Research And Implementation On Automatic Whole Heart Segmentation Algorithm In CT Images

Cardiovascular disease is one of the most common diseases in the world,and it is the leading direct or contributing cause of non-accidental deaths.The heart,in the cardiovascular system,is an important organ.Therefore,the heart is accurately segmented plays a vital role in early diagnosis and treatment.However,fully automatic whole heart segmentation is difficult due to the two main challenges.On the one head,because of the complexity of the heart anatomical structure,there is a certain range of overlap in the gray value,such as atrium,ventricle,coronary artery,inferior vena cava,pericardium.On the other head,the low-quality image includes heavy motion artifacts,intensity inhomogeneity and noise due to the continuous beat and blood flow within the heart.Hence,the algorithms of accurate positioning and segmentation need to be improved.In order to solve the problem,which is difficult to locate the heart in the CT images,in this paper,an automatic whole heart segmentation method,which is on the basis of the existing medical image segmentation algorithms,is proposed based on localizing region-based active contour model and watershed algorithm in CT images.And the specific work is as follows:(1)The gray features of CT cardiac images were analyzed,and the Gauss filter and bilateral filtering are used to perform the experiments.At the same time,two typical deformable models in medical image processing are researched analyzed,which lays the foundation for the design and optimization of CT image segmentation algorithm.(2)Researching on several typical localized active contour models,and we can find that redundant curves are easily produced by localized active contour models during curve evolution.Therefore,a localizing region-based active contour model is adopted,and we improved the Heaviside function and Dirac function.And the same time,we make analysis and comparison of the segmentation results with the typical localized active contour models,so it can be better applied to whole heart segmentation in CT images.(3)Studying and analyzing the watershed algorithm,and it is equipped with fast running speed,high accuracy and the ability to obtain the contour of the closed target area.According to the characteristics of watershed algorithm and CT cardiac images,we propose an initial contour setting algorithm of CT whole heart images based on improved watershed.At the same time,the automatic extraction of the initial contour of CT whole heart images is realized.(4)In order to solve the difficulty of the whole heart segmentation in CT images,we propose a hybrid model based on localizing region-based active contour model and watershed algorithm.And it is applied to whole heart segmentation in CT images.The experimental results on several groups of clinical cardiac CT data show that the proposed model has a wide range of practicability,good accuracy and robustness to noise.