Segmentation Of Pulmonary Vascular In CT Images Based On 3D Region Growing Method

Computed tomography(C T)imaging has the advantage of high spatial resolution,and it has become the first choice for imaging diagnosis of pulmonary diseases.The majority of doctors use pulmonary CT images to diagnosis and treat pulmonary diseases.The segmentation of pulmonary vascular in enhanced CT images which has important research significance and application value is the base and key of computer aided detection and diagnosis of pulmonary diseases such as pneumonia and pulmonary embolism.Combined with the characteristics of pulmonary vascular,this paper proposes a new algorithm for pulmonary vascular segmentation based on 3D region growing method on the basis of literature reading.There are three main research work.(1)The study and research of related theoretical basis.Learning about the medical and medical imaging of the pulmonary and the distribution of pulmonary vascular and trachea.Studying the segmentation method of pulmonary vascular at present,analyzed the research results of pulmonary vascular segmentation based on region growing algorithm further.Understanding the current research status of pulmonary vascular segmentation domestic and foreign by reading literature deeply.(2)Algorithm design and analysis.The whole algorithm consists of 4 parts: image preprocessing,alveolar segmentation,pulmonary trachea segmentation,pulmonary vascular segmentation and so on.Aiming at alveolar segmentation,the algorithm based on Water filling was proposed which solved the problem of error segmentation and excessive segmentation.Aiming at pulmonary trachea segmentation,the 26-neighborhood 3D region growing algorithm was proposed which can segment pulmonary trachea accurately.Aiming at pulmonary vascular segmentation,remove the trachea pixel from the CT image at first,then we can use 26-neighborhood 3D region growing segmentation algorithm to segment pulmonary vascular.(3)Algorithm implementation and verification.Using the software platform of Visual Studio 2010 combined with the third open source library of OpenCV and VTK,using C++ language programming to implement the algorithm.Verifying the feasibility of the algorithm by the experimental results.Giving quantitative evaluation by using image segmentation evaluation index.The innovation of the research is that(1)the use of double threshold as the growth rules of three-dimensional region growth,after the seed points manually selected,according to the mean variance and gray as growth rules,so as to improve the accuracy of segmentation.(2)Aiming at the problem of error segmentation and excessive segmentation,a fully automatic method based on water filling algorithm is proposed.The deficiency of the research is that the segmentation of pulmo nary capillaries is not complete,and the degree of automation of segmentation needs to be further improved.