Fast analytical scatter estimation using graphics processing 			 units

Ingleby, Harry; Lippuner, Jonas; Rickey, Daniel W.; Li, Yue; Elbakri, Idris A.

doi:10.3233/xst-150475

Cited by 6 publications

(6 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Since the majority of calculation time is spent estimating the singly and multiply scattered components based on the large number of scatter centers, and the phase‐space information of all the interaction centers is known, such a calculation can be potentially completed by parallel computing using, for example, GPU (graphics processing units) parallelism. The GPU parallelism with a single NVIDIA 9800 GX2 was applied for fast analytical calculation for a singly scattered fluence map in low energy KV imaging, and it accomplished a 32 3 voxel calculation in 4.3 s. 18 The GPU in that earlier work had only 128 cores. In the current market, GPUs with over 4000 cores are available at low cost, and therefore we expect reprogramming the TH method to take advantage of GPU processing will significantly accelerate the calculation (to about 1 s with 4000 cores).…”

Section: Discussionmentioning

confidence: 99%

Performance optimization of a tri‐hybrid method for estimation of patient scatter into the EPID

Guo

Ingleby

Uytven

et al. 2021

J Applied Clin Med Phys

Self Cite

View full text Add to dashboard Cite

On‐treatment EPID images are contaminated with patient‐generated scattered photons. If this component can be accurately estimated, its effect can be removed, and therefore a corresponding in vivo patient dose estimate will be more accurate. Our group previously developed a "tri‐hybrid" (TH) algorithm to provide fast but accurate estimates of patient‐generated photon scatter. The algorithm uses an analytical method to solve for singly‐scattered photon fluence, a modified Monte Carlo hybrid method to solve for multiply‐scattered photon fluence, and a pencil beam scatter kernel method to solve for electron interaction generated scattered photon fluence. However, for efficient clinical implementation, spatial and energy sampling must be optimized for speed while maintaining overall accuracy. In this work, the most significant sampling issues were examined, including spatial sampling settings for the patient voxel size, the number of Monte Carlo histories used in the modified hybrid MC method, scatter order sampling for the hybrid method, and also a range of energy spectrum sampling (i.e., energy bin sizes). The total predicted patient‐scattered photon fluence entering the EPID was compared with full MC simulation (EGSnrc) for validation. Three phantoms were tested with 6 and 18 MV beam energies, field sizes of 4 × 4, 10 × 10, and 20 × 20 cm2, and source‐to‐imager distance of 140 cm to develop a set of optimal sampling settings. With the recommended sampling, accuracy and precision of the total‐scattered energy fluence of the TH patient scatter prediction method are within 0.9% and 1.2%, respectively, for all test cases compared with full MC simulation results. For the mean energy spectrum across the imaging plane, comparison of TH with full MC simulation showed 95% overlap. This study has optimized sampling settings so that they have minimal impact on patient scatter prediction accuracy while maintaining maximum execution speed, a critical step for future clinical implementation.

show abstract

Section: Discussionmentioning

confidence: 99%

Performance optimization of a tri‐hybrid method for estimation of patient scatter into the EPID

Guo

Ingleby

Uytven

et al. 2021

J Applied Clin Med Phys

Self Cite

View full text Add to dashboard Cite

show abstract

“…graphics processing unit parallelism). For example, GPU parallelism with a single NVIDIA 9800 GX2 (circa 2009) was applied for analytical photon scatter calculations in KV imaging, and completed a 32 3 voxel calculation in 4.3 s (Ingleby et al 2015).…”

Section: Discussionmentioning

confidence: 99%

“…bremsstrahlung and positron annihilation). Several groups have used analytical methods to estimate the SS energy fluence (Poletti et al 2002, Kyriakou et al 2008, Star-Lack et al 2009, Lippuner et al 2011, Ingleby et al 2015. The MS component is known to be a smooth, broad function and has been treated as proportional to the SS photon distribution (Yao and Leszczynski 2009).…”

Section: Introductionmentioning

confidence: 99%

A tri-hybrid method to estimate the patient-generated scattered photon fluence components to the EPID image plane

et al. 2020

Self Cite

View full text Add to dashboard Cite

In vivo dosimetry methods can verify the prescription dose is delivered to the patient during treatment. Unfortunately, in exit dosimetry, the megavoltage image is contaminated with patient-generated scattered photons. However, estimation and removal of the effect of this fluence improves accuracy of in vivo dosimetry methods. This work develops a ‘tri-hybrid’ algorithm combining analytical, Monte Carlo (MC) and pencil-beam scatter kernel methods to provide accurate estimates of the total patient-generated scattered photon fluence entering the MV imager. For the multiply-scattered photon fluence, a modified MC simulation method was applied, using only a few histories. From each second- and higher-order interaction site in the simulation, energy fluence entering all pixels of the imager was calculated using analytical methods. For photon fluence generated by electron interactions in the patient (i.e. bremsstrahlung and positron annihilation), a convolution/superposition approach was employed using pencil-beam scatter fluence kernels as a function of patient thickness and air gap distance, superposed on the incident fluence distribution. The total patient-scattered photon fluence entering the imager was compared with a corresponding full MC simulation (EGSnrc) for several test cases. These included three geometric phantoms (water, half-water/half-lung, computed tomography thorax) using monoenergetic (1.5, 5.5 and 12.5 MeV) and polyenergetic (6 and 18 MV) photon beams, 10 × 10 cm2 field, source-to-surface distance 100 cm, source-to-imager distance 150 cm and 40 × 40 cm2 imager. The proposed tri-hybrid method is demonstrated to agree well with full MC simulation, with the average fluence differences and standard deviations found to be within 0.5% and 1%, respectively, for test cases examined here. The method, as implemented here with a single CPU (non-parallelized), takes ∼80 s, which is considerably shorter compared to full MC simulation (∼30 h). This is a promising method for fast yet accurate calculation of patient-scattered fluence at the imaging plane for in vivo dosimetry applications.

show abstract

“…Besides, with an average runtime of 100 ms, it was more than twice as fast as the U-Net with a runtime of 240 ms (CPU, Intel(R) i7-8850H), suggesting that it can be used in a clinical setup once current limitations are solved. Promising counter-measures include the incorporation of more prior knowledge, such as deriving scatter probabilities based on the patient shape model, or the combination with a first-order scatter estimation algorithm [8,11,30].…”

Section: Methodsmentioning

confidence: 99%

Simultaneous Estimation of X-ray Back-Scatter and Forward-Scatter using Multi-Task Learning

Roser¹,

Zhong²,

Birkhold³

et al. 2020

Preprint

View full text Add to dashboard Cite

Scattered radiation is a major concern impacting X-ray imageguided procedures in two ways. First, back-scatter significantly contributes to patient (skin) dose during complicated interventions. Second, forward-scattered radiation reduces contrast in projection images and introduces artifacts in 3-D reconstructions. While conventionally employed anti-scatter grids improve image quality by blocking X-rays, the additional attenuation due to the anti-scatter grid at the detector needs to be compensated for by a higher patient entrance dose. This also increases the room dose affecting the staff caring for the patient. For skin dose quantification, back-scatter is usually accounted for by applying pre-determined scalar back-scatter factors or linear point spread functions to a primary kerma forward projection onto a patient surface point. However, as patients come in different shapes, the generalization of conventional methods is limited. Here, we propose a novel approach combining conventional techniques with learning-based methods to simultaneously estimate the forward-scatter reaching the detector as well as the back-scatter affecting the patient skin dose. Knowing the forwardscatter, we can correct X-ray projections, while a good estimate of the back-scatter component facilitates an improved skin dose assessment. To simultaneously estimate forward-scatter as well as back-scatter, we propose a multi-task approach for joint back-and forward-scatter estimation by combining X-ray physics with neural networks. We show that, in theory, highly accurate scatter estimation in both cases is possible. In addition, we identify research directions for our multi-task framework and learning-based scatter estimation in general.

show abstract

Fast analytical scatter estimation using graphics processing units

Cited by 6 publications

References 25 publications

Performance optimization of a tri‐hybrid method for estimation of patient scatter into the EPID

Performance optimization of a tri‐hybrid method for estimation of patient scatter into the EPID

A tri-hybrid method to estimate the patient-generated scattered photon fluence components to the EPID image plane

Simultaneous Estimation of X-ray Back-Scatter and Forward-Scatter using Multi-Task Learning

Contact Info

Product

Resources

About