Research Of Vascular Segmentation Algorithms And3D Visualization For Thoracic CT Scans

With the rapid development of the medical imaging technology, the research on computer-aided diagnosis (CAD) system based on it is becoming a hot topic in the field. With the emergence of multi-slice spiral CT, the patients’ information provided by medical images becomes increasingly rich, and the medical imaging information that doctors need to process also increases dramatically. On the one hand, doctors’workload is increasing; on the other hand, it provides accurate information for the high-tech operation method.In the CAD system, the medical image segmentation is the key technology. Because of the complexity and diversity of the medical images, it’s difficult for the traditional segmentation method to achieve good results, especially for the segmentation of pulmonary vessels. The vessels have tiny structure, fuzzy vascular boundaries and are affected by the noise of CT itself, which makes the segmentation of pulmonary vessels become a challenge for lung segmentation. This thesis mainly studies the segmentation algorithms of the pulmonary vessels and the algorithms of three-dimensional visualization.In terms of the vessel segmentation, this thesis does research on the image preprocessing first, including image denoising and the segmentation of lung parenchyma images. Then extract the blood vessels of the lung parenchyma, including the enhancement algorithms respectively based on Hessian matrix and counter-proliferation model, which have been experimentally analyzed. The results show that the effect of vessels enhancement by Hessian matrix could extract vessels without of non-vessel tissues, but the computation is huge. The enhancement algrithms of counter-proliferation model has the advangtage of rapid and simple. So this method is available on vessels segmentation.The proposal of the level set method greatly contributes to the development of active contour models, the combination of level set method and curve evolution theory, and overcomes many disadvantages of traditional Snakes models. This thesis mainly analyses the segmentation of vessels with active contour models, which is based on the level set method. The method has an obvious advantage to the structure change of the topology with complex vessels. On this basis, it comes up with an improved point. The improved model expland the range of the contour detection, with which the segmentation can shows a better result.At last, this thesis has also studied the application of3D reconstruction technology. Based on the theoretical analysis of vessels segmentation and3D reconstruction, a medical image vessels segmentation platform system has implemented, which is combined with the medical image segmentation development tools of ITK and the visualization development tools of VTK. The platform system help get good visual effects towards the3D reconstruction of the segmentation result and human computer interaction.