TY - GEN
T1 - Delay Compression
T2 - 2019 IEEE International Ultrasonics Symposium, IUS 2019
AU - West, Brendan L.
AU - Zhou, Jian
AU - Chakrabarti, Chaitali
AU - Wenisch, Thomas F.
N1 - Funding Information:
This work was supported in part by NSF CCF-1406739 and the Applications Driving Architectures (ADA) Research Center, a JUMP Center cosponsored by SRC and DARPA.
Publisher Copyright:
© 2019 IEEE.
PY - 2019/10
Y1 - 2019/10
N2 - In 3D plane-wave ultrasound, computational requirements are directly proportional to the number of focal points in a volume. For a receive aperture of size of MxMy transducers, a beamforming aperture size of NxNy, and a depth of Mz focal points, Mx My Mz Nx Ny round-trip delays must be computed. To reduce this requirement, we decompose the planar transmit distance into two parts: (1) from the plane-wave's origin to the first point in each focal line, and (2) from the plane when it is touching the first point in each focal line to each subsequent focal point along that line. The latter distance, as well as the reflection distances, are symmetric across beamforming apertures, and thus their computation can be shared. This decomposition results in up to a Mx My reduction in the number of unique delays while retaining full image quality. Using our technique, precomputing delays and storing them in look-up tables (LUTs) is now possible for 3D plane-wave ultrasound for the first time, opening new doors for computational architectures in this field. Our method works with 2D, 3D, and 3D-separable variants of plane-wave ultrasound.
AB - In 3D plane-wave ultrasound, computational requirements are directly proportional to the number of focal points in a volume. For a receive aperture of size of MxMy transducers, a beamforming aperture size of NxNy, and a depth of Mz focal points, Mx My Mz Nx Ny round-trip delays must be computed. To reduce this requirement, we decompose the planar transmit distance into two parts: (1) from the plane-wave's origin to the first point in each focal line, and (2) from the plane when it is touching the first point in each focal line to each subsequent focal point along that line. The latter distance, as well as the reflection distances, are symmetric across beamforming apertures, and thus their computation can be shared. This decomposition results in up to a Mx My reduction in the number of unique delays while retaining full image quality. Using our technique, precomputing delays and storing them in look-up tables (LUTs) is now possible for 3D plane-wave ultrasound for the first time, opening new doors for computational architectures in this field. Our method works with 2D, 3D, and 3D-separable variants of plane-wave ultrasound.
KW - accelerator
KW - beamforming
KW - plane-wave
KW - ultrasound
UR - http://www.scopus.com/inward/record.url?scp=85077575814&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85077575814&partnerID=8YFLogxK
U2 - 10.1109/ULTSYM.2019.8925725
DO - 10.1109/ULTSYM.2019.8925725
M3 - Conference contribution
AN - SCOPUS:85077575814
T3 - IEEE International Ultrasonics Symposium, IUS
SP - 1278
EP - 1281
BT - 2019 IEEE International Ultrasonics Symposium, IUS 2019
PB - IEEE Computer Society
Y2 - 6 October 2019 through 9 October 2019
ER -