Temporal change detection of three dimensional CT images using non-rigid co-registration

本研究的目的是在两次连续CT扫描之间生成减影CT图像。在第一步中，我们新开发了一个带有大型廉价数据服务器的DICOM图像数据库，以保存大量临床薄层CT数据供研究之用。在新的服务器上存储了超过30TB的CT图像数据，并且三年没有发生数据丢失。减影图像的生成使用了非刚性图像配准软件，但非刚性方法并不能很好地配准周围解剖结构。更精确的联合注册被认为是临床使用所必需的。

The purpose of this study was to generate subtracted CT image between two serial CT scans. In the first step, we newly developed a DICOM image database with a large and cheap data server to save a lot of clinical thin slice CT data for research. Over 30TB CT image data were stored in the new server and data loss was not happened for three years. A non-rigid image co-registration software was used for the subtracted image generation, but peripheral anatomy was not perfectly co-registered by the non-rigid method. More precise co-registration was thought to be necessary for the clinical use.