Image-derived input function obtained in a 3TMR-brainPET

Abstract

Aim: The combination of a high-resolution MR-compatible BrainPET insert operated within a 3T MAGNETOM Trio MR scanner is an excellent tool for obtaining an image derived input function (IDIF), due to simultaneous imaging. In this work, we explore the possibility of obtaining an IDIF from volumes of interest (VOI) defined over the carotid arteries (CAs) using the MR data. Material and methods: FDG data from three patients without brain disorders were included. VOIs were drawn bilaterally over the CAs on a MPRAGE image using a 50% isocontour (MR50VOI). CA PET/MR co-registration was examined based on an individual and combined CA co-registration. After that, to estimate the IDIF, the MR50VOI average (IDIF-A), four hottest pixels per plane (IDIF-4H) and four hottest pixels in VOI (IDIF-4V) were considered. A model-based correction for residual partial volume effects involving venous blood samples was applied, from which partial volume (PV) and spillover (SP) coefficients were estimated. Additionally, a theoretical PV coefficient (PVt) was calculated based on MR50VOI. Results: The results show an excellent co-registration between the MR and PET, with an area under the curve ratio between both co-registration methods of 1.00±0.04. A good agreement between PV and PVt was found for IDIF-A, with PV of 0.39±0.06 and PVt0.40±0.03, and for IDIF-4H, with PV of 0.47±0.05 and PVt0.47±0.03. The SPs were 0.20±0.03 and 0.21±0.03 for IDIF-A and IDIF-4H, respectively. Conclusion: The integration of a high resolution BrainPET in an MR scanner allows to obtain an IDIF from an MR-based VOI. This must be corrected for a residual partial volume effect.