Technical note: A PET/MR coil with an integrated, orbiting 511 keV transmission source for PET/MR imaging validated in an animal study.

BACKGROUND: MR-based methods for attenuation correction (AC) in PET/MRI either neglect attenuation of bone, or use MR-signal derived information about bone, which leads to a bias in quantification of tracer uptake in PET. In a previous study, we presented a PET/MRI specific MR coil with an integrated transmission source (TX) system allowing for direct measurement of attenuation. In phantom measurements, this system successfully reproduced the linear attenuation coefficient of water. PURPOSE: The purpose of this study is to validate the TX system in a clinical setting using animals and to show its applicability compared to standard clinical methods. METHODS: As test subject, a 15-kg piglet was injected with 53 MBq of 18F-NaF. The µ-map obtained with the TX system and the reconstructed activity distribution were compared to four established AC methods: a Dixon sequence, an ultra-short echo time (UTE) sequence, a CT scan, and a 511 keV transmission scan using a Siemens ECAT EXACT HR+ as the reference. The PET/MRI measurements were performed on a Siemens Biograph mMR to obtain the µ-map using the TX system as well as the Dixon and UTE sequence directly followed by the CT and ECAT measurements. RESULTS: The reconstructed activity distribution using the TX system for AC showed similar results compared to the reference (<5% difference in hot regions) and outperformed the MR-based methods as implemented in the PET/MRI system (<10% difference in hot regions). However, the additional hardware of the TX system adds complexity to the acquisition process. CONCLUSION: Our porcine study demonstrates the feasibility of post-injection transmission scans using the developed TX system in a clinical setting. This makes it a useful tool for PET/MRI in cases where transmission information is needed for AC. Potential applications are studies using larger animals where state-of-the-art atlas-based or artificial intelligence AC methods are not available.