TY - JOUR
T1 - Near real-time measurement of myocardial contraction and relaxation based on high-frame-rate ultrasound with a graphical processing unit system
AU - Asai, Takuma
AU - Taki, Hirofumi
AU - Kanai, Hiroshi
N1 - Publisher Copyright:
© 2017 The Institute of Electrical Engineers of Japan.
PY - 2017
Y1 - 2017
N2 - Several speckle tracking methods have been proposed for noninvasive and quantitative evaluation of tissue motion. Since the low temporal resolution causes a large myocardial motion in the elevational direction and a large deformation, two-dimensional (2D) speckle tracking at a high frame rate is desirable for accurate estimation of myocardial contraction and relaxation. 2D speckle tracking at a high frame rate requires a high computational load, and the large suppression of calculation time is, therefore, essential for clinical use. In the present study, we investigate the minimum frame rate required for the estimation of myocardial contraction and relaxation. Furthermore, we employ a parallel computing principle using a graphical processing unit (GPU) system with 2,496 streaming processors to decrease the calculation time effectively. The employment of a parallel computing principle with a GPU system successfully decreased the calculation time to 1/50 of that using a desktop PC with a CPU. When the number of tracking points is 64, the calculation time was decreased to 28.7 s for the estimation during 1 s at a frame rate of 287 Hz, indicating that the proposed method with a GPU system has a potential to realize a near real-time estimation of myocardial contraction and relaxation.
AB - Several speckle tracking methods have been proposed for noninvasive and quantitative evaluation of tissue motion. Since the low temporal resolution causes a large myocardial motion in the elevational direction and a large deformation, two-dimensional (2D) speckle tracking at a high frame rate is desirable for accurate estimation of myocardial contraction and relaxation. 2D speckle tracking at a high frame rate requires a high computational load, and the large suppression of calculation time is, therefore, essential for clinical use. In the present study, we investigate the minimum frame rate required for the estimation of myocardial contraction and relaxation. Furthermore, we employ a parallel computing principle using a graphical processing unit (GPU) system with 2,496 streaming processors to decrease the calculation time effectively. The employment of a parallel computing principle with a GPU system successfully decreased the calculation time to 1/50 of that using a desktop PC with a CPU. When the number of tracking points is 64, the calculation time was decreased to 28.7 s for the estimation during 1 s at a frame rate of 287 Hz, indicating that the proposed method with a GPU system has a potential to realize a near real-time estimation of myocardial contraction and relaxation.
KW - Graphical processing unit (GPU)
KW - Myocardial contraction/relaxation property
KW - Speckle tracking
KW - Two-dimensional (2D) displacement estimation
UR - http://www.scopus.com/inward/record.url?scp=85016993063&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85016993063&partnerID=8YFLogxK
U2 - 10.1541/ieejeiss.137.573
DO - 10.1541/ieejeiss.137.573
M3 - Article
AN - SCOPUS:85016993063
VL - 137
SP - 573
EP - 578
JO - IEEJ Transactions on Electronics, Information and Systems
JF - IEEJ Transactions on Electronics, Information and Systems
SN - 0385-4221
IS - 4
ER -