Development of an Image Analyzing System for Cardiovascular Morphorogy and Dynamics using Synchrotron Radiation Imaging

该项目的最终目标是在早期发现心血管异常。为此，同步辐射成像（SR）在寻找运动器官的精确结构方面是非常有用的，它给我们提供了物体的光谱信息。另一方面，我们可以从时间序列图像中得到更精确的器官结构和功能。在这里我们开发了一个系统和几个算法来实现这种分析。在2ms（即心脏运动可忽略的时间间隔）内获得两幅频谱图像，将图像数据序列与所需的参考时间序列信号记录在IC存储器中。每个图像数据为512x512大小，深度为12bit，通常两个频谱图像序列的时间差为2ms。为了快速校正图像失真和实时DSA，准备了快速数据处理板，并在与该记录系统相连的工作站上进行更复杂的处理。针对双翼图像时间序列的三维冠状结构识别、运动目标的检测与跟踪、光谱图像中器官特定物质的成像等相关实际问题，开发了几种用于器官评价的图像序列分析与信息提取的基本算法。本文还对SR成像光学系统的几个问题进行了研究。对于消能成像中散射x射线通量的抑制和x射线光谱的快速切换，我们找到了解决方案，并在继续实施。改进诊断算法和对该系统的临床评价是下一步要解决的问题。

The final goal of this project is to detect cardiovascular abnormality in its early stage. For this purpose synchrotron radiation (SR) imaging is quite useful to find precise structure on moving organs, and it gives us spectral information of objects. On the other hand, we can get more exact structures and functions of living organs from time sequence images. Here we developed a system and several algorithms to realize such analysis.Two spectrum images obtained in 2ms, i.e. ignorable time interval for cardiac motion, image data sequences are recorded in IC memory bank with required reference time series signal. Each image data is 512x512 size with 12bit depth, and the usual time difference between two spectrum image sequences is 2ms. For fast correction of image distortion and real time DSA, a fast data processing board is prepared and more sophisticated processing are executed on a work station connected to this recording system.Several basic algorithms for the analysis of image sequence and information extraction for evaluation of organs are developed on related practical problems, e.g. 3D coronary structure identification from biplane image time sequences, detection and tracking moving targets and imaging of specific materials in organs from spectrum images.A few problems on the optical system for SR imaging is also studied. For the suppression of scattering X-ray flux and the fast switching of X-ray spectrum for energy subtraction imaging, we found solutions, and their implementation are continuing. Improvement of diagnosing algorithm and clinical evaluation of this system is the next problem.