The Design And Implementation Of Lung Parenchyma Segmentation And Tube Path Extraction Algorithms

With the flourish development of computer application techniques, many high precision and fast digital medical imaging devices were well up, and the Computer-aided Diagnosis (CAD) which based on these techniques was popularized rapidly and applied in the clinical diagnosis. Computer-aided Diagnosis is a complex process which include image enhancing, segmentation, pathology area detection, three-dimensional display and so on, and some CAD systems also provide virtual endoscopy function. These function models can satisfy doctor’s diversified needs for aided diagnosis.Against the background above, this pater select CT image as source data, mainly research the lung structures, design and implementate the lung segmentation and tube path extraction algorithms that involved in the CAD, and get better applications.First, pater discuss the methods of image segmentagion, design and implementate a segmentation algorithm which based on watershed and combined with region adjacency graph for lung parenchyma, the experiment testify this algorithm can segment the contour of lung parenchyma, and we got a better effect, this is one of the most important steps of lung disease detection in CAD. Then paper research the algorithms of image interpolation, after segment the trachea contour by region growing algorithm, design and implementate a interpolation algorithm based on shape for trachea, the experiment improve this algorithm can reconstruct smooth and completely three-dimensional trachea model, and finally we extract the three-dimensional trachea, the result can used in some visible detection. At the end of this pater, we research the algorithms of center path extraction of tubular objects, design and implementate a center path extraction algorithm based on distance transform, and experiment it on the three-dimensional trachea model we segment before, the experiment testifiy this algorithm can extract ideal trachea center paths, and provide the basic for following virtual endoscopy research.