基于不同图像分割法构建3D打印右室双出口模型  

Three-dimensional printing double-outlet right ventricle model based on different image segmentation methods

作　　者：胡立伟[1] 白凯[2] 钟玉敏[1] 孙爱敏[1] 王谦[1] 刘金龙 邱海嵊 朱铭[1] 

机构地区：[1]上海交通大学医学院附属上海儿童医学中心影像诊断中心,上海200127 [2]上海交通大学医学院附属新华医院小儿心血管科,上海200092 [3]上海市儿童先天性心脏病研究所,上海200127

出　　处：《中国医学物理学杂志》2016年第12期1272-1275,共4页Chinese Journal of Medical Physics

基　　金：上海市卫生局课题(20134306);上海市级医院新兴前沿技术项目(SHDC12015128);上海交通大学科技创新专项(YG2015QN25)

摘　　要：目的:通过区域生长法和阈值法分割CT图像并构建3D打印先天性心脏病模型,分析不同图像分割方法打印模型对心内结构显示的精准程度,并探讨3D打印对右室双出口疾病的诊断价值。方法:对1例两个月的先天性心脏病患儿行64排前门控增强CT扫描后,基于迭代算法进行重建。将CT的DICOM数据导入Mimics 17.0软件,分别使用区域生长方法进行血池分割以及通过阈值法提取心肌组织图像,得到不同分割图像之后,进行图像处理。模型A:包裹1 mm后对血池强化,进行挖空处理,构建表面模型;模型B:使用分割的心肌组织图像建模并通过平滑处理。将模型A和模型B生成光固化立体造型术文件数据,并将数据导入Objet 260 3D打印机后打印模型,完成建模过程。结果:两种不同分割方法构建的3D打印心脏模型都能清晰地显示异常的解剖结构,同时可观察室间隔缺损和两根大血管之间的空间位置关系。模型A和模型B的心内结构准确度没有明显差异,主动脉内径宽度分别为8.44、8.42 mm,肺动脉分别为12.81、12.73 mm,室间隔缺损分别为14.51、14.18 mm。模型A的圆锥组织比模型B的长度更长,分别为7.21、6.32 mm。结论:通过挖空和包裹等后处理方法,模型A可清楚地显示大血管的空间位置关系,在图像后处理过程中,3D打印模型可能忽略微小的解剖结构。模型B可精确地显示心脏内的解剖结构,包括心脏圆锥、三尖瓣位置、乳头状肌等。图像处理的关键是将DICOM数据转换成精确的3D打印的心脏模型。对于准确诊断复杂先天性心脏病,基于心脏CT打印的3D模型是一个非常有效的方法。Objective To analyze the accuracy of three-dimensional （3D） printing cardiac models with different image segmentation methods, and to discuss on the value of 3D printing model in the diagnosis of double- outlet right ventricle by segmenting CT image with region growing method and threshold method, and building a 3D printing model for congenital heart disease （CHD）. Methods A two-month-old child with CHD was scanned by 64-slice CT scanner （GE Discovery CT750 HD）. The obtained CT image was reconstructed based on iterative algorithm. The DICOM data of CT were input into Mimics 17.0 software. Region growing method and threshold method were used for segmenting blood pool image and extracting myocardial image, respectively. For model A, a 1 mm-thick wall was wrapped to the outside of the blood pool and the blood pool was hollowed for creating a surface model. For model B, segmented myocardium image was used for model building and smoothing. The 3D printer, OBJET 260, was utilized to print 3D models. Results Two 3D printing cardiac models with different segmentation methods demonstrated not only the anatomic abnormalities, but also the spatial position of ventricular septal defect and two great arteries. No difference was found between model A and model B in the width of inner diameter in the aorta （8.44 and 8.42 mm, respectively）, and pulmonary artery （12.81 and 12.73 mm, respectively）, and ventricular septal defect （14.51 and 14.18 mm, respectively）. The conus artery of model A was significantly longer than that of model B （7.21 and 6.32 mm, respectively）. Conclusion Model A clearly shows the spatial position of the great vessels through hollowing and wrapping, but the 3D printing model A might hide or mislead some intracardiac information during the post- processing. Model B precisely shows the intracardiac structure, including conus, the location of tricuspid valve, papillary muscle, etc. Converting the DICOM data into accurate 3D printing cardiac models is the key of image processing. Th

关 键 词：先天性心脏病 3D打印技术 图像分割 区域生长 阈值法 

分 类 号：R816.2[医药卫生—放射医学] Q819[医药卫生—临床医学]