肾功能评价模型改进及通用影像平台构建

发布时间：2018-07-08 15:13

  本文选题：肾小球滤过率 + 软件平台　； 参考：《东北大学》2014年硕士论文

【摘要】：利用动态增强核磁影像研究肾脏功能是目前的一个热点,因为肾脏功能的评价方法是肾脏疾病诊疗过程中的一个重要环节,核磁相对于CT或者核医学等方法又有着无辐射,无损伤,效率高的优势。肾小球滤过率(glomerular filtration rate, GFR)是评价肾脏功能的一个重要参数。现阶段基于动态增强核磁影像测定肾小球滤过率尚处于研究阶段,因此在模型推导,模型计算,图像分割,软件工具等方面还有很多不完善的地方。软件工具方面,本文考察了比较流行的几种免费的医学图像处理平台,由于动态增强核磁所采集的图像是4D(3D+时间)多序列的DICOM图像,目前的软件平台大多数在导入多序列图像时会出现问题,也缺乏对4D图像分析的支持。针对这些需求同时考虑到软件的可扩展性,本文设计并实现了一个拥有插件系统的医学图像处理平台,支持图像导入、分割、显示,时间-密度曲线绘制,GFR计算等功能,并且提供软件接口可以供其他开发者开发插件扩展本平台的功能。模型方面,本文考察了目前国内外学者对于肾脏滤过系统所建立的数学模型,这些模型大多有参数多不稳定或者考虑因素较少等缺点,因此还有着很大的改进余地。本文也同时探讨了一种利用体素的时间密度曲线来分割肾脏的血管房室的思路,并利用该分割结果去简化目前的三室法模型。最后利用本文构建的软件平台在天津医科大学提供的图像上测试了改进后的分割.算法和计算GFR的模型,并将计算结果与目前的GFR金标准做了对比,证明了改进后的模型准确度更高。
[Abstract]:The use of dynamic enhanced nuclear magnetic resonance imaging to study renal function is a hot topic at present, because the evaluation method of renal function is an important link in the diagnosis and treatment of renal diseases, and nuclear magnetic field has no radiation in comparison with CT or nuclear medicine. No damage, high efficiency advantage. Glomerular filtration rate (glomerular filtration rate,) is an important parameter in evaluating renal function. At present, glomerular filtration rate measurement based on dynamic enhanced nuclear magnetic imaging is still in the research stage, so there are still many imperfections in model derivation, model calculation, image segmentation, software tools and so on. In terms of software tools, several popular medical image processing platforms are investigated in this paper. Because the images collected by dynamic enhanced nuclear magnetic field are 4D (3D time) multi-sequence DICOM images, Most of the current software platforms have problems in importing multi-sequence images and lack of support for 4D image analysis. Considering the extensibility of the software, this paper designs and implements a medical image processing platform with plug-in system, which supports image import, segmentation, display, time density curve drawing and GFR calculation, etc. And provides the software interface for other developers to develop plug-ins to extend the functionality of the platform. In terms of models, this paper examines the mathematical models established by scholars at home and abroad for renal filtration systems. Most of these models have many disadvantages such as unstable parameters or few factors, so there is still much room for improvement. At the same time, this paper also discusses a method of using the time density curve of voxel to segment the vascular atrioventricular of the kidney, and simplifies the current three-compartment model with the result of the segmentation. Finally, the improved segmentation is tested on the images provided by Tianjin Medical University using the software platform constructed in this paper. The algorithm and the model for calculating GFR are compared with the current GFR gold standard, and the improved model is proved to be more accurate.
【学位授予单位】：东北大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：R692;R445.2
，

本文编号：2107839