Abstract
2320
Objectives The prototype 3TMR-BrainPET combines high resolution PET and anatomical MRI. This study explores the possibility of using an image-derived input function (IDIF) from VOIs placed over the carotid artery in dynamic FDG-studies acquired with the BrainPET insert.
Methods In 3 patients without brain disease BrainPET data were acquired post injection (p.i.) of 370 MBq FDG in listmode for 1 h and reconstructed with OP-OSEM including all corrections. There were 23 frames from 8 frames per 5 sec to 3 frames per 10 min. Six venous blood samples were drawn from 5 to 55 min. MPRAGE-MRI was recorded shortly p.i.. A 50% isocontour VOI (MR50VOI) was defined in the MR image over the carotid arteries and copied to the PET image to obtain IDIF-A from the VOI average and IDIF-4P from the VOI mean values of the 4 hottest pixels per plane. To correct for the partial volume effect (PVE) and spillover (SP) from FDG within the brain the late part of each IDIF was calibrated to the blood samples. Additionally an IDIF-4V was created from the mean values of the 4 hottest pixels in MR50VOI. Finally, a 0/1-mask was created from the MR50VOI and filtered with a 3D 3mm Gaussian corresponding to the BrainPET’s central resolution. The mean value of the 4 hottest pixels in the filtered image represents a theoretical PVEt for IDIF-4V.
Results Calibrating the IDIFs to the blood samples yielded the results shown in the table. Comparing the blood peaks found in the calibrated IDIF-A and in the IDIF-4V yielded a ratio of 0.93+/-0.26 which is similar to 0.81 +/-0.03, the value of PVEt for the 4 hottest pixels within MR50VOI.
Conclusions Our results show that MRI simultaneously acquired with PET in an MR-BrainPET scanner is a promising tool for determining an image-derived input function