Evaluation of state-of-the-art hardware architectures for fast cone-beam CT reconstruction

Scherl, Holger; Kowarschik, Markus; Hofmann, Hannes; Keck, Benjamin; Hornegger, Joachim

doi:10.1016/j.parco.2011.10.004

Cited by 24 publications

(9 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To the point that currently there are many examples of their usage for accelerating scientific applications. For instance, Scherl et al used up to six GPU families, and also field programmable gate arrays, to reduce the reconstruction time. They concluded that the main problem of GPUs resides in the memory bandwidth available to in/out data and the computing resources.…”

Section: Related Workmentioning

confidence: 99%

A comparative study of an X‐ray tomography reconstruction algorithm in accelerated and cloud computing systems

Serrano

Blas

Carretero

2015

Concurrency and Computation

View full text Add to dashboard Cite

SUMMARYWith the increase of resolution in medical image scanners and the need of faster reconstruction methods, new ways of exploiting the inherent parallelism of reconstruction algorithms have arisen. In this paper, we present Mangoose++, an application to perform X-ray computed tomography that supports multiple grades of parallelism. This parallelism is tackled with two different approaches: the usage of parallel nodes with multicore CPUs in a cloud environment and the usage of high-performance computing (HPC)-based parallel architectures such as general-purpose computing on graphics processing unit (GPGPU) or Intel Xeon Phi. In this paper, we show the design and implementation of the application in three types of platforms related to the previous mentioned approaches, comparing and analyzing the performance, resource utilization, and scalability of each platform. Accelerators offer high performance for data sizes that fit inside the accelerator memory. This is the main advantage of Intel Xeon Phi that, in this work, obtains similar performance results than compute unified device architecture (CUDA)-based GPGPU versions, comparing with cards with less memory capacity. In our evaluation experiments, we additionally analyze and discuss the costs and efficiency of Mangoose++ over Amazon Compute Cloud platform, demonstrating that lower times can be achieved in a reasonable price compared with owned HPC-based hardware. A comparison between distinct hardware configurations is provided for emphasizing on the advantages and disadvantages of each one.

show abstract

Section: Related Workmentioning

confidence: 99%

A comparative study of an X‐ray tomography reconstruction algorithm in accelerated and cloud computing systems

Serrano

Blas

Carretero

2015

Concurrency and Computation

View full text Add to dashboard Cite

show abstract

“…Listing 1 Generic FBP algorithm for CL process. A discussion on SIMD-accelerated reconstructions can also be found in [31], [46]. • Another approach to speed up the reconstruction procedure is to organize concurrent computations in the frame of a so-called shared memory model.…”

Section: Implementation Detailsmentioning

confidence: 99%

“…In principle it could also be used for parallel-beam reconstruction tasks with an inclined axis [18], [30] and an impressive number of publications present very efficient FDK implementations [25], [31]- [33]. This algorithm consists of three principal steps which are the cosine weighting (to compensate for the divergent beam), the ramp filtering, and backprojection.…”

Section: Introductionmentioning

confidence: 99%

Efficient Volume Reconstruction for Parallel-Beam Computed Laminography by Filtered Backprojection on Multi-Core Clusters

Myagotin

Voropaev

Helfen

et al. 2013

IEEE Trans. on Image Process.

View full text Add to dashboard Cite

Computed laminography (CL) was developed to use X-rays from synchrotron sources for high-resolution imaging of the internal structure of a flat specimen from a series of 2-D projection images. The projections are acquired by irradiation of the sample under different rotation angles where the object rotation axis is inclined with respect to the beam direction. This yields for laterally extended objects a more uniform average transmitted intensity during sample rotation compared with computed tomography (CT). The reconstruction problem of CL cannot be reduced to a data-efficient 2-D case (as for parallel-beam CT) since each single slice perpendicular to the rotation axis requires a 2-D region on the detector as input data for all projection directions. This paper describes a computationally efficient reconstruction procedure based on filtered backprojection (FBP) adapted to the CL acquisition geometry. From the Fourier slice theorem, we derive a framework for analytic image reconstruction and outline implementation details of the generic FBP algorithm. Different approaches reducing the reconstruction time by means of parallel and distributed computations are considered and evaluated.

show abstract

“…As shown in Fig. 1 [8], the patient or entity is positioned between the X-beam source and a detector.The majority of prior research has applied to accelerate FDK focused on optimizing reconstruction via different accelerators such as Cell-board Engine [9], Field-programmablearray FPGA [10,11] has been widely studied and surveyed [12]. Graphics Processing Units (GPUs) which are considered as massively parallel engines for computation play an important role in reconstruction acceleration [13,14].…”

Section: Introductionmentioning

confidence: 99%

An Acceleration Strategy for Generating Cone-Beam CT Images Based on "Multi-core" Systems

Abbas¹,

Youness²,

Hassan³

2021

Journal of Advanced Engineering Trends

View full text Add to dashboard Cite

Evaluation of state-of-the-art hardware architectures for fast cone-beam CT reconstruction

Cited by 24 publications

References 10 publications

A comparative study of an X‐ray tomography reconstruction algorithm in accelerated and cloud computing systems

A comparative study of an X‐ray tomography reconstruction algorithm in accelerated and cloud computing systems

Efficient Volume Reconstruction for Parallel-Beam Computed Laminography by Filtered Backprojection on Multi-Core Clusters

An Acceleration Strategy for Generating Cone-Beam CT Images Based on "Multi-core" Systems

Contact Info

Product

Resources

About