A Study Of Ultrasound Imaging Based On Compressed Sensing

In recent years, the compressive sensing(CS) theory has becoming a new researching hot spot in the signal processing domain. The theory, which provides a great prospects in reducing the signal sampling rate and improving the reconstruction quality, has even been explored in the ultrasound imaging domain. The quality of ultrasound imaging is relatively lower than other medical imaging methods because of its working mode and fundamental, and it is hard to achieve fast imaging speed with the reason of the ultrasound speed limit.In order to improve imaging quality and real-time performance, we can modify the imaging fundamental theory by introducing the CS theory. However, a large-scale measurement matrix and a complicated iterative optimization process are needed for the image reconstruction when the CS theory is applied to the UI, which is not conducive to the real-time performance. In this case, it is probably useful to combine parallel computing to the reconstruction process in the purpose of speed up the image reconstruction.In this paper, we focus on the fast UI issue under the CS framework. On the premise of good reconstruction quality, we first improve the UI fundamental theory by establishing the plane wave scattering model. Next, we apply the CS theory, in which a mathematic model is built and the Fast Iterative Shrinkage-Thresholding Algorithm(FISTA) is modified to reduce the computation complexity and the iterative times. Finally, we adopt the idea of parallel computation to the reconstruction process. Several parallel platforms’ performance, including Matlab, MPI and CUDA with different data size, is tested, and the results show that the highest speed-up ratio is achieved by the CUDA platform. The simulation experiments prove that our method can not only acquire high-quality ultrasound image, but also achieves a better real-time performance.