Fast convolution‐superposition dose calculation on graphics hardware

Hissoiny, Sami; Ozell, Benoı̂t; Després, Philippe

doi:10.1118/1.3120286

Cited by 53 publications

(57 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recently, the rapid development of GPU technology for scientific computing has offered a promising prospect to speed up computationally heavy tasks in radiotherapy. [18][19][20][21][22][23][24][25][26][27][28][29][30][31] Although GPU has been utilized in the TNLM-based 4D-CT reconstruction study, the implementation was not optimal. A new implementation of the algorithm in the 4D-CBCT context is necessary to fully explore the GPU's computational power.…”

Section: Introductionmentioning

confidence: 99%

Four‐dimensional cone beam CT reconstruction and enhancement using a temporal nonlocal means method

et al. 2012

View full text Add to dashboard Cite

Purpose: Four-dimensional cone beam computed tomography (4D-CBCT) has been developed to provide respiratory phase-resolved volumetric imaging in image guided radiation therapy. Conventionally, it is reconstructed by first sorting the x-ray projections into multiple respiratory phase bins according to a breathing signal extracted either from the projection images or some external surrogates, and then reconstructing a 3D CBCT image in each phase bin independently using FDK algorithm. This method requires adequate number of projections for each phase, which can be achieved using a low gantry rotation or multiple gantry rotations. Inadequate number of projections in each phase bin results in low quality 4D-CBCT images with obvious streaking artifacts. 4D-CBCT images at different breathing phases share a lot of redundant information, because they represent the same anatomy captured at slightly different temporal points. Taking this redundancy along the temporal dimension into account can in principle facilitate the reconstruction in the situation of inadequate number of projection images. In this work, the authors propose two novel 4D-CBCT algorithms: an iterative reconstruction algorithm and an enhancement algorithm, utilizing a temporal nonlocal means (TNLM) method. Methods: The authors define a TNLM energy term for a given set of 4D-CBCT images. Minimization of this term favors those 4D-CBCT images such that any anatomical features at one spatial point at one phase can be found in a nearby spatial point at neighboring phases. 4D-CBCT reconstruction is achieved by minimizing a total energy containing a data fidelity term and the TNLM energy term. As for the image enhancement, 4D-CBCT images generated by the FDK algorithm are enhanced by minimizing the TNLM function while keeping the enhanced images close to the FDK results. A forward-backward splitting algorithm and a Gauss-Jacobi iteration method are employed to solve the problems. The algorithms implementation on GPU is designed to avoid redundant and uncoalesced memory access, in order to ensure a high computational efficiency. Our algorithms have been tested on a digital NURBS-based cardiac-torso phantom and a clinical patient case. Results: The reconstruction algorithm and the enhancement algorithm generate visually similar 4D-CBCT images, both better than the FDK results. Quantitative evaluations indicate that, compared with the FDK results, our reconstruction method improves contrast-to-noise-ratio (CNR) by a factor of 2.56-3.13 and our enhancement method increases the CNR by 2.75-3.33 times. The enhancement method also removes over 80% of the streak artifacts from the FDK results. The total computation time is 509-683 s for the reconstruction algorithm and 524-540 s for the enhancement algorithm on an NVIDIA Tesla C1060 GPU card. Conclusions: By innovatively taking the temporal redundancy among 4D-CBCT images into consideration, the proposed algorithms can produce high quality 4D-CBCT images with much less streak artifacts than the FDK results, in the situa...

show abstract

Section: Introductionmentioning

confidence: 99%

Four‐dimensional cone beam CT reconstruction and enhancement using a temporal nonlocal means method

et al. 2012

View full text Add to dashboard Cite

show abstract

“…11) GPU clusters are more affordable and require lighter maintenance than traditional CPU clusters. GPU algorithms have been used for dose calculations by Hissoiny et al 12) and Jacques et al, 13) who implemented superposition convolution algorithms for dose calculations on GPUs. A GPU-based Monte Carlo code for coupled electron-photon transport was implemented by Jia et al, 14) who reported a calculation time reduction up to a factor of 6.6.…”

Section: Introductionmentioning

confidence: 99%

A GPU-Based Track-Repeating Algorithm for Dose Calculation for Photon Radiotherapy

YEPES

2011

Progress in Nuclear Science and Technology

View full text Add to dashboard Cite

An essential ingredient in radiotherapy is the calculation of the dose to be delivered to the patient. Analytical algorithms are commonly used for such a task, however their accuracy is not always satisfactory. Monte Carlo techniques provide higher accuracy, but they often require large computational times. Track-repeating algorithms, for example the Fast Dose Calculator, have shown promise for achieving the accuracy of the Monte Carlo approach for proton radiotherapy dose calculations, while considerably reducing the calculation time. We report on the implementation of the Fast Dose Calculator for photon dose calculations on a GPU architecture. As in the proton case, this implementation reproduces the full Monte Carlo dose calculations within 2%, while achieving a statistical uncertainty of 2% in less than one minute utilizing one single GPU card.

show abstract

“…The acceleration factor provided by general purpose parallel computing on graphical processing units [57][58][59][60][61] has made the method applicable. Following implementation for external beam treatment planning, Oncentra® Brachy v. 4.4 (Nucletron, an Elekta company, Veenendaal, Netherlands) was released in 2013 including the capability for 192 Ir dosimetry calculations based on the CCS method, under the commercial name Oncentra-ACE (Advanced Collapsed cone Engine).…”

Section: Collapsed Cone Superposition: Clinical Implementationmentioning

confidence: 99%

Current state of the art brachytherapy treatment planning dosimetry algorithms

Papagiannis¹,

Pantelis²,

Karaiskos³

2014

BJR

View full text Add to dashboard Cite

Following literature contributions delineating the deficiencies introduced by the approximations of conventional brachytherapy dosimetry, different model-based dosimetry algorithms have been incorporated into commercial systems for 192 Ir brachytherapy treatment planning. The calculation settings of these algorithms are pre-configured according to criteria established by their developers for optimizing computation speed vs accuracy. Their clinical use is hence straightforward. A basic understanding of these algorithms and their limitations is essential, however, for commissioning; detecting differences from conventional algorithms; explaining their origin; assessing their impact; and maintaining global uniformity of clinical practice.Conventional, Task Group (TG)43-based 1 dosimetry marked an improvement over prior dose calculation formalisms for brachytherapy treatment planning by advocating the use of a source strength quantity traceable to international standards, the introduction of two-dimensional (2D) source anisotropy, and global uniformity in source characterization as well as clinical dosimetry practice. In the past decade, brachytherapy has progressed from the traditional surgical paradigm to modern three-dimensional (3D) image-based treatment planning systems (TPSs) and dose delivery. The information available through patient imaging, however, had not been fully exploited since TG43-based dosimetry relies on sourcespecific data pre-calculated in a standard homogeneous water geometry.1-3 Hence, it disregards patient-specific radiation scatter conditions and the radiological differences of tissue or applicator materials from water.In response to literature on the effect of these shortcomings, which has been reviewed in several recent publications, [4][5][6][7] TPSs have become commercially available that include improved dosimetry algorithms, collectively referred to as model-based dosimetry algorithms (MBDCAs). At the time of writing, these include a deterministic solver of the linear Boltzmann transport equation (LBTE) 8-10 and a collapsed cone superposition (CCS) algorithm 11-17 for 192 Ir high-doserate (HDR) applications.This work reviews the basic features of these algorithms and their clinical implementation and presents illustrative results of their performance. Monte Carlo (MC) simulation is also briefly discussed since, besides being a candidate MBDCA for clinical implementation, it is used for obtaining input data for MBDCAs, as well as for their testing. DOSIMETRY IN THE BRACHYTHERAPY PHOTON ENERGY RANGEThe vast majority of brachytherapy sources can be considered pure photon emitters.1-3 Figure 1 readily implies that mainly density heterogeneities affect dosimetry of higher photon energy-emitting sources like 192 Ir, since attenuation and energy absorption per unit mass is comparable for all materials except for bone. On the contrary, attenuation and energy absorption differences exist between different tissue compositions for lower photon energies. Literature on the effect...

show abstract

Fast convolution‐superposition dose calculation on graphics hardware

Cited by 53 publications

References 18 publications

Four‐dimensional cone beam CT reconstruction and enhancement using a temporal nonlocal means method

Four‐dimensional cone beam CT reconstruction and enhancement using a temporal nonlocal means method

A GPU-Based Track-Repeating Algorithm for Dose Calculation for Photon Radiotherapy

Current state of the art brachytherapy treatment planning dosimetry algorithms

Contact Info

Product

Resources

About