Implementing acoustic radiation force imaging on GPU using OpenCL

Acoustic Radiation Force Impulse (ARFI) imaging is a novel method of tissue stiffness estimation. By using radiation force to generate and measure the small displacement of tissue, it can detect and calculate the stiffness of tissues. Since tissue elasticity is known to be associated with pathological condition, ARFI can used to diagnose disease. Normally, massive computation is required in order to finish data processing, so it will be computationally expensive to implement with traditional platform and methods. In this paper, a new method of ARFI imaging based on OpenCL (Open Computing Language) parallel processing platform is introduced. Based on this platform and Loupas' two-dimensional (2-D) autocorrelation [1] algorithm, some optimization has been made and compared. The results of the comparison demonstrate the speedup using GPU and each time the optimization been made.