On The Trajectory Tracking Of Right Ventricles Using Two-Dimensional Ultrasound Images

Currently, cardiovascular disease has become a serious threat to human health, standing first in all kinds of death. Medical imaging technology can provide significant on the clinical diagnosis and treatment of heart disease. The geometry and movement of right ventricular is so complex that conventional means of ultrasound detection on its feature movement is insufficient. In recent years, there is a speckle-tracking imaging technology which can extract and match the speckles on right ventricular myocardium, thereby tracking the trajectory of myocardial tissue.Based on the motion sequence, this paper analyzes the motion of the right ventricular myocardium by speckle tracking imaging to enable a two-algorithm analysis. The main work is divided into four parts:1. Feature detection and tracking based on optical flow. According to the characteristics of the right ventricle ultrasound images, introduced a concept of optical flow method and did a detailed description of its basic principles and three traditional optical flow methods which are Horn & Schunck algorithm, Lucas-Kanade algorithm and pyramidal LK algorithm, finally, analyzed the results of feature points detecting and tracking based on optical flow method.2. Ultrasound image preprocessing. This paper took into the principle of ultrasound imaging, analyzed the reason of erroneous judgment on trajectory of some point and three filtering:linear mean filtering, median filtering and nonlinear adaptive Wiener filter, meanwhile excluded the area which impacted on matching feature points.3. Ultrasound image feature point extraction. In this paper, three pivotal characteristic detection operators were adopted to achieve the feature extraction, and briefly summarized on cardiac ultrasound image feature extraction by way of experiments and analysis, finally choose the SIFT algorithm as the best one.4. Experimental analysis of right ventricular point tracking. This paper achieved the matching of right ventricle ultrasound images using SIFT algorithm, analyzed the matching number of points, accuracy and trends to derive the movement of the right ventricular myocardium, and improved the coherence of SIFT matching algorithm on the multi-frame trace. Compared this algorithm against optical flow algorithm on the computational time and the tracking number of points and summarized.