Research On Left Atrial Appendage Image Segment Method

The left atrial appendage is one of the important organizational structures in the human heart.It is connected to the left atrium and communicates internal.There are many branches crisscross each other intemal.The changing of the sharp of the left atrial appendage are closely related to heart disease.Segmentation of the left atrial appendage is one of the important steps in the analysis of the left atrial appendage.In clinical,left atrial appendage occlusion,use biological 3D printing technology to manufacture artificial left atrial appendages studies is based on left atrial appendage image segmentation.This paper focuses on the study of image segmentation algorithms for the left atrial appendage.The left atrial appendage is special heart structure,which connected with the left atrium in physiological,and without obvious boundary between the left atrial and the left atrium on the CT image.The traditional image segmentation method is based on the difference in the image gray value or gray distribution at the boundary as the feature to identify the boundary.However,the left atrial appendage is a structure which have no boundaiy,since it is impossible to obtain the boundary feature from the grayscale.It is difficult to accurately determine the segmentation boundary.For this reason,this paper proposes three improved algor:ithms for the structural characteristics of left atrial appendage without border and left atrium adhesion:1.The new segment method is based on the regional growth algorithm,where an improved pixel growth strategy is proposed for the left atrial appendage structure.The Bresenham algorithm is used to set the growth stop surface at the junction between the left atrium and the left atrial appendage,and then the region growing algorithm stopped pixel growth at the adhesion boundary.The experimental results show that the left atrial appendage structure can be segmented using this method.But there are also deficiencies in human interaction.2.In order to solve the problem that Bresenham region growing algorithm needs manual interaction,according to the characteristics of location of left atrial appendage image,this paper proposes a new level set algorithm based on distance to offset the missing of boundary feature.The experimental results show that this method has better segmentation results than the common distance watershed algorithm.3.The Atlas method can use prior knowledge effectively since used for image segmentation with weak boundary and adhesion problem widely.However,the traditional Atlas algorithm requires a large number of label images and a large number of registration processes,which is inefficient and cannot be widely applied to sequence segmentation and three-dimensional segmentation.In this paper,the similarity between adjacent images in the sequence is used to realize the rough segmentation with single label image get segmentation of a sequence images.Then the segmentation resultis used as the initial contour of the distance level set algorithm to achieve accurate segmentation of the left atrial appendage image sequence.The experimental results show that the Bresenham algorithm can be segmented semi-automatically for the left atrial appendage without border features.The distance feature level set algorithm can improve the robustness of the algorithm.Applying the contour obtained from the adjacent information Atlas to the distance level set algorithm can improve the accuracy of left atrial appendage image segmentation.This paper aims at the location characteristics of the left atrial appendage image,proposing an accurate method for segmenting the left atrial appendage and achieves a good result.