基于医学图像处理的肝纤维化分期诊断研究

Nowadays the clinical diagnosis of hepatic fibrosis and cirrhosis is still a hard work. The purpose of this study is to develop the algorithms for noninvasive computer-aided diagnosis (CAD) of the stages of fibrosis, using integrates information from the patient such as image features, the position and shape of a lesion, or even physical characteristic. As Guangxi province has a high incident rate of liver cancer and CAD products on liver is very real, this research is to construct the mathematical model of fibrosis by analyzing the abdominal CT/MR images according to the situation of Guangxi, and provide accurate staging and early detection of cirrhosis. We proposed a fully automatic method to integrate the texture features to the morphology changes to overcome the shortcomings of the low accuracy detection rate by traditional approaches. This work mainly contains several parts: Segmentation of the liver region with tumor tissues; Staging the fibrosis by texture features; Establishing the shape model to describe the irregularity of liver surface; Classifying the fibrosis by changes of volume among the different segments; and Calculating the stiffness of liver using MR tag images.

目前肝纤维化、早期肝硬化的临床诊断比较困难。本课题的目标为研究计算机辅助诊断（CAD）算法来实现对肝纤维化的无创性分级诊断，它的依据主要来自于患者的图像特征，病变位置及形状，甚至物理特性等综合信息。广西是肝癌的重灾区，而现阶段肝脏 CAD 的产品极少。本课题将通过对肝脏的CT 和MRI图像进行计算机自动解析，结合适合于广西及周边肝病高发区的资料数据库及病理诊断信息，建立肝纤维化的数理模型，从而达到对肝纤维化的准确分级和肝硬化的早期诊断，以及对主要慢性病病人的长期动态监控等目的。我们提出的基于纹理特征结合边缘形态检测的算法将克服目前采用单一算法检测准确率偏低的不足。研究内容包括：肝脏领域的全自动抽出；基于纹理特征量的纤维化分级；基于肝脏边缘凹凸度的表面特征量提取及分类；基于肝脏变形程度的体积特征的纤维化分级及基于肝脏硬度特征的肝硬变程度的检出等。

目前肝纤维化、早期肝硬化的临床诊断比较困难。本课题的目标为研究计算机辅助诊断（CAD）算法来实现对肝纤维化的无创性分级诊断，它的依据主要来自于患者的图像特征，病变位置及形状，甚至物理特性等综合信息。广西是肝癌的重灾区，而现阶段肝脏 CAD 的产品极少。本课题通过对肝脏的CT 和MRI图像进行计算机自动解析，结合适合于广西及周边肝病高发区的资料数据库及病理诊断信息，建立肝纤维化的数理模型，从而达到对肝纤维化的准确分级和肝硬化的早期诊断，以及对主要慢性病病人的长期动态监控等目的。我们提出的基于纹理特征结合边缘形态检测的算法克服了目前采用单一算法检测准确率偏低的不足。研究内容包括：肝脏领域的全自动抽出；基于纹理特征量的纤维化分级；基于肝脏边缘凹凸度的表面特征量提取及分类；基于肝脏变形程度的体积特征的纤维化分级及基于肝脏硬度特征的肝硬变程度的检出等。