CT Image Reconstruction Based on GPUs

Flores, Liubov Alexandrovna; Vidal, Vicente; Mayo, Patricia; Ródenas, F.; Verdú, G.

doi:10.1016/j.procs.2013.05.308

Cited by 18 publications

(15 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Efficiency. Reconstructions of high-quality CT images are extremely computationally demanding tasks (Flores et al, 2013). Recently, image processing using the GPU of a computer has become more popular as the computational performance of GPUs has surpassed that of the CPU with the advent of powerful GPUs for computer games (Eklund et al, 2013).…”

Section: Software Parameters For Image Calibrations and Artifact Removalmentioning

confidence: 99%

“…While there are many different reconstruction algorithms, most can be sub-classified into two main categories; filtered back-projection (FBP) and algebraic iterative algorithms (Flores et al, 2013). FBP methods are based on analytical algorithms using the inverse Fourier transform on equally spaced projections (Flores et al, 2013) while algebraic iterative methods fit predictor models to data through iterations (Deá k et al, 2013) and do not require evenly spaced projections.…”

Section: Introductionmentioning

confidence: 99%

“…FBP methods are based on analytical algorithms using the inverse Fourier transform on equally spaced projections (Flores et al, 2013) while algebraic iterative methods fit predictor models to data through iterations (Deá k et al, 2013) and do not require evenly spaced projections. FBP has long been the standard reconstruction method running on a central processing unit (CPU); however, with the advent of third-generation synchrotron sources and new detectors, larger datasets significantly increase reconstruction times making new high-performance computing systems mandatory (Marone & Stampanoni, 2012).…”

Section: Introductionmentioning

confidence: 99%

“…FBP has long been the standard reconstruction method running on a central processing unit (CPU); however, with the advent of third-generation synchrotron sources and new detectors, larger datasets significantly increase reconstruction times making new high-performance computing systems mandatory (Marone & Stampanoni, 2012). Many software programs are now making use of graphics processing units (GPUs) to increase efficiency for these demanding algorithms or for large datasets for tomographic reconstructions (Flores et al, 2013). Use of a GPU can significantly reduce the data processing time.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Assessment of freeware programs for the reconstruction of tomography datasets obtained with a monochromatic synchrotron-based X-ray source

Wolkowski

Snead

Wesolowski

et al. 2015

J Synchrotron Radiat

View full text Add to dashboard Cite

Synchrotron-based in-line phase-contrast computed tomography (PC-CT) allows soft tissue to be imaged with sub-gross resolution and has potential to be used as a diagnostic tool. The reconstruction and processing of in-line PC-CT datasets is a computationally demanding task; thus, an efficient and user-friendly software program is desirable. Four freeware programs (NRecon, PITRE, H-PITRE and Athabasca Recon) were compared for the availability of features such as dark-and flat-field calibration, beam power normalization, ring artifact removal, and alignment tools for optimizing image quality. An in-line PC-CT projection dataset (3751 projections, 180 rotation, 10.13 mm Â 0.54 mm) was collected from a formalin-fixed canine prostate at the Biomedical Imaging and Therapy Bending Magnet (BMIT-BM) beamline of the Canadian Light Source. This dataset was processed with each of the four software programs and usability of the program was evaluated. Efficiency was assessed by how each program maximized computer processing power during computation. Athabasca Recon had the least-efficient memory usage, least user-friendly interface, and lacked a ring artifact removal feature. NRecon, PITRE and H-PITRE produced similar quality images, but the Athabasca Recon reconstruction suffered from the lack of a native ring remover algorithm. The 64-bit version of NRecon uses GPU (graphics processor unit) memory for accelerated processing and is userfriendly, but does not provide necessary parameters for in-line PC-CT data, such as dark-field and flat-field correction and beam power normalization. PITRE has many helpful features and tools, but lacks a comprehensive user manual and help section. H-PITRE is a condensed version of PITRE and maximizes computer memory for efficiency. To conclude, NRecon has fewer imaging processing tools than PITRE and H-PITRE, but is ideal for less experienced users due to a simple user interface. Based on the quality of reconstructed images, efficient use of computer memory and parameter availability, H-PITRE was the preferred of the four programs compared.

show abstract

Section: Software Parameters For Image Calibrations and Artifact Removalmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Assessment of freeware programs for the reconstruction of tomography datasets obtained with a monochromatic synchrotron-based X-ray source

Wolkowski

Snead

Wesolowski

et al. 2015

J Synchrotron Radiat

View full text Add to dashboard Cite

show abstract

“…The cost per iteration increases quadratically with the size and number of input sinograms, and the slow convergence of the iterative approach requires many of those expensive iterations. There are some more efficient approaches that implement algebraic methods by using graphic processing units (GPUs) [14,15]. In addition, the number of the iterations in the algebraic approaches is a tuning parameter, which typically determines how much detailed features are included in the resulting reconstruction; a too large choice would give a noisy reconstruction, and a too small choice would not give a reconstruction with many missing details.…”

Section: Introductionmentioning

confidence: 99%

Sparse filtered SIRT for electron tomography

Park

2020

Pattern Recognition

View full text Add to dashboard Cite

Electron tomographic reconstruction is a method for obtaining a threedimensional image of a specimen with a series of two dimensional microscope images taken from different viewing angles. Filtered backprojection, one of the most popular tomographic reconstruction methods, does not work well under the existence of image noises and missing wedges. This paper presents a new approach to largely mitigate the effect of noises and missing wedges. We propose a novel filtered backprojection that optimizes the filter of the backprojection operator in terms of a reconstruction error. This data-dependent filter adaptively chooses the spectral domains of signals and noises, suppressing the noise frequency bands, so it is very effective in denoising. We also propose the new filtered backprojection embedded within the simultaneous iterative reconstruction iteration for mitigating the effect of missing wedges. Our numerical study is presented to show the performance gain of the proposed approach over the state-of-the-art.

show abstract

Accelerating iterative CT reconstruction algorithms using Tensor Cores

Nourazar

Goossens

2021

J Real-Time Image Proc

View full text Add to dashboard Cite

Tensor Cores are specialized hardware units added to recent NVIDIA GPUs to speed up matrix multiplication-related tasks, such as convolutions and densely connected layers in neural networks. Due to their specific hardware implementation and programming model, Tensor Cores cannot be straightforwardly applied to other applications outside machine learning. In this paper, we demonstrate the feasibility of using NVIDIA Tensor Cores for the acceleration of a non-machine learning application: iterative Computed Tomography (CT) reconstruction. For large CT images and real-time CT scanning, the reconstruction time for many existing iterative reconstruction methods is relatively high, ranging from seconds to minutes, depending on the size of the image. Therefore, CT reconstruction is an application area that could potentially benefit from Tensor Core hardware acceleration. We first studied the reconstruction algorithm’s performance as a function of the hardware related parameters and proposed an approach to accelerate reconstruction on Tensor Cores. The results show that the proposed method provides about 5 $$\times $$ × increase in speed and energy saving using the NVIDIA RTX 2080 Ti GPU for the parallel projection of 32 images of size $$512\times 512$$ 512 × 512 . The relative reconstruction error due to the mixed-precision computations was almost equal to the error of single-precision (32-bit) floating-point computations. We then presented an approach for real-time and memory-limited applications by exploiting the symmetry of the system (i.e., the acquisition geometry). As the proposed approach is based on the conjugate gradient method, it can be generalized to extend its application to many research and industrial fields.

show abstract

CT Image Reconstruction Based on GPUs

Cited by 18 publications

References 12 publications

Assessment of freeware programs for the reconstruction of tomography datasets obtained with a monochromatic synchrotron-based X-ray source

Assessment of freeware programs for the reconstruction of tomography datasets obtained with a monochromatic synchrotron-based X-ray source

Sparse filtered SIRT for electron tomography

Accelerating iterative CT reconstruction algorithms using Tensor Cores

Contact Info

Product

Resources

About