基于颅部医学图像的三维重建技术研究

发布时间：2018-03-16 21:25

本文选题：医学图像　切入点：立方体算法　出处：《哈尔滨工程大学》2016年硕士论文　论文类型：学位论文

【摘要】：由于人体组织器官的形态多样、结构复杂,传统的二维断层数据已经不能满足人们在医学诊断方面的需求,使得三维重建技术成为了当今医学研究的一个热点。利用三维重建技术,将计算机断层扫描、核磁共振等二维医学数据构造成可交互的三维图像,可以多角度全方位显示人体内部的复杂结构。因此医学三维重建技术在医疗诊断、手术规划等方面具有很好的学术价值和应用前景。基于人体头部结构的复杂性,研究了体绘制、面绘制等三维重建算法,根据“可对头部二维医学图像进行三维重建、且能对重建结果切割”的需求,设计了一套三维医学重建系统。首先从数据源、算法、软件平台三个方面综述当前三维重建技术的发展,讨论系统设计中要使用的算法和算法平台;对Marching Cubes算法展开探索与研究,深入讨论了 Marching Cubes算法的二义性问题,并在渐近线判别法的基础上提出一种改进方法。接着,阐述了光线投射法的原理和流程,对传统的基于体素标量值和梯度幅值的二维体绘制传递函数进行了研究,针对其特征少、不能有效区分物质类别的缺点,设计了一种基于极限学习机的多维体绘制传递函数并进行实验,绘制效果得到明显改善。最后,设计开发了一个医学图像三维重建系统,并对多组医学数据进行实验。系统可以很好的实现病人信息的提取、单幅DICOM的显示和调整、面绘制、体绘制和虚拟切割等功能。
[Abstract]:Due to the diversity and complexity of human tissues and organs, the traditional two-dimensional tomography data can no longer meet the needs of medical diagnosis. 3D reconstruction technology has become a hot topic in modern medical research. By using 3D reconstruction technology, 2D medical data such as computer tomography, nuclear magnetic resonance and other two-dimensional medical data are constructed into interactive 3D images. Therefore, 3D medical reconstruction technology has good academic value and application prospect in medical diagnosis, surgical planning, etc. Based on the complexity of human head structure, Based on the requirement of "3D reconstruction of two-dimensional medical image of head and cutting of reconstruction result", a set of 3D medical reconstruction system is designed, which is based on data source, algorithm, and so on. This paper summarizes the development of 3D reconstruction technology in three aspects of software platform, discusses the algorithm and algorithm platform to be used in the system design, explores and studies the Marching Cubes algorithm, and discusses the ambiguity of Marching Cubes algorithm. An improved method is proposed on the basis of asymptote discriminant method. Secondly, the principle and flow of ray-casting method are described, and the traditional transfer function of two-dimensional volume rendering based on volume scalar and gradient amplitude is studied. In order to solve the problem that it has few characteristics and can not distinguish the material category effectively, a multi-dimensional volume rendering transfer function based on the ultimate learning machine is designed and experimented, and the rendering effect is improved obviously. A medical image 3D reconstruction system is designed and developed, and many groups of medical data are tested. The system can achieve the functions of patient information extraction, single DICOM display and adjustment, surface rendering, volume rendering and virtual cutting.
【学位授予单位】：哈尔滨工程大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：R318;TP391.41

【参考文献】