Extraction And Analysis Of Pulmonary Vascular Based On MDCT Images

Pulmonary hypertension is a common disease of clinical, because pulmonary vascular remodeling induced pulmonary hemodynamic changes, and ultimately can lead to right heart failure, making it a high rate of morbidity and mortality, serious harm to physical and mental health of patients, increasing the burden of social health major diseases. Pulmonary vascular is one of the most important tissues with the most complex topological structure. It is the target for human beings research to analyze the thoracic 3D CT images with computer to realize the aided diagnosis of pulmonary diseases. Correct segmentation of pulmonary vascular is the key to construct the functional mapping and aided diagnosis system.Measuring the pulmonary artery and its branches can prompt diagnosis of the disease which caused by lung disease and pulmonary vascular morphological changes. For example, there is some intrinsic link between the pulmonary artery and pulmonary artery pressure, when pulmonary artery pressure increased to a certain extent, has also come to pulmonary artery diameter expand, thus can be predicted the pulmonary artery hypertension by the diameter. Because of the uniqueness of Hessian matrix’s eigenvalues and eigenvectors in different structures, Hessian matrix always gets splendid results in the algorithm designation of the segmentation of tissues with complex structures. The article will have a deep research on Hessian matrix’s eigenvalues and eigenvectors and give an algorithm of 3D images segmentation based on Hessian matrix. The algorithm includes the threshold segmentation of lung based on morphology, the design of multi-scale filters, the vascular traversal for connecting the break between two segments and the analysis of the connected objects of pulmonary vascular. The algorithm combines the morphological characteristics of pulmonary vascular fully, has strict logic and clear orderliness and provides great convenience to the designation of program.The measurement of pulmonary artery reference to domestic and foreign scholars in the measurement method of the cross-sectional images, which using a combination of coronal and sagittal measurements, make it more accurate of positioning the pulmonary artery and its branches.The algorithm has been applied to 28 cases. All the cases’ left and right pulmonary vascular is segmented successfully. Furthermore, the segmented vascular has sharp structure, the branches of vascular can be seen clearly and tiny vessel is also segmented completely. In clinical diagnosis, thick vessel has more medical value. Users can keep the thick vessel and drop out the thin vessel by changing the parameter setting of the segmentation algorithm. Then the segmented vascular structure will be compared with the healthy and normal vascular structure to find out the abnormal symptoms and get the diagnosis results. The 20 patients were divided into normal and diseased group, analysis the measured data:pulmonary artery, left pulmonary artery, right pulmonary artery in each group is significant gender difference, and there is also statistically significant between normal and diseased groups.