基于多源信息融合技术的一体化TOF-PET/MRI衰减校正方法研究

一体化PET/MRI是用于疾病诊断和生命科学研究的新型双模态成像技术。其中光子的衰减问题极大地影响PET的重建图像质量和定量精度；由于缺少校正源和CT图像，使得PET/MRI的衰减校正问题成为关键性难题。国际上现有的方法都存在一定的局限性。本项目采用多源信息融合技术，在不增加透射源和额外MRI扫描规程条件下研究并实现PET/MRI衰减校正。以蒙特卡罗模拟和实验相结合的方法进行研究。通过作用深度信息建模和511keV光子向下散射干扰校正对PET探测器镥本底符合数据进行优化；基于镥本底符合数据实现MRI线圈与其标准衰减模板配准；建立多源信息融合模型，充分利用镥本底符合数据、飞行时间(TOF)信息、MRI图像、扫描床与线圈衰减模板，进行联合重建，获取更为精确的衰减信息。项目预期可以有效提高PET/MRI同时成像系统的成像质量，具有重要的临床医学应用价值，期望能够进一步推进PET/MRI的发展。

Integrated PET/MRI is a novel dual-modality imaging technology for disease diagnosis and life science research. The attenuation problem significantly affects the image quality and quantitative accuracy of PET. How to perform attenuation correction is a major technical challenge in PET/MRI due to the lack of calibration source and CT images. In general the existing methods also have some limitations. In this program, we propose to introduce multi-source information fusion technology for PET/MRI attenuation correction, without additional transmission sources or additional MRI scanning procedures. This study will be carried out by combining Monte Carlo simulations and imaging experiments. The research content include: optimizing the lutetium intrinsic coincidence data of PET through depth information modeling and 511 keV photons down-scatter correction; registration of the RF coil images and its standard attenuation template based on intrinsic coincidence data; establishing the multi-source information fusion model with the full information of the lutetium intrinsic coincidence data, TOF information, MRI images, scanning bed and coil attenuation template, to acquire more accurate attenuation coefficients. The proposed method is expected to improve the image quality of PET/MRI that has significant clinical value. Hope our research could promote the development and progress of PET/MRI technology.