CT图像肺肿瘤分割系统的设计与实现

发布时间：2018-07-29 06:51

【摘要】：众所周知,癌症已成为人类死亡的一大重要原因,而肺癌更是位列恶性肿瘤死亡原因之首。肺肿瘤的分割是计算机辅助诊断的重要组成部分。目前,临床上主要通过医生手工勾勒的方式对肺肿瘤进行分割,然而这种方式不仅效率低下,且具有一定的主观性。近年来,人们对肺肿瘤的分割进行了大量研究,由于肿瘤的多样性,使得现阶段仍然不能使用一种算法将所有类型的肺肿瘤准确、自动地分割出来。因此,本文通过对肺肿瘤分割算法的研究,利用VTK和Qt工具包设计并实现了一个完整的肺肿瘤分割系统。本文主要工作内容如下:(1)采用边界约束的区域生长算法对肺肿瘤进行交互分割,实现了粘连型肺肿瘤的二维分割,解决了区域生长算法的过分割问题,通过大量的分割实验与医生手工分割结果进行对比,验证了分割结果的准确性。(2)根据肺肿瘤分割系统的实际需求,将系统分为读入保存模块、显示模块、交互模块和分割模块。其中读入保存模块通过VTK接口实现了图像数据的统一管理;显示模块在实现图像二维及三维可视化的同时,还实现了缩放等交互显示功能;交互事件响应模块实现了系统外部事件及内部事件的检测与响应;分割模块通过人机交互实现了肺肿瘤的二维分割和孤立型肺肿瘤的三维分割及三维重建。(3)在系统中通过VOI的提取和多种子点的选择,解决了区域生长算法在三维CT图像的数据量较大时速度过慢的问题,实现了孤立型肺肿瘤的三维分割和三维重建,通过等值面阈值调节工具给用户提供了最佳的三维重建显示效果。本文最后完成了系统的功能测试,并使用系统进行分割实验。结果表明,本文所设计的系统可以达到预期的设计目标,不仅对肺肿瘤具有良好的分割效率,而且有一定的应用价值。
[Abstract]:As we all know, cancer has become an important cause of human death, and lung cancer is the leading cause of death from malignant tumors. Lung tumor segmentation is an important part of computer aided diagnosis. At present, lung tumors are segmented by doctors' manual drawing. However, this method is not only inefficient, but also subjective. In recent years, people have done a lot of research on the segmentation of lung tumors. Due to the diversity of tumors, it is still not possible to use an algorithm to segment all types of lung tumors accurately and automatically. Therefore, through the research of lung tumor segmentation algorithm, a complete lung tumor segmentation system is designed and implemented by using VTK and QT toolkits. The main work of this paper is as follows: (1) the boundary constrained region growth algorithm is used to segment lung tumor interactively. The two-dimensional segmentation of adherent lung tumor is realized, and the over-segmentation problem of region growth algorithm is solved. The accuracy of the segmentation results is verified by comparing a large number of segmentation experiments with manual segmentation results by doctors. (2) according to the actual needs of lung tumor segmentation system, the system is divided into read and save module, display module, interactive module and segmentation module. The reading and saving module realizes the unified management of image data through the VTK interface, and the display module realizes the interactive display function such as zooming while realizing the visualization of 2D and 3D images. The interactive event response module realizes the detection and response of the external and internal events of the system. The segmentation module realizes the two-dimensional segmentation of lung tumor and three-dimensional reconstruction of solitary lung tumor through human-computer interaction. (3) extraction of VOI and selection of multiple seed points in the system. It solves the problem that the region growth algorithm is too slow when the data of 3D CT image is large, and realizes the 3D segmentation and reconstruction of solitary lung tumor. The optimal display effect of 3D reconstruction is provided by the equal-surface threshold adjustment tool. Finally, the function test of the system is completed, and the system is used for the segmentation experiment. The results show that the system designed in this paper can achieve the desired design goal, not only has a good segmentation efficiency for lung tumors, but also has certain application value.
【学位授予单位】：河北大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R734.2;TP391.41

【参考文献】