Intelligent CAD Based on Quantitative Analysis of Spatio-Temporal Change in Medical Image Sequence

我们的挑战是开发一个智能的计算机辅助诊断（CAD）系统的基础上定量分析的时空变化的高分辨率多维图像序列。我们也专注于4d影像的成像与影像处理，以及协助观察连续多维影像的影像资料库系统。虽然我们应该开发很多技术来实现我们的目的，但我们不能在项目的有限时间内做所有的事情，然后，我们主要集中在以下三个系统，并取得了成果。利用4D MR标记图像进行心功能分析的CAD（1）我们研制了回顾性心脏-呼吸双门控MR标记系统。利用我们新开发的k-t插值方法对所获得的数据进行了校正和重建。我们还利用TARD和丽莎技术开发了无瞬态伪影的4D标记序列。(2)提出了一种在5d（x，y，z，t，I）空间中进行标签检测和分析的新方法。(3)获得了约100个高分辨率4d胸部MRI数据集.人类胚胎序列三维图像数据库及其数据检索与显示系统（1）开发了胚胎序列三维图像数据库的原型系统，该系统可以用于观察不同发育阶段或不同受试者之间的差异。看来，该系统的概念可以用于诊断图像序列的CAD系统。(2)提出了一种新的多维图像描述方法（RBCT），该方法适用于多维图像的分割和显示. 3D X线CT图像主动脉改变的计算机辅助设计系统我们研制了主动脉改变的计算机辅助设计系统。该系统具有主动脉区域分割、主动脉周围钙化区域检测、主动脉直径、钙化区域数量和体积等参数测量等功能。该系统作为PLUTO的插件系统实现。

Our challenge is to develop an intelligent computer aided diagnosis (CAD) systems based on quantitative analysis of spatio-temporal change in high resolution multidimensional image sequence. We also focused on imaging and image processing of 4d images and on image database system for aiding observing sequential multidimensional images. Although we should develop a lot of techniques to achieve our purpose, we could not do everything in the restricted period of the projects and we, then, mainly focused on the following three systems and obtained results.1. CAD for analyzing cardiac function by using 4d MR tagging images (1) We developed retrospective cardiac-respiratory dual gating system for MR tagging. Obtained data were corrected and reconstructed by our newly developed k-t interpolation method. We also developed transient artifact free 4d tagging sequence by using TARD and LISA technique. (2) We developed a new tag detection and analysis method in 5d (x, y, z, t, I) space. (3) About 100 data set of high resolution 4d chest MRI were obtained.2. Sequential 3d image database and its data retrieval and displaying system of human embryos (1) We developed prototype system, and it was found that the system was useful for observing difference between subjects or between different growing stages. It seems that concept of this system can be used for CAD system for diagnosing image sequence. (2) We developed a new multidimensional image description method (RBCT), and it was useful for segmentation and displaying of multidimensional images.3. CAD system for aortic change in 3d X-ray CT imagesWe developed CAD for observing aortic change. The system has functions of segmenting aortic region, detecting calcification region around aorta, and measuring several parameters (aortic diameter, number and volume of calcification region, and so on). The system was implemented as a plug-in system for PLUTO.