Research On Key Issues Of Pulmonary Function Evalution Using CT Images

Lung disease more and more threaten people's health and safety, and the early detection of lung disease becomes the focus of prevention of lung diseases. Along with the development of CT image and medical CAD technology, using CT image to help diagnose disease attracts more experts' attention, which acquired rapid progress. Pulmonary function check, one necessary step of respiratory disease check, has a significant guide on the early detection of lung and airway disease, assessment of the severity of the disease, identification of the cause of breath difficulties, diagnosis of other lesion. Based the sequence of lung CT images, this paper studies some key issues of the assessment of pulmonary function and mainly consist of following four parts:(1) Research on segmentation and modification Algorithm for lung CT images sequence. Based on advanced 8-connected boundary tracking algorithm,8-step method is proposed for automatic accurate segmentation to sections of CT images sequence. To the lung parenchymal after segmentation, rolling ball method and Ray casting algorithm are individually used to modify the outside and inside edge of the lung, and automatically identify whether left and right lung are attached, and separate the attached lung parenchyma using up integral algorithm,and obtain intact lung parenchymal. Experimental results show that the algorithms this paper proposing can achieve the desired results.(2) Segmentation of pulmonary trachea tree and blood vessel tree. Firstly, based on the structure of the tracheal anatomy, using advanced adaptive 3-D region growth algorithm, this paper proposes the way of "point to point" to find seeds in the next image, and combines shape analysis and template matching method to segment high-precision trachea trees. According to the feature in gray value of blood vessel on the CT image, more intact blood vessel tree including th artery of the lung'shilus can be segmented based on improved SMDC-Connection Cost algorithm. Experimental results show that the proposed algorithms can achieve the expected result.(3) Through studying on the connection between pulmonary function indexs, pulmonary's texture and lung diseases, using the intact lung parenchymal after segmentation, experiments obtain four pulmonary function indexs and eigenvalues based on GLCM and fractal dimension. Through calculating and comparing to different patients'CT images, this paper gets the feasibility in the application of lung disease diagnosis.(4) Through deep integration of MFC and VTK, the built of platform for evaluation of pulmonary function is accomplished and realization of function is achieved baed on VC6.0.