基于肺实质分割的肺纤维化辅助诊断系统研究与实现

发布时间：2018-08-03 10:29

【摘要】：肺纤维化疾病是威胁人体健康的四大疾病之一，严重影响着患者的生活和工作。目前尚无有效方法来治疗，尽早发现病情，控制纤维化程度至关重要。医生常通过肺功能检查来发现病情，患者很有可能被误诊，导致错过治疗时间。因此开发一种可以辅助检测纤维化疾病的系统，就显得非常必要。而如何实现肺部CT图像准确分割和三维重建，以准确进行纤维化定量分析，辅助医生诊断病情并尽早制定治疗方案，是需要深入研究的基础和关键。本文重点研究了医学图像分割与三维重建算法，借助医学图像处理技术，在Matlab平台上搭建了肺纤维化辅助检测系统，根据肺部图像信息特征来进行定量分析，并进行病人资料记录。本文所做的主要工作如下： ①在图像分割方面，介绍了阈值法、区域生长法、边缘检测法和基于特定理论的方法等常用的分割方法，通过在Matlab环境下运用各个算法对肺部CT图像进行了分割，分析了各个算法的优缺点。针对传统方法很难满足课题要求的问题，，提出了基于改进Snake模型的肺部图像分割算法。将原图像进行阈值分割得到大致边界，然后利用Live-wire模型得到初始轮廓，最后利用Snake模型提取肺实质。文章分别对Live-wire算法和Snake模型的缺点进行了相关的改进。利用图像分割算法的评价方法，对改进的Snake算法进行了评价和分析。 ②在三维重建方面，介绍了三维重建中的面绘制和体绘制两种方法，分析了这两种方法的特点，最后在Matlab环境下对肺部序列图像使用体绘制方法进行了三维重建。 ③在肺纤维化辅助检测方面，根据分割得到的肺实质，从灰度特征、肺野面积等定量分析肺纤维化；根据得到的肺三维模型从不同角度观察有无纤维化病灶。 ④在Matlab环境下对系统进行了设计与开发。主要对图像读取显示、图像分割、三维重建、定量分析、数据库资料记录等模块进行了设计实现。本文通过研究图像分割和重建，构造了肺纤维化辅助诊断系统，能辅助医生定量分析肺纤维化疾病，准确诊断轻度肺纤维化、进展期肺纤维化和重度纤维化，取得了良好的效果。
[Abstract]:Pulmonary fibrosis is one of the four major diseases threatening human health, which seriously affects the life and work of patients. At present, there is no effective method to treat the disease. It is very important to find the disease and control the degree of fibrosis as soon as possible. Doctors often use pulmonary function tests to find the condition, patients are likely to be misdiagnosed, resulting in missed treatment time. Therefore, it is necessary to develop a system that can assist the detection of fibrosis diseases. How to realize accurate segmentation and 3D reconstruction of lung CT images, to accurately carry out quantitative analysis of fibrosis, assist doctors to diagnose the disease and formulate treatment plans as soon as possible, is the foundation and key to further study. This paper focuses on the medical image segmentation and 3D reconstruction algorithm. With the help of medical image processing technology, a pulmonary fibrosis detection system is built on the Matlab platform, and quantitative analysis is carried out according to the characteristics of lung image information. Patient data were recorded. The main work of this paper is as follows: 1 in the aspect of image segmentation, some common segmentation methods, such as threshold method, region growth method, edge detection method and method based on specific theory, are introduced. In this paper, lung CT images are segmented in Matlab environment, and their advantages and disadvantages are analyzed. Aiming at the problem that the traditional method is difficult to meet the requirements of the subject, a lung image segmentation algorithm based on improved Snake model is proposed. The original image is segmented by threshold, then the initial contour is obtained by using Live-wire model, and the lung parenchyma is extracted by Snake model. In this paper, the shortcomings of Live-wire algorithm and Snake model are improved. Using the evaluation method of image segmentation algorithm, the improved Snake algorithm is evaluated and analyzed. 2 in 3D reconstruction, two methods of surface rendering and volume rendering are introduced, and the characteristics of these two methods are analyzed. Finally, 3D reconstruction of lung sequence image using volume rendering method was carried out under Matlab environment. 3 in the aspect of pulmonary fibrosis assistant detection, quantitative analysis of pulmonary fibrosis was carried out according to the segmented lung parenchyma, such as gray feature, lung field area and so on. According to the three dimensional lung model obtained, the fibrosis lesions were observed from different angles. 4 the system was designed and developed under Matlab environment. The modules of image reading and display, image segmentation, 3D reconstruction, quantitative analysis, database recording and so on are designed and implemented. Through the study of image segmentation and reconstruction, an auxiliary diagnostic system for pulmonary fibrosis is constructed, which can assist doctors in quantitative analysis of pulmonary fibrosis diseases, accurate diagnosis of mild pulmonary fibrosis, progressive pulmonary fibrosis and severe fibrosis. Good results have been achieved.
【学位授予单位】：重庆大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP391.41;R563

【相似文献】