基于水平集方法的医学影像血管分割

发布时间：2018-06-26 19:45

本文选题：医学影像分割 + 磁共振图像　；参考：《北京化工大学》2017年硕士论文

【摘要】：近年来,医学成像技术的快速发展使人们能够获得大量高分辨率的医学影像数据,分别来自X射线、CT、超声及MRI等,这些成像技术现已在医学诊断、治疗及术前规划等阶段得到了广泛的应用,研究如何从医学影像中获得的我们所需要的信息和数据进行分析,使这些宝贵的信息得以有效利用,是具有十分重要的意义。在医学图像的处理和分析中,医学图像分割尤为关键,它能够提取医生所需要的结构和组织,为医生诊断提供帮助或制定更为精确的手术计划。本文在分析医学图像分割领域的研究现状的基础上,研究了基于水平集的肝血管图像分割技术。首先对主动轮廓模型和水平集算法的基本理论进行了系统的解释和研究,分析了肝脏血管磁共振图像的特点及其分割的局限性,以及传统的水平集分割方法存在的优点和缺陷。同时,针对肝脏血管磁共振图像血管末梢对比度低的问题,提出了改进的方法:结合非局域稳健统计量和偏移校正,通过提高种子点信息的稳健性并降低图像亮度不均匀产生的影响来提高分割的精确度,改进后的水平集算法提高了血管末梢的分割精确度。此外,本文还研究了基于大津阈值法和马尔科夫随机场的大脑图像分割技术。本文针对大脑磁共振图像病变区域边缘不清晰的问题,对马尔科夫随机场提出了改进的方法:结合马尔科夫随机场和迭代大津阈值法,分割不同成像方法得到的病变区域和正常组织,并对混叠的边缘进行分类,改进后的分割算法提高了病变区域提取的精确度。
[Abstract]:In recent years, the rapid development of medical imaging technology has enabled people to obtain a large number of high-resolution medical image data, respectively from X-ray CTS, ultrasound, MRI and so on. These imaging techniques have been used in medical diagnosis. Treatment and preoperative planning have been widely used. It is of great significance to study how to analyze the information and data we need from medical images so that these valuable information can be used effectively. In medical image processing and analysis, medical image segmentation is particularly important. It can extract the structure and tissue needed by doctors, provide help for doctors' diagnosis or make more accurate surgical plans. Based on the analysis of the current situation of medical image segmentation, the level set based hepatic vascular image segmentation technique is studied in this paper. Firstly, the basic theory of active contour model and level set algorithm is systematically explained and studied. The characteristics of hepatic vascular magnetic resonance imaging and its segmentation limitations are analyzed, as well as the advantages and disadvantages of traditional level set segmentation methods. At the same time, aiming at the problem of low contrast of vascular endings in hepatic vascular magnetic resonance imaging, an improved method is proposed: combining non-local robust statistics and offset correction, The accuracy of segmentation is improved by improving the robustness of seed point information and reducing the influence of uneven image brightness. The improved level set algorithm improves the segmentation accuracy of vascular endings. In addition, the brain image segmentation technique based on Otsu threshold method and Markov random field is studied. In this paper, an improved method of Markov random field is proposed to solve the problem that the edge of the lesion region is not clear in magnetic resonance image of the brain: combining the Markov random field with the iterative threshold method, The lesion regions and normal tissues obtained from different imaging methods are segmented and the edges of the aliasing are classified. The improved segmentation algorithm improves the accuracy of the lesion region extraction.
【学位授予单位】：北京化工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R445;TP391.41

【相似文献】