Clinical comparison of a whole-body PET-MR system with PET-CT

Objectives PET-MR is a potential alternative to PET-CT. We present an evaluation in clinical patients(pts) of the first combined whole-body PET-MR unit in the US, in comparison with PET-CT.

Methods After injection of F-18 FDG,36 pts had routine PET-CT at 1.2±0.4 hrs,followed by PET-MR of whole body and/or body parts at 2.7±0.9 hrs,with a Philips TF PET-MR scanner.T1 MR images for PET coregistration were segmented into air,lungs,or soft tissues for attenuation correction (atMR).The PET-CT,PET-MR,and atMR images were reviewed for clinical findings,quality,and artifacts.

Results 46 of the 52 PET-MR studies(92%) were of good quality.All significant lesions on PET-CT were also seen on PET-MR. 27 of 40(68%) head images had bitemporal MR signal droput(SDO) due to headphones, and nearly all body MR images had SDO at field of view edges, with no effect on PET. 16 of 40(20%) head-neck and 12 of 37(32%) body MR images had SDO due to metallic materials.Of these,only 5 PET images were affected. 17 of 35(48%) body MR images had aortic blood motion artifacts in the lungs or liver,which caused mild PET lesions in 2 studies. 15 of 35 (43%) body MR images had motion artifacts without effect on PET. 7 of 52(13%) atMR images showed "air" artifact in the skull base,chest,or thigh,with no effect on PET.

Conclusions The PET images from PET-MR were generally of good quality with identification of all lesions seen on PET-CT.The MR images were adequate for anatomical correlation for the brain, pelvis,lower extremities,and the neck,except for metallic artifacts.The MR images were of poor quality in the chest and abdomen,due to organ or blood motion artifacts.The atMR was adversely affected by occasional mistaken segmentation. These MR artifacts did not generally affect the PET images. As the challenges for atMR continue to be addressed, anatomical correlation will need to be met with additional regionally-optimized MR imaging.

Research Support Philips Healthcar