Research On Speckle Tracking Algorithm For Quantitative Analysis Of Echocardiography

At present, cardiovascular disease has become one of the diseases with the highest incidence and mortality rates in our country. Cardiovascular disease not only brings physical and psychological distress to patients, but also brings heavy pressure to society. Ultrasonic imaging technology is one of the most direct and effective methods, which can prevent, treat cardiovascular disease, and it can make noninvasive observation and detection of cardiac motion. To calculate accurate myocardial motion parameters more directly, effectively and accurately, speckle tracking algorithm is put forward.This paper is mainly based on speckle tracking algorithm to carry out the research work.Firstly, the principle of ultrasonic imaging and image speckle is introduced, and the feasibility of the speckle tracking technology’s application in ultrasonic image is analyzed. Then, this paper introduces two main methods of speckle tracking algorithm: optical flow algorithm and block matching algorithm, and the main characteristics of two main methods are given. Compared with the speckle tracking algorithm based on optical flow algorithm, the speckle tracking algorithm based on block matching algorithm has advantages of higher tracking accuracy and faster computating speed and better anti-jamming immunity. According to the requirements of clinical application, the full search block matching algorithm whose accuracy is the highest is adopted as the basic algorithm of the speckle tracking algorithm.Secondly, the flow and the tracking results in the actual cardiac ultrasound images of the speckle tracking algorithm based on full search block matching algorithm are given and analyzed briefly. In view of the shortcomings of the full search block matching algorithm, the algorithm based on calculating tracking error is proposed. The speckle tracking algorithm based on calculating tracking error has good tracking results, which shows that this algorithm can effectively improve the applicability, stability and accuracy of tracking in different ultrasound images and multi-point tracking in the same ultrasound images.Thirdly, in order to meet the real-time requirements of ultrasonic medical clinical better, GPU parallel computing is applied, the speckle tracking algorithm based on full search block matching algorithm is parallelized and implemented by CUDA. Several modes of different template block, tracking distance and different image storage schemes are designed to conduct experiment of a cceleration, and the results are analyzed. The highest accelerated storage scheme is applied to the CUDA implementation of the algorithm based on calculating tracking error, which greatly accelerated the computing speed of the speckle tracking algorithm.Finally, according to the tracking data in the ultrasound images of the algorithm based on calculating tracking error, quantitative myocardial parameters curve, which can be used in clinical, is calculated. In order to avoid possible tracking errors which influence the accuracy of myocardial parameters, random sample consensus algorithm is adopted to remove the error tracking data, which can reduce the calculation error caused by tracking errors, parameter calculation results prove that the random sample consensus algorithm can effectively guarantee the calculation accuracy of parametric curves.