Ultrasonic Texture Evaluation Segmentation Method Based On Three-dimensional Block Matching

Medical imaging as an important part of medicine in medical theory as well as clinical diagnosis Ultrasound images with the features of non-destructive,inexpensive,real-time operation,becomes the one of the most important medical imaging system Compared with other medical imaging techniques,the application of ultrasound in clinical diagnosis has a significant popularity.Therefore,the studies of ultrasound image processing has booming.Image segmentation as a key step in image analysis,information medicine,has got the attention of many researchers.However,because the ultrasound image has a relatively poor quality compare with CT and MRI technology.This caused the ultrasound image segmentation of a major difficulty.Ultrasound images' specific speckle texture increases the difficulty of ultrasound segmentation.In the field of ultrasound image processing has a lot of literatures on how to remove speckle.In addition,ultrasonic speckle is also pointed out that has the relationship with the nature of the organizations' position.How to utilize speckle to provide information for tissue segmentation while reducing the impact of speckle on ultrasound images is one of the focus of ultrasound segmentation research.In this paper,two algorithms based on nonlocality are proposed for the randomness and ambiguity of ultrasonic speckle texture.'Ambiguity' means that speckle affects the quality of ultrasound imaging at the same time in the distribution of the organization are closely linked.First,this paper proposes a non-local block matched feature extraction algorithm for randomness of ultrasonic speckle texture,which solves the problem that it is difficult to extract uniform eigenvalue when the random speckle texture is strong.Secondly,this paper proposes an adaptive feature extraction algorithm for the ambiguity of ultrasonic speckle texture,which solves the problem that the tissue feature information of the speckle is easy to be removed when the general denoising is preprocessed.And the validity of the two methods is proved by comparison experiments.Finally,this paper combines the two methods,proposed based on three-dimensional block matching ultrasonic texture extraction method,and the ultrasound image segmentation.In addition,this method is improved to solve the problem that the method is unstable at the weak echo of ultrasound.And the validity of these two methods was analyzed by comparison experiments.