P3H-8 A Scalable Parallel Implementation of a k-Space Method for Large-Scale Ultrasound Imaging Simulations
Mohammad I. Daoud, Yi‐Ting Shen, James C. Lacefield · 2006
Two-dimensional (2D) and three-dimensional (3D) simulations of ultrasound pulse propagation in soft tissue provide an important tool in medical imaging research. However, large-scale 2D and 3D ultrasound simulations require long execution times on serial computers. Parallel computing provides a promising approach to large-scale ultrasound simulations with feasible execution times. In this paper, we present two parallel implementations of a k-space method used for ultrasound simulation. The first implementation divides the simulation grid between the nodes of a computer cluster, while the second implementation runs multiple simulated scan lines concurrently on a computer cluster. The results demonstrate the efficiency of parallel computing for ultrasound simulations with short running times. For example, with 40 processors, the ultrasound simulator that computes the simulated scan lines concurrently is 39.57 plusmn 0.43 faster than the serial ultrasound simulator