基于医学CT的人体肝脏三维建模关键技术研究与实现

发布时间：2018-06-29 04:07

本文选题：肝脏灰度分离 + 腹部CT图像　；参考：《浙江大学》2017年硕士论文

【摘要】：随着近几年来肝脏切除技术的飞速发展和人们生活水平的提高,个体化治疗是目前医疗的发展趋势,但是从肝脏二维医学影像到三维模型的重建仍是一个未能完全解决的问题。图像分割算法和三维软件建模为解决这一问题提供了有效的途径,其过程包括:肝脏原始数据处理、肝脏特征序列生成、肝脏模型分层重构以及肝脏实体模型打印。本文针对人体肝脏三维建模关键技术展开深入研究,主要成果如下:研究了现有肝脏特征提取算法无法完全提取肝脏边缘的理论根源,提出灰度分离的肝脏分割算法,加上对现有算法的整合和改进,成功地解决了肝脏边缘特征提取不明显以及无法只保留肝脏部分而不显示其它组织的问题,从而得到了肝脏特征序列,使得肝脏三维重建过程简化、建模时间极大的缩减以及肝脏模型准确性得到了进一步的保证。在肝脏分割算法的研究过程中,利用水平集算法提取出肝脏轮廓特征,然后通过灰度分离算法对轮廓内的区域进行处理,最后针对区域增长算法需要手动选取种子点的弊端,对其进行了改进,从而成功用改进后的区域增长算法自动将灰度分离后的图像变成仅含有肝脏特征而不含其它组织的图像,使算法的效率和精度得到了提升。针对肝脏三维建模需要手动修改蒙板、耗时巨大及精度无法保证等问题,分析了传统建模方法的不足,结合图像分割算法利用处理后的肝脏特征序列进行三维重建的系统,借助于Matlab提取并生成肝脏特征序列,导入Mimics后无需进行阈值分割和蒙板手动修改,选择所有灰度值不为0的区域即可直接计算出三维模型,最后可将三维模型打印出肝脏实体,从而解决了肝脏三维建模的效率和精度等问题。根据上述研究内容,以甲、乙两病人的腹部CT图像进行实验验证。首先,灰度分离算法处理效果良好,解决了肝脏边缘特征无法自动分割及分割效果不佳的问题;其次,先生成肝脏特征序列再进行分层重构的方法经试验使得肝脏三维模型重建的时间大大缩短;最后,通过肝脏实体模型3D打印完成了整个实验最终验证。
[Abstract]:With the rapid development of hepatectomy technology and the improvement of people's living standard in recent years, individualized treatment is the development trend of medical treatment at present, but the reconstruction from two-dimensional medical image of liver to three-dimensional model is still a problem that can not be solved completely. Image segmentation algorithm and 3D software modeling provide an effective way to solve this problem. The process includes: liver raw data processing, liver feature sequence generation, liver model hierarchical reconstruction and liver entity model printing. In this paper, the key technologies of 3D modeling of human liver are deeply studied. The main results are as follows: firstly, the theoretical roots of the existing liver feature extraction algorithms that can not completely extract the liver edges are studied, and a liver segmentation algorithm based on gray-scale separation is proposed. Combined with the integration and improvement of the existing algorithms, the problem of liver edge feature extraction is solved successfully, and the liver feature sequence is obtained, which can not only retain the liver part without displaying other tissues. The 3D liver reconstruction process is simplified, the modeling time is greatly reduced and the accuracy of the liver model is further guaranteed. In the research process of liver segmentation algorithm, the level set algorithm is used to extract the liver contour features, and then the gray level separation algorithm is used to process the region in the contour. Finally, aiming at the drawbacks of the region growth algorithm, we need to manually select the seed points. The improved region growth algorithm is successfully used to automatically transform the gray-scale image into an image with only liver features and no other tissues, which improves the efficiency and accuracy of the algorithm. In order to solve the problems such as the need to modify the mask manually, the time consuming and the precision can not be guaranteed, the deficiency of the traditional modeling method is analyzed, and the 3D reconstruction system based on the processed liver feature sequence is combined with the image segmentation algorithm. With the help of Matlab, the liver feature sequence is extracted and generated. After importing mimics, the 3D model can be directly calculated by selecting all regions with gray value not equal to zero, and the 3D model can be printed out to the liver entity without the need of threshold segmentation and manual modification of the mask. Thus, the efficiency and accuracy of liver 3D modeling are solved. According to the above study, abdominal CT images of patients A and B were used for experimental verification. Firstly, the gray level separation algorithm has good processing effect, which solves the problems that the liver edge feature can not be segmented automatically and the segmentation effect is not good. The method of stratified reconstruction of liver characteristic sequence is used to shorten the time of liver 3D model reconstruction. Finally, the whole experiment is verified by 3D printing of liver entity model.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R816.5;TP391.41

【参考文献】