TY - GEN
T1 - OpenACC GPU implementation of double-stage delay-multiply-and-sum algorithm
T2 - Photons Plus Ultrasound: Imaging and Sensing 2019
AU - Miri Rostami, Seyyed Reza
AU - Mozaffarzadeh, Moein
AU - Hariri, Ali
AU - Jokerst, Jesse V.
AU - Ghaffari-Miab, Mohsen
N1 - Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.
PY - 2019
Y1 - 2019
N2 - Double-stage delay-multiply-and-sum (DS-DMAS) is one of the algorithms proposed for photoacoustic image reconstruction where a linear-array transducer is used to detect signals. This algorithm provides a higher contrast image in comparison with the conventional delay-multiply-and-sum (DMAS) and delay-and-sum (DAS), but it imposes a high computational complexity. In this paper, open accelerators (OpenACC) GPU computation parallel approach is used to lessen the computational time and address the high computational time of the DSDMAS for photoacoustic image reconstruction process. Compared with sequential execution of the DS-DMAS on CPU, a speed-up of approximately 74× is achieved (for an image having 1024 × 1024 pixels). The proposed approach provides possibility to have an accurate reconstructed photoacoustic image with a reasonable frame rate. In addition, the higher the number of the image pixels, the higher speed-up is achieved. Using the suggested GPU implementation, it is feasible to reconstruct photoacoustic images having a size of 128 × 128, and 256 × 256 with a frame rate of 3 and 2, respectively.
AB - Double-stage delay-multiply-and-sum (DS-DMAS) is one of the algorithms proposed for photoacoustic image reconstruction where a linear-array transducer is used to detect signals. This algorithm provides a higher contrast image in comparison with the conventional delay-multiply-and-sum (DMAS) and delay-and-sum (DAS), but it imposes a high computational complexity. In this paper, open accelerators (OpenACC) GPU computation parallel approach is used to lessen the computational time and address the high computational time of the DSDMAS for photoacoustic image reconstruction process. Compared with sequential execution of the DS-DMAS on CPU, a speed-up of approximately 74× is achieved (for an image having 1024 × 1024 pixels). The proposed approach provides possibility to have an accurate reconstructed photoacoustic image with a reasonable frame rate. In addition, the higher the number of the image pixels, the higher speed-up is achieved. Using the suggested GPU implementation, it is feasible to reconstruct photoacoustic images having a size of 128 × 128, and 256 × 256 with a frame rate of 3 and 2, respectively.
KW - Beamforming
KW - Central Processing Unit (CPU)
KW - Double-stage delay-multiply-and-sum (DS-DMAS)
KW - Eigenspace-based minimum variance
KW - Graphics Processing Unit (GPU)
KW - Linear-array imaging
KW - OpenACC (Open accelerators)
KW - Parallel computing
KW - Photoacoustic imaging
UR - http://www.scopus.com/inward/record.url?scp=85065424177&partnerID=8YFLogxK
U2 - 10.1117/12.2511115
DO - 10.1117/12.2511115
M3 - Conference contribution
AN - SCOPUS:85065424177
VL - 10878
T3 - PHOTONS PLUS ULTRASOUND: IMAGING AND SENSING 2019
BT - Proceedings of SPIE
A2 - Wang, Lihong V.
A2 - Oraevsky, Alexander A.
PB - SPIE
Y2 - 3 February 2019 through 6 February 2019
ER -