Motion Estimation Of Ultrasound Images Based On Shape Model And Discrete Optimization

Currently, cardiovascular disease has become one of the world’s highest incidence of disease, which is a serious threat to the human health. The rapid development of medical imaging technology makes the means of clinical diagnosis of cardiovascular disease become more and more rich, and the accuracy is higher and higher. The main diagnostic methods include:Electrocardiogram, Echocardiogram, MRI, CT, Coronary Angiography and Radionuclide Examination, etc.Ultrasonic imaging has been used as an important method for diagnosis of heart disease and heart function evaluation because of its advantages of non-infringement, low cost, convenience and real-time motion display of organs and tissues. Medical dynamic image sequences can provide more information for the doctor’s diagnosis, the echocardiogram motion analysis can extract the motion information of the heart, thus to help doctors improve the diagnostic efficiency and accuracy, and compensate for the lack of low spatial resolution of ultrasound images.Optical flow method and the block matching method as two kind of classical motion estimation algorithm have been widely used in the motion estimation of ultrasound image sequences. But because of the complexity of the cardiac motion and the dramatically movement between the opening and closing of atrioventricular valve, motion estimation of echocardiographic image sequence is faced with many difficulties. The assumption of the time interval between adjacent frames is small enough by the optical flow method when calculating the optical flow is challenged due to large motion range of the heart, and the block matching algorithm’s time complexity is very high because of the searching and regional image matching.Given the traditional approach in cardiac ultrasound image sequences motion estimation vulnerable to the complicated movement of the heart, the ASM segmentation methods was applied to the early stage of motion estimation in this paper, and further optimize the atrioventricular region contours obtained by ASM method, making it with better atrioventricular edge joint. On this basis, we designed a double guide mechanism of search strategy based on the principle of the ACO with neighboring frame image contours by ASM method, through the use of a priori contour information to improve the efficiency of the motion search, and ensure the motion vector obtained with better availability and high precision at the same time. Experimental results show the motion vectors obtained by the method in this paper can be easily applied to independently extract accurate motion information of every atrioventricular region, thus can compare and comprehensive analysis.