Abstract
Purpose: In this study, we created a whole body fusion image of PET and MRI using automated software for fusion imaging, and assessed the accuracy of the software. Materials and methods: Twenty patients with abnormal FDG-PET findings underwent whole body MRI. Images from both modalities were automatically fused in two ways using software (Fusion Viewer, Nihon Medi-Physics Co., Ltd., Nishinomiya, Japan) with a registration algorithm based on maximum mutual information. One was to create a whole body fusion image at once, named whole body fusion (WBF). The other was to create fusion images of the head and body separately, named partial body fusion (PBF). Two radiologists measured the misregistration between PET and MRI in the fusion images at nine landmarks (brain, cervical spine, chest, heart, liver, right kidney, left kidney, vertebra, base of bladder). Results: When fusion images were created using WBF at once, misregistration was observed in the head and neck area in approximately half of the cases, whereas almost no misregistration was observed in the body. When fusion images were created using PBF, the misregistration in the head and neck areas was significantly smaller than in those using WBF, and misregistration in the body was very small. Conclusion: The PBF technique that creates highly accurate whole body PET/MRI fusion images is easy to use and may provide clinically useful information.
