Pulmonary Vascular Segmentation Methods In CT Images

With the development of medical imaging technology, CT(Computed Tomography)image has become an important basis for the diagnosis of diseases. Lung contains a large number of blood vessels, pulmonary artery goes along with pulmonary vein. The distribution of vascular in the lung CT image gives an important basis for diagnosis and surgical planning. segmenting pulmonary vascular correctly can give a more accurate diagnosis and treatment information and has important significance.This paper mainly studies the pulmonary vascular segmentation methods. Firstly, the lung parenchyma region is segmented by a method combing threshold segmentation with morphological repair. Then smoothing is used to remove the noise. We describes two pulmonary vessels extraction methods, the first method uses Hessian enhancement and a given threshold, and the second method uses OTSU and gradient operator, then the results of the two methods are compared. In order to improve the accuracy of segmentation, the bronchial walls and the impurities in the segmentation results are removed. Finally, the roots of pulmonary vessels are expanded, so that vessel branches can be merged into a tree.Experiments are used to verify the feasibility of pulmonary vascular segmentation method proposed in this study. In the experiment, 60 sets of CT images of patients were tested. Experimental results show that pulmonary vessel segmentation method used in this paper has a good result, Non-vascular pixels are removed correctly and the extracted vessels have obvious tree characteristics. Meanwhile the proposed method has a smaller calculation amount and a faster speed.