中国科学院分子影像重点实验室知识库

PET/CT image fusion has emerged as a new and promising research area in recent years, whereas the current standard method for PET/CT fusion, Alpha-blending, often blurs fine anatomical structure. In this paper, a new fusion method based on the Pansharp model is presented. Our proposed method consists of first up-scaling the PET image to the resolution of the CT data by bilinear interpolation and registering them with mutual information, then integrating the anatomical information into the original PET image and transforming it to a multi-channel RGB image using a color table, and then simulating the panchromatic image with multiple regression analysis, and finally obtaining the fused image according to the Pansharp model. Our method is verified based on 45 real data sets and both qualitative and quantitative analyses of the fusion results demonstrate theeffectiveness and robustness of the proposed method.