A3D Interactive Segmentation System Of Lung Nodules CT Images

Pulmonary nodule segmentation is the key step of computer aided diagnosis ofpulmonary nodules, its accuracy is related to the subsequent nodule quantification anddiagnosis conclusion credibility. But there is no one segmentation algorithm canaccurately segment pulmonary nodules from lung CT image, this paper presents a3Dinteractive segmentation system, allows users adjust the results manually based on theautomatic segmentation.Select the Visualization Toolkit (VTK, Visualization Toolkit) to develop3Dvisualization environment of the system. First of all, on the VTK encapsulation of various3D reconstruction algorithms are compared, on the basis of the experimental results tochoose proper algorithm for pulmonary nodules and other tissue reconstruction display,according to show the results of adjustment parameters and optimize the process, finally toachieve the optimal effect of reconstruction. Then, the design algorithm of the user on thepulmonary nodules were manually segmented operation for lung nodule segmentationresults, adhesion and segmentation is insufficient two cases were handled respectively.First, the nodules adhesion portion defines resection operation; the combination of theVTK package some display component (Widget) generates a spherical plane and severalremoval tool, and for these tools to define the corresponding response function of thenodules were excised, the user through the removal tool to specify nodules to removal of aportion of the system response function, and then call processing user operations tomodify the nodules, the final reconstruction display new nodules. In second, thesegmentation of insufficient nodules defines drag growth operation; a drag-and-dropoperation by the user by clicking the mouse and moving is completed, then the program inresponse to user operation to complete the nodule growth, mainly defined the followingfunction response: determines whether the user specified growth starting point andterminal point, determine which point to growth, growth pattern (growth direction andgrowth distance) and the growth process of the real time display.Through the tools and environment the system provided, Users can observe andsegment the pulmonary nodules by3D interaction, so the segmentation results will bemore accurate, the reliability of computer aided diagnosis will be improved.