Parallelization and Runtime Prediction of the ListMode OSEM Algorithm for 3D PET Reconstruction

Schellmann, Maraike; Kösters, Thomas; Gorlatch, Sergei

doi:10.1109/nssmic.2006.354349

Cited by 7 publications

(7 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Different computer architectures coupled with dedicated memory access strategies are used to accelerate the BP step of an analytic or iterative reconstruction, including: general purpose processors [5], [6], [7], graphical processors [8], [9], [10], [11], [12], [13], [14], the Cell processor [4], [15] or ASIC/FPGA architectures [16], [17], [18], [19], [20]. While most of these works have investigated cone beam BP, only a few of them have investigated 3D parallel beam BP [2], [5], [8], [9].…”

Section: B Acceleration Of Reconstructionmentioning

confidence: 99%

“…The parallelisation of reconstruction algorithms on shared memory parallel general purpose computer [5] stays efficient only up to 4 processing units, because of conflictual accesses on the memory bus. Considering clusters of heterogeneous PCs [6], [7], efficiency of parallelisation drops down quickly because of the costly communication between PCs.…”

Section: B Acceleration Of Reconstructionmentioning

confidence: 99%

See 1 more Smart Citation

High Speed 3D Tomography on CPU, GPU, and FPGA

et al. 2009

View full text Add to dashboard Cite

Back-Projection (BP) is a costly computational step in tomography image reconstruction such as Positron Emission Tomography (PET). To reduce the computation time, this paper presents a Pipelined, Pre-fetch and Parallelized Architecture for PET BP (3PA-PET).The key feature of this architecture is its original memory access strategy, masking the high latency of the external memory. Indeed, the pattern of the memory references to the data acquired hinder the processing unit.The memory access bottleneck is overcome by an efficient use of the intrinsic temporal and spatial locality of the BP algorithm. A loop reordering allows an efficient use of general purpose processor's caches, for software implementation, as well as the 3D Predictive and Adaptive Cache (3D-AP Cache), when considering hardware implementations. Parallel hardware pipelines are also efficient thanks to a hierarchical 3D-AP Cache: each pipeline performs a memory reference in about one clock cycle to reach a computational throughput close to 100%.The 3PA-PET architecture is prototyped on a System on Programmable Chip (SoPC) to validate the system and to measure its expected performances. Time performances are compared with a desktop PC, a workstation and a GPU (Graphic Processor Unit).

show abstract

Section: B Acceleration Of Reconstructionmentioning

confidence: 99%

Section: B Acceleration Of Reconstructionmentioning

confidence: 99%

High Speed 3D Tomography on CPU, GPU, and FPGA

et al. 2009

View full text Add to dashboard Cite

show abstract

“…The challenge is to combine this (and possibly other features) with methods of parallelization like in [6].…”

Section: The Issue Of Implementationmentioning

confidence: 99%

Scatter Correction in PET Using the Transport Equation

Kösters

Natterer

Wübbeling

2006

2006 IEEE Nuclear Science Symposium Conference Record

Self Cite

View full text Add to dashboard Cite

In the presence of a strong scatterer, it is inappropriate to perform scatter correction by subtracting scattered from measured events and work on unscattered events only. Take, for example, a strong scatterer that completely surrounds the reconstruction volume. In this case, all events are scattered and removal simply would leave nothing. We propose an algorithm for PET reconstruction that is based on modeling the scanning process by the full transport equation, incorporating (multiple) scatter and attenuation. Then, a solution to the full forward model is sought. The concept sounds simple, but it is hard to implement since the forward model becomes arbitrarily complex. Also, care has to be taken to use expectation maximization techniques to retain image quality.

show abstract

“…There are several implementations of reconstruction systems to speed up its computation : PC clusters 1, 2 , DSP 3 , multi-processor machines [4][5][6] , Cell 7, 8 , GPU 9 or FPGA [10][11][12] . All these implementations have to face the memory bottleneck due to the limited bandwidth of the main memory.…”

Section: Introductionmentioning

confidence: 99%

A hardware projector/backprojector pair for 3D PET reconstruction

et al. 2008

View full text Add to dashboard Cite

Forward and Backward projections are two computational costly steps in tomography image reconstruction such as Positron Emission Tomography (PET). To speed-up reconstruction time, a hardware projection/backprojection pair has been built following algorithm architecture adequacy principles. Thanks to an original memory access strategy based on an 3D adaptive and predictive memory cache, the external memory wall has been overcome. Thus, for both projector architectures several units run efficiently. Each unit reaches a computational throughput close to 1 operation per cycle.In this paper, we present how from our hardware projection/backprojection pair, an analytic (3D-RP) and an iterative (3D-EM) reconstruction algorithms can be implemented on a System on Programmable Chip (SoPC). First, an hardware/software partitioning is done based on the different steps of each algorithm. Then the reconstruction system is composed of two hardware configurations of the programmable logic resources (FPGA). Each one corresponds mainly to the projection and backprojection step.Our projector/backprojector has been validated with a software 3D-RP and 3D-EM reconstruction on simulated PET-SORTEO data. A reconstruction time evaluation of these reconstruction systems are done based on the measured performances of our projectors IPs and the estimated performances of the additional simple hardware IPs. The expected reconstruction time is compared with the software tomography distribution STIR. A speed-up of 7 can be expected for the 3D-RP algorithm and a speed-up of 3.5 for the 3D-EM algorithm. For both algorithms, the architecture cycle efficiency expected is largely greater than the software implementation : 120 times for 3D-RP and 60 times for 3D-EM.

show abstract

Parallelization and Runtime Prediction of the ListMode OSEM Algorithm for 3D PET Reconstruction

Cited by 7 publications

References 7 publications

High Speed 3D Tomography on CPU, GPU, and FPGA

High Speed 3D Tomography on CPU, GPU, and FPGA

Scatter Correction in PET Using the Transport Equation

A hardware projector/backprojector pair for 3D PET reconstruction

Contact Info

Product

Resources

About