Research And Implementation Of Algorithm For CT Image Segmentation Of Lung Lobe

Due to air pollution,smoking,aging population and other factors,the incidence of chronic obstructive pulmonary disease and mortality is increasing in recent years,it has become a serious public health problem and social problem that endanger human health and endanger social and economic sustainable development.Chest CT examination can provide high-density resolution,high contrast images of the lungs,it has obvious advantages in the clinical diagnosis of chronic obstructive pulmonary disease.But the current routine CT examination is mainly the qualitative diagnosis of lung tissue morphology,subjectivity,easily lead to misdiagnosis and omission,and it is mainly for the diagnosis of single lung,unable to make an effective quantitative assessment of lobectomy.It has important value in making accurate assessment,reasonable operation scheme of accurate segmentation of lung lobe in patients with chronic obstructive pulmonary disease.Because of the incomplete of the fissures between the lung lobes and comparison of the fissures between the lobes of the lung on CT images,it is difficult to directly use the fissures to segment the lobes.Through the comprehensive analysis of the anatomical structure of the lung,ofutually independent features of vessels in different lobes,a method is proposed by automaticly search the interface of vessel in two adjacent lobe by classification method to complete lung lobe segmentation in this paper.Firstly,the left and right lung tissues were segmented by 3D region growing in this paper.And then eliminate the adhesion between left lung and right lung based on the adhesion characteristics to get two lung tissues that are not connected to each other.A method is proposed to eliminate the adhesion of the left and right lung with the approximate boundary of the corrosion.The blood vessels were filled by morphological operation.And then segment the lung vessels by fast marching method.And the vessels of different lobes are marked by different value,then,to reduce the amount of calculation,the central pathway of pulmonary vessels was extracted.And the vessels of different adjacent lobes are used to train the two classification support vector machine model to get a two classification support vector machine classifier.At last,classification of lung tissue with support vector machine classifier.The experimental results show that the lung lobe segmentation method proposed in this paper is fast and accurate,and has important clinical application value.