Imaging and image processing for integrated understanding of multi-modal medical images

建立可靠的多模式图像评价，构建与器官或组织生理动力学相对应的诊断支持系统，是先进医学和防护医学的重要研究内容。为此，成像和图像处理技术必须成为获得形态信息和功能信息，并利用互信息改善每一次观测的关键技术，开始实现新的功能图像和形态图像的CT成像技术。在同步辐射成像中，我们已经成功地利用目标器官上的荧光和康普顿散射现象来同时描述它们，并具有相当大的互信息效应。结合荧光X线CT和相位对比CT图像，可以较好地评价器官功能改变的情况。我们还开发和讨论了透射式光学CT，利用透射光的衰减系数和光谱展宽来表征人体手指关节等靶器官。为了临床评价心壁功能，我们已经开发了一个交互式三维显示系统，将SPECT的功能信息叠加到MRI的三维形态描述上，并成功地验证了综合呈现的有效性。为此，我们开发了一个自动处理程序，利用目标器官附近的清晰描述的器官，我们还开发了几个用于PET/MRI和SPECT/MRI图像融合的自动3D配准程序，并获得了一些用于定量临床评估的参数。进一步的努力现在集中在调整与临床使用的单个靶点相对应的工艺参数上。

For advanced and protective medicine, it is an important study to establish reliable multi-modal image evaluation and to construct a supporting system for diagnosis corresponding to the physiological dynamics of organ or tissue. On this purpose, imaging and image processing techniques must be key-technologies to obtain both morphological and functional information and to improve each observation by using mutual information.It is started to realize new CT imaging techniques for functional and morphological image. In synchrotron radiation imaging, we have succeeded to describe them simultaneously by using fluorescent and Compton scattering phenomena on the target organ with considerable effect of mutual information. In case of the combination with fluorescent X-ray CT and phase-contrast CT images, we can evaluate properly the situation of functional changes of organ. Trans-illumination optical CT is also developed and discussed to utilize attenuation coefficient and spectrum broadening of transmitted light to characterize target organ such as human finger joint.For clinical evaluation of cardiac wall function, we have already developed an interactive 3D display system where functional information from SPECT are superposed on 3D morphological description from MRI, and succeeded to demonstrate the validity of integrated presentation. Here we have developed an automatic processing procedure for this purpose by using clear described organs near the target organ.We have also developed several procedures in automatic 3D registration on image fusion for PET/MRI and SPECT/MRI images, and to get a few parameters for quantitative clinical evaluation. Further effort is now concentrated on tuning processing parameters corresponding individual target for clinical use.