty space leapings between colon walls are performed to increase computational efficiency further near the haustral folds. Experiments were performed to illustrate the efficiency of the proposed scheme compared with the conventional GPU-based method, which has been known to be the fastest algorithm. The experimental results showed that the rendering speed of our method was 7.72 fps for translucency rendering of 1024×1024 colonoscopy image, which was about 3.54 times faster than that of the conventional method. Since our method performed the fully optimized empty space leaping for any kind of colon inner shapes, the frame-rate variations of our method were about two times smaller than that of the conventional method to guarantee smooth navigation. The proposed method could be successfully applied to help diagnose colon cancer using translucency rendering in virtual colonoscopy.     

CT/MRI混合配准方法及其在放疗计划系统中的应用

旨在研究放疗中图像配准方法,特别是针对放疗中常用的CT、MRI,提出基于混合框架的配准方法,该方法主要包 括两个方面:（1）采用掩膜（Mask）提取感兴趣区域、形态学运算等图像处理方法以及CPU多线程并行技术,大幅度提高配 准速度;（2）采用由全局到局部的混合配准策略,首先利用基于仿射变换的刚性配准整体配准,以防止图像间偏差过大,在 此基础上针对感兴趣区域采用B样条弹性配准,调整局部形变。通过实验表明,采用预处理及加速策略的刚性配准,在保 持其精度的情况下,提速比可达10倍,测试结果已达到临床需求;此外,采用基于GPU加速的混合配准策略,其配准速度 提至约4 min。     

Fast Monte Carlo simulation for patient-specific CT/CBCT imaging dose calculation

X-ray imaging dose from computed tomography (CT) or cone beam CT (CBCT) scans has become a serious concern. Patient-specific imaging dose calculation has been proposed for the purpose of dose management. While Monte Carlo (MC) dose calculation can be quite accurate for this purpose, it suffers from low computational efficiency. In response to this problem, we have successfully developed a MC dose calculation code, gCTD, on GPU architecture under the NVIDIA CUDA platform for fast and accurate estimation of the x-ray imaging dose received by a patient during a CT or CBCT scan. Techniques have been developed particularly for the GPU architecture to achieve high computational efficiency. Dose calculations using CBCT scanning geometry in a homogeneous water phantom and a heterogeneous Zubal head phantom have shown good agreement between gCTD and EGSnrc, indicating the accuracy of our code. In terms of improved efficiency, it is found that gCTD attains a speed-up of ～400 times in the homogeneous water phantom and ～76.6 times in the Zubal phantom compared to EGSnrc. As for absolute computation time, imaging dose calculation for the Zubal phantom can be accomplished in ～17 s with the average relative standard deviation of 0.4%. Though our gCTD code has been developed and tested in the context of CBCT scans, with simple modification of geometry it can be used for assessing imaging dose in CT scans as well.     

立体定向粒子植入治疗计划系统的开发及其关键技术研究

目的设计一套适用于立体定向放射性粒子植入肿瘤治疗手术的计划系统。方法采用线性布源原则,设计结合立体定向手术方式的粒子植入治疗计划设计模块;采用基于图形处理单元的粒子剂量快速计算方法,进行放射剂量场的实时计算和更新;研究术前MRI图像和术后CT图像的自动配准和融合方法,实现术后剂量的定量验证,并提出一种基于体积重合率的剂量比较方法。结果实现了一套设计合理、功能完善的神经外科放射性粒子植入治疗计划系统软件。经临床试用,完全满足临床需求。结论该系统能够有效地辅助医生进行放射性粒子植入手术的术前治疗计划设计、优化和术后的放射剂量验证,具有重要的临床价值和广泛的应用前景。     

Using graphics processing units to accelerate perturbation Monte Carlo simulation in a turbid medium

We report a fast perturbation Monte Carlo (PMC) algorithm accelerated by graphics processing units (GPU). The two-step PMC simulation [Opt. Lett. 36, 2095 (2011)] is performed by storing the seeds instead of the photon's trajectory, and thus the requirement in computer random-access memory (RAM) becomes minimal. The two-step PMC is extremely suitable for implementation onto GPU. In a standard simulation of spatially-resolved photon migration in the turbid media, the acceleration ratio between using GPU and using conventional CPU is about 1000. Furthermore, since in the two-step PMC algorithm one records the effective seeds, which is associated to the photon that reaches a region of interest in this letter, and then re-run the MC simulation based on the recorded effective seeds, radiative transfer equation (RTE) can be solved by two-step PMC not only with an arbitrary change in the absorption coefficient, but also with large change in the scattering coefficient.     

Fast Monte Carlo simulations of ultrasound-modulated light using a graphics processing unit

Ultrasound-modulated optical tomography (UOT) is based on "tagging" light in turbid media with focused ultrasound. In comparison to diffuse optical imaging, UOT can potentially offer a better spatial resolution. The existing Monte Carlo (MC) model for simulating ultrasound-modulated light is central processing unit (CPU) based and has been employed in several UOT related studies. We reimplemented the MC model with a graphics processing unit [(GPU), Nvidia GeForce 9800] that can execute the algorithm up to 125 times faster than its CPU (Intel Core Quad) counterpart for a particular set of optical and acoustic parameters. We also show that the incorporation of ultrasound propagation in photon migration modeling increases the computational time considerably, by a factor of at least 6, in one case, even with a GPU. With slight adjustment to the code, MC simulations were also performed to demonstrate the effect of ultrasonic modulation on the speckle pattern generated by the light model (available as animation). This was computed in 4 s with our GPU implementation as compared to 290 s using the CPU.     

Speed versus accuracy in a fast convolution photon dose calculation for conformal radiotherapy

A convolution dose calculation for megavoltage photon beams is described and the compromise between speed and accuracy examined. The algorithm is suitable for treatment planning optimization, where the need is for a fast, flexible method requiring minimal beam data but providing an accurate result. The algorithm uses a simple tabular beam model, together with a discrete scatter kernel. These beam parameters are fitted either to a measured dose distribution, or to a dose distribution calculated using a more accurate dose calculation algorithm. The calculation is then applied to pelvic and thoracic conformal plans, and the results compared with those provided by a commercial radiotherapy treatment planning system (Pinnacle3, Philips Radiation Oncology Systems, Milpitas, CA), which has been verified against measurements. The calculation takes around 4 s to compute a 100 x 100 mm field, and agreement of the dose-volume histograms with the commercial treatment planning system is to within 5% dose or 8% volume. Use of a grid resolution coarser than 5 x 5 x 5 mm is found to be inaccurate, whereas calculating primary dose on a coarse grid and interpolating is found to increase speed without significantly reducing accuracy. Kernel resolution influences the speed and accuracy, but using 12 discrete points provides a fast result with a limited error. Thus, the algorithm is suitable for optimization applications.