基于CT图像的左心室建模及参数统计分析

发布时间：2018-02-27 15:37

本文关键词： 心脏CT图像左心室点分布模型动态参数统计分析　出处：《浙江工业大学》2012年硕士论文　论文类型：学位论文

【摘要】：左心室的泵血功能维持着人体的血液循环，因而对左心室的运动分析可以辅助诊断心脏疾病。通过医学图像建立左心室三维模型，然后基于模型来分析左心室的运动状态和评价心室功能成为近年来的研究热点，而通过对反映心室局部功能的动态参数进行统计分析可以实现疾病的早期诊断和治疗，因此相关的研究工作尤为重要。心脏的CT图像是临床诊断的重要手段与依据，图像中左心室内外壁轮廓为不规则椭圆，且造影时由于左心房对左心室的遮挡使心室内外壁轮廓不闭合，因此，单纯基于灰度对图像进行搜索和匹配建模的方法并不适合心脏的CT图像。现有的文献中缺乏针对心脏CT图像的心脏运动分析研究成果，并且缺少对心脏的动态参数统计分类心脏疾病的研究。本文基于心脏CT图像进行了左心室建模，提取了左心室的动态参数，并对动态参数做了统计分析来评价心脏功能。本文主要的工作如下： 1．提出了一种基于心脏CT图像建立左心室点分布模型的方法。该方法主要包括手动分割，点云获取，NURBS表面拟合，重采样和点云匹配这五个步骤。其中根据左心室的解剖结构和心肌分布的特点对手动分割进行了轮廓标记的优化，并且根据控制点的分布特征对NURBS拟合进行了点序列的优化，以物理相位时间划分为依据，可以得到7个主要相位时刻的左心室点分布模型。 2．针对得到的左心室点分布模型，提取了左心室心肌扭转角度和扭转角速度两个动态参数，并且计算了心肌速度，加速度，扭转角及扭转角速度这几个参数在不同物理相位时刻的值。 3.将点分布模型中计算出的左心室动态参数与临床上统计的范围值作对比，根据临床上的运动异常状态分析左心室运动情况。并且由于数据源对于心脏是否病变存在缺失，，因此引入了无监督的聚类方法对动态参数进行统计分类，分析了心脏运动异常的疑似区域。
[Abstract]:The pump function of the left ventricle maintains the blood circulation of the human body, so the motion analysis of the left ventricle can assist in the diagnosis of heart disease. Then it is a hot topic to analyze the motion state of left ventricle and evaluate ventricular function based on the model. The early diagnosis and treatment of the disease can be realized by statistical analysis of dynamic parameters reflecting the regional function of the left ventricle. Therefore, the relevant research work is particularly important. The CT image of the heart is an important method and basis for clinical diagnosis. The contour of the left ventricular wall is irregular ellipse in the image, and the contour of the inner and outer wall of the left ventricle is not closed because of the occlusion of the left atrium to the left ventricle. The method of searching and matching the image based on gray level is not suitable for the CT image of the heart. In this paper, the left ventricular model is established based on the CT images of the heart, and the dynamic parameters of the left ventricle are extracted, and the dynamic parameters are statistically analyzed to evaluate the heart function. The main work of this paper is as follows:. 1. A method of establishing a left ventricular point distribution model based on cardiac CT image is proposed, which includes manual segmentation, point cloud acquisition, and surface fitting of Nurbs. Resampling and point cloud matching are five steps. According to the anatomical structure of left ventricle and the characteristics of myocardial distribution, the manual segmentation is optimized by contour marking, and the point sequence of NURBS fitting is optimized according to the distribution characteristics of control points. Based on the physical phase time division, the distribution model of left ventricular points at 7 main phase times can be obtained. 2. Two dynamic parameters of left ventricular torsion angle and torsion angular velocity were extracted according to the model of left ventricular point distribution, and the myocardial velocity and acceleration were calculated. The values of torsion angle and torsional angular velocity at different physical phase time. 3. The dynamic parameters of the left ventricle calculated in the point distribution model were compared with the range values of the clinical statistics, and the left ventricular motion was analyzed according to the abnormal state of the clinical motion. So the unsupervised clustering method is introduced to classify the dynamic parameters and analyze the suspected areas of abnormal cardiac motion.
【学位授予单位】：浙江工业大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：R814.42

【参考文献】