Optimized GPU Framework for Pulsed Wave Doppler Ultrasound

Pulsed Wave (PW) spectrum Doppler ultrasound is a valuable tool for clinical diagnosis for flow velocity distribution in vessels. However, real-time processing of PW spectrum is computationally intensive, involving wall filtering, Fast Fourier Transform (FFT), column filtering and linear averaging. In this paper a very efficient implementation of a PW Doppler spectrum ultrasound using the Compute Unified Device Architecture (CUDA™) platform developed by NVIDIA^® is presented. By exploiting the explicit parallelism exposed in the graphics hardware we obtain more than one order speed-up gain compared with that from standard CPUs. Finally, we get a rate of 7.60 μs with one line of 256 samples, which is about 92 times faster than the CPU implementation.