Research On High Frame Rate Ultrasonic Doppler Imaging Based On Compressed Sensing

In recent years,researchers have developed a variety of new ultrasonic imaging modes using related post-processing algorithms.These new ultrasound imaging models can obtain more tissue information and quantitatively evaluate tissue lesions,thus helping doctors to make more accurate clinical diagnosis.However,it also puts forward new requirements on the frame frequency and image quality of ultrasonic imaging.In view of the successful application of compressed sensing in rapid magnetic resonance imaging in the medical field in recent years,many people began to study its feasibility in the field of ultrasound.Although some progress has been made in some areas of ultrasound,the research on ultrasonic Doppler imaging is far from enough.Ultrasonic Doppler itself is a relatively complex imaging mode,which can see the internal structure of the body and blood flow information of a certain area of the body at the same time.Moreover,due to the fast blood flow velocity,in order to improve the maximum detectable blood flow velocity and avoid phase winding artifacts,the imaging frame frequency needs to be increased.The reason why the frame frequency of Doppler mode is much lower than that of B mode imaging is that it needs to emit multiple times at the same position in the imaging process,and calculate the phase change information between each ultrasonic echo,so as to obtain the blood flow velocity.Obviously,if you can reduce the number of shots per location while maintaining the similar imaging effect under the condition of many shots,you can greatly increase the frame rate of the imaging.This is the main content of this paper.In order to achieve this goal,this paper,based on the Principle of Doppler Imaging,firstly designed the transmission strategy based on compressed sensing technology to solve the problem of how to waste as little transmission time as possible.After that,the sampling strategy is designed to recover the complete data from scratch under the premise that the slow time data points are very few.Meanwhile,the problem that sparsity is not very good in common sparse domain based on slow time data is solved by constructing over-complete dictionary through training.In addition,the measurement matrix designed in the sampling strategy is optimized to reduce its correlation with the constructed dictionary to improve the performance of compressed sensing recovery.Finally,the effect of different parameters on reconstruction quality is studied in detail,and its feasibility is verified by simulation and in vivo experiment.Compared with other algorithms,this algorithm can reduce the sampling time,improve the frame rate of Doppler imaging without losing the image quality as much as possible,and reach a good compromise between the frame rate and image quality.