Please use this identifier to cite or link to this item: http://localhost/handle/Hannan/594105
Title: An FPGA-Based Real-Time Maximum Likelihood 3D Position Estimation for a Continuous Crystal PET Detector
Authors: Yonggang Wang;Yong Xiao;Xinyi Cheng;Deng Li;Liwei Wang
subject: depth-of-interaction|PET detector|Continuous crystal|FPGA|maximum likelihood estimation|position estimation
Year: 2016
Publisher: IEEE
Abstract: For the continuous crystal-based positron emission tomography (PET) detector built in our lab, a maximum likelihood algorithm adapted for implementation on a field programmable gate array (FPGA) is proposed to estimate the three-dimensional (3D) coordinate of interaction position with the single-end detected scintillation light response. The row-sum and column-sum readout scheme organizes the 64 channels of photomultiplier (PMT) into eight row signals and eight column signals to be readout for X- and Y-coordinates estimation independently. By the reference events irradiated in a known oblique angle, the probability density function (PDF) for each depth-of-interaction (DOI) segment is generated, by which the reference events in perpendicular irradiation are assigned to DOI segments for generating the PDFs for X and Y estimation in each DOI layer. Evaluated by the experimental data, the algorithm achieves an average X resolution of 1.69 mm along the central X-axis, and DOI resolution of 3.70 mm over the whole thickness (0-10 mm) of crystal. The performance improvements from 2D estimation to the 3D algorithm are also presented. Benefiting from abundant resources of FPGA and a hierarchical storage arrangement, the whole algorithm can be implemented into a middle-scale FPGA. By a parallel structure in pipelines, the 3D position estimator on the FPGA can achieve a processing throughput of 15 M events/s, which is sufficient for the requirement of real-time PET imaging.
URI: http://localhost/handle/Hannan/175586
http://localhost/handle/Hannan/594105
ISSN: 0018-9499
1558-1578
volume: 63
issue: 1
Appears in Collections:2016

Files in This Item:
File Description SizeFormat 
7407516.pdf1.16 MBAdobe PDFThumbnail
Preview File
Title: An FPGA-Based Real-Time Maximum Likelihood 3D Position Estimation for a Continuous Crystal PET Detector
Authors: Yonggang Wang;Yong Xiao;Xinyi Cheng;Deng Li;Liwei Wang
subject: depth-of-interaction|PET detector|Continuous crystal|FPGA|maximum likelihood estimation|position estimation
Year: 2016
Publisher: IEEE
Abstract: For the continuous crystal-based positron emission tomography (PET) detector built in our lab, a maximum likelihood algorithm adapted for implementation on a field programmable gate array (FPGA) is proposed to estimate the three-dimensional (3D) coordinate of interaction position with the single-end detected scintillation light response. The row-sum and column-sum readout scheme organizes the 64 channels of photomultiplier (PMT) into eight row signals and eight column signals to be readout for X- and Y-coordinates estimation independently. By the reference events irradiated in a known oblique angle, the probability density function (PDF) for each depth-of-interaction (DOI) segment is generated, by which the reference events in perpendicular irradiation are assigned to DOI segments for generating the PDFs for X and Y estimation in each DOI layer. Evaluated by the experimental data, the algorithm achieves an average X resolution of 1.69 mm along the central X-axis, and DOI resolution of 3.70 mm over the whole thickness (0-10 mm) of crystal. The performance improvements from 2D estimation to the 3D algorithm are also presented. Benefiting from abundant resources of FPGA and a hierarchical storage arrangement, the whole algorithm can be implemented into a middle-scale FPGA. By a parallel structure in pipelines, the 3D position estimator on the FPGA can achieve a processing throughput of 15 M events/s, which is sufficient for the requirement of real-time PET imaging.
URI: http://localhost/handle/Hannan/175586
http://localhost/handle/Hannan/594105
ISSN: 0018-9499
1558-1578
volume: 63
issue: 1
Appears in Collections:2016

Files in This Item:
File Description SizeFormat 
7407516.pdf1.16 MBAdobe PDFThumbnail
Preview File
Title: An FPGA-Based Real-Time Maximum Likelihood 3D Position Estimation for a Continuous Crystal PET Detector
Authors: Yonggang Wang;Yong Xiao;Xinyi Cheng;Deng Li;Liwei Wang
subject: depth-of-interaction|PET detector|Continuous crystal|FPGA|maximum likelihood estimation|position estimation
Year: 2016
Publisher: IEEE
Abstract: For the continuous crystal-based positron emission tomography (PET) detector built in our lab, a maximum likelihood algorithm adapted for implementation on a field programmable gate array (FPGA) is proposed to estimate the three-dimensional (3D) coordinate of interaction position with the single-end detected scintillation light response. The row-sum and column-sum readout scheme organizes the 64 channels of photomultiplier (PMT) into eight row signals and eight column signals to be readout for X- and Y-coordinates estimation independently. By the reference events irradiated in a known oblique angle, the probability density function (PDF) for each depth-of-interaction (DOI) segment is generated, by which the reference events in perpendicular irradiation are assigned to DOI segments for generating the PDFs for X and Y estimation in each DOI layer. Evaluated by the experimental data, the algorithm achieves an average X resolution of 1.69 mm along the central X-axis, and DOI resolution of 3.70 mm over the whole thickness (0-10 mm) of crystal. The performance improvements from 2D estimation to the 3D algorithm are also presented. Benefiting from abundant resources of FPGA and a hierarchical storage arrangement, the whole algorithm can be implemented into a middle-scale FPGA. By a parallel structure in pipelines, the 3D position estimator on the FPGA can achieve a processing throughput of 15 M events/s, which is sufficient for the requirement of real-time PET imaging.
URI: http://localhost/handle/Hannan/175586
http://localhost/handle/Hannan/594105
ISSN: 0018-9499
1558-1578
volume: 63
issue: 1
Appears in Collections:2016

Files in This Item:
File Description SizeFormat 
7407516.pdf1.16 MBAdobe PDFThumbnail
Preview File