基于Web架构的医学影像三维可视化处理研究

发布时间：2018-01-17 00:16

本文关键词：基于Web架构的医学影像三维可视化处理研究　出处：《中国科学院研究生院(上海技术物理研究所)》2016年博士论文　论文类型：学位论文

【摘要】：本论文根据当前医学影像信息临床需求、医疗“互联网+”的发展趋势以及本实验室多年来在医学影像三维可视化领域的工作积累,开展了可在“云计算”环境下运行基于Web架构的医学影像三维可视化以及病人医疗信息3D可视化显示系统设计及关键技术研究,通过国际合作研发了肺/肝/脑肿瘤3D定量化诊断组件以及自主研发了脑出血定位诊断3D可视化算法与组件,主要研究内容和创新点如下:(1)研究设计了基于Web架构的医学影像三维可视化系统原型机,该样机具有医学影像存储、处理、分析及可视化计算等功能在“云端”(服务器端)实现、而3D图像显示操作在客户端进行的新型医学影像三维可视化技术特性。该系统主要由图形用户界面GUI、客户端/服务器端通信层、服务器端应用层组成,其中图形用户界面GUI基于HTML5实现,具有瘦客户端、零足迹、跨平台的优点;客户端/服务器端通信层基于WebSocket协议实现,全双工通信,提升用户交互体验;服务器端应用层基于GPU加速技术优化可视化算法,能够快速响应客户端请求。(2)设计了一种2D/3D图像处理集成中间件,使得本文研发的三维可视化功能可与临床PACS二维可视化系统AxVS进行无缝集成,该中间件基于Java和Python语言实现,通过Java开源缓存框架Ehcache与PostgreSQL数据库协同工作,降低了影像数据库的负担,提升了用户的访问体验。(3)将基于Web架构的医学影像三维可视化系统移植到病人医疗信息可视化索引系统VIS,对基于Web架构的病人医疗信息3D可视化显示技术展开了探索性研究,初步实现了“云端”(服务器端)解析及绘制3DS可视化人体模型,然后将每帧绘制结果传送至客户端浏览器进行显示,图形用户界面GUI支持用户对可视化人体模型旋转、缩放、平移、部位选取高亮等操作。(4)通过国际合作开发了“肺/肝/脑肿瘤3D定量化诊断组件”以及自主研发了“脑出血定位算法与软件组件”,将这两种组件与本实验室研发的PACS医学影像三维可视化组件MedViewCtrl集成并进行了初步临床测试评估。肺/肝/脑肿瘤3D定量化诊断组件主要针对肺/肝/脑肿瘤3D定量化诊断,用户通过选取病灶ROI,然后调用相关分割算法实现对病灶的3D分割,最后通过三维可视化算法重建得到肿瘤模型,计算出其一系列定量化指标(如肿瘤体积、密度等)作为医生诊断的依据。脑出血定位算法与软件组件主要是帮助医生实现脑出血区域的定位,用户在多平面重建MPR的帮助下找到感兴趣切面,然后通过该组件提供的测量控件计算出颅脑半径、额距、颞距和高距,方便医生临床快速对脑出血患者实行救治。
[Abstract]:According to the current clinical demand of medical image information, the development trend of medical "Internet" and the accumulation of our laboratory's work in the field of 3D visualization of medical image for many years. The design and key technology of 3D visualization system of medical image based on Web architecture and 3D visualization display system of patient medical information are carried out in "cloud computing" environment. The 3D quantitative diagnosis module of lung / liver / brain tumor and 3D visualization algorithm and module of location diagnosis of intracerebral hemorrhage were developed by international cooperation. The main research contents and innovations are as follows: 1) A prototype medical image 3D visualization system based on Web architecture is designed. The prototype has medical image storage and processing. The functions of analysis and visual computing are realized in the "cloud" (server side). This system is composed of GUI, client / server communication layer and server-side application layer. The graphical user interface (GUI) is implemented on the basis of HTML5, which has the advantages of thin client, zero footprint and cross-platform. Client / server communication layer is based on WebSocket protocol, full duplex communication, improve user interaction experience; A 2D / 3D image processing integration middleware is designed in the server application layer, which is based on the GPU acceleration technology to optimize the visualization algorithm, which can quickly respond to the client's request. The 3D visualization function developed in this paper can be seamlessly integrated with clinical PACS two-dimensional visualization system AxVS. The middleware is based on Java and Python language. Through Java open source caching framework Ehcache and PostgreSQL database work together, reduce the burden of image database. Improve the user's access experience. 3) Transplant the medical image 3D visualization system based on Web architecture to the patient medical information visual index system (VIS). The 3D visualization display technology of patient medical information based on Web architecture is studied, and the "cloud" (server side) analysis and rendering of 3DS visual human model are preliminarily realized. Then the rendering results of each frame are transmitted to the client browser for display. The graphical user interface (GUI) supports users to rotate scale and translate the visual human model. By international cooperation, we developed a 3D quantitative diagnostic module for lung, liver and brain tumors, and developed a localization algorithm and software module for intracerebral hemorrhage (ICH). The two components were integrated with MedViewCtrl, a 3D visualization module for PACS medical images developed by our laboratory, and the preliminary clinical tests were evaluated. A 3D quantitative diagnosis Group for Lung / liver / brain tumors. Three dimensional quantitative diagnosis of lung / liver / brain tumors was performed. The user selects the focus ROI, then calls the correlation segmentation algorithm to realize the 3D segmentation of the focus. Finally, the 3D visualization algorithm is used to reconstruct the tumor model. A series of quantitative indexes (such as tumor volume, density, etc.) are calculated as the basis of doctor diagnosis. ICH localization algorithms and software components are mainly to help doctors achieve the location of intracerebral hemorrhage region. With the help of multi-plane reconstruction MPR, the user finds the section of interest, and then calculates the brain radius, frontal distance, temporal distance and high distance through the measurement control provided by the component. It is convenient for doctors to treat patients with intracerebral hemorrhage quickly.
【学位授予单位】：中国科学院研究生院(上海技术物理研究所)
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TP391.41

【相似文献】