Motion Analysis Of Carotid Ultrasound Image Sequence

Cardiovascular diseases such as atherosclerosis have been a great threat to health. The cardiovascular diseases are highly be related with the CIMT (Carotid Intima-Medial Thickness) and motion feature. The motion feature can reflect the condition of carotid whether it has hardened or expanded and contracted in the right way, and then helps doctors to diagnose Cardio-vascular diseases.Ultrasound detection has been widely used because of its non-invasion, low-cost, but Ultrasound images have its own speckle noise. Although speckle has influenced the quality of images, they can be used to track the motion related information because of its unique and stable feature. This paper gives the complete theme of analyzing the Carotid motion feature based on the improved speckle motion tracking algorithm. The main work of this paper is as follows:(1) There are two commonly used methods to track the motion in the ultrasound images. They are optical flow method and block matching algorithm, where the block matching algorithm is easy to implement, not as the optical flow method which needs the complicated iterations costing too much time. So we also choose the block matching algorithm to track the motion information from the images. Before using the block matching algorithm, we use the ultrasound linear model to simulate the ultrasound images to validate the accuracy of the block matching algorithm while using it to track the motion information. The result shows that it can detect the motion vector correctly.(2) The block matching algorithm includes two aspects, searching strategy and matching criterion. There are several searching strategies such as full searching, diamond searching and three step method and so on. The three step method is more efficient, but it needs that most of the motion vectors are next to the centre of the searching range. Simulation shows the distribution of motion vectors of ultrasound images also has the feature. We use the new three step searching method (NTSS) and find the new three step method has more efficiency. In the other aspect, there are also a lot of criterions such as NCC (normalized cross correlation), MAD (MAD, mean absolute of differences), MSD (mean square of differences). We use the MAD because it is easier with no complex multiplication operations. Besides that, while comparing the best candidate block using MAD, we can use the BSP (block sum pyramid) fast algorithm. We modify the block pyramid algorithm (MBSP) by adding several MAD values, and the simulation result shows that the modified algorithm can improve the speed to a certain extent. Eventually we consider combining NTSS and MBSP algorithms to enhance the speed of motion tracking.(3) We provide the qualitative and quantitative tools to analyze the Carotid in clinical application based on the above efficient speckle motion tracking method. The qualitative tool is to add the motion vectors to the original image sequence in suitable proportion while displaying dynamically. This tool let doctors can observe the state of Carotid within the heart cycle. The quantitative tool is the interpolated curve of motion vector, strain and strain rate within the heart cycle. From the curves, we can extract the peak values of motion vector, strain and strain rate which can reflect the state of Carotid, and then use the statistical knowledge analyzes these parameters acquiring from different groups (Normal group, CIMT thickened group, Plaque formation group). Eventually hope these tools can assist doctors to diagnose the state of Carotid comprehensively.This paper improves the current speckle motion tracking algorithm and then uses them to extract the motion feature of Carotid Ultrasound images. Results show that the improved algorithm is more efficient than before. Besides that, other qualitative and quantitative tools are also provided to hope that they can help doctors to analyze the relationship between the motion feature of Carotid and the cardiovascular diseases.