Four-dimensional Image Processing Softwares for Computer Assisted Diagnosis or Ireatment Procedures

在这个项目中，我们主要关注心脏图像中的四维成像和图像处理方法。研究了边缘保持滤波、区域生长分割、三维-二维动态配准等四维图像处理方法。利用这些技术，我们还开发了一些计算机辅助诊断或治疗程序的软件。具体开发的方法或程序如下。快速心功能成像。提出了基于TrueFISP序列的多相三维MR标记方法。这种方法仍然受到工件问题的困扰。然而，它将在不久的将来得到解决。（伪影抑制在其他项目中进行了研究。）4维图像分割。开发了四维CT或MRI数据的自动提取和体积测量程序。2 .提出了基于分支的区域生长方法，用于提取三维MRA数据中的血管（空间三维+血管结构一维）。三维图像（CT/GSPECT）与x射线图像序列的动态配准。为了辅助主动脉支架移植，开发了一种术前3DCTA和术中透视图像序列注册系统。该系统是基于DRR（数字重建射线图）的3D-2D图像配准。开发了心脏门控SPELT与透视图像的3D-2D图像配准系统。在该系统中，通过评估心外膜表面（GSPECT）冠状动脉（2D x射线图像）的三维重建来进行配准。

In this project, we were concerned about 4-dimensional imaging and image processing methods mainly in cardiac images. We investigated about 4-dimensional image processing methods such as edge preserving filters, segmentation by using region growing, 3D-2D dynamic registration, and so on. By using these techniques, we also developed some softwares for computer assisted diagnosis or treatment procedures.In detail, developed methods or programs were as follows.1. Rapid Cardiac Functional Imaging.Multi phase 3D MR tagging method based on TrueFISP sequence was developed. This method still suffers from problem of artifacts. However, it will be solved in the near future. (Artifact suppression is investigated in other project.)2. 4-dimensional Image Segmentation.Automated extraction and volume measurement program from 4-dimensional CT or MRI data was developed. Branch based region growing method was also developed for extracting blood vessels in 3D MRA data(3-dimensions in space plus 1-dimension in structure of blood vessels).3. Dynamic Registration between 3D Image(CT/GSPECT) and X-ray Image Sequences.For assisting aortic stmt grafting, a system for registration of pre-operative 3DCTA and intra-operative fluoroscopic image sequence was developed. This system is based on DRR (Digitally Re-constructed Radiograph) based 3D-2D image registration. 3D-2D image registration system of Cardiac Gated SPELT and fluoroscopic images was also developed. In this system registration is performed by evaluate 3D reconstruction of coronary arteries(appeared in 2D X-Ray images) on epicaridal surface (appeared in GSPECT).