Comparison of attenuation collection methods in brain PET/MRI study

Objectives: The accuracy of attenuation correction (AC) of PET/MRI still remains a major problem. A segmentation-based method using the two-point Dixon gradient echo sequence is currently used to attain MR-based PET AC of PET/MRI of the whole body. This method provides a segmentation-based attenuation map showing air, lungs, soft tissues, and fat, however, bone is considered as soft tissue in this method. Another AC method using the ultrashort echo time (UTE) sequence can detect bone signal of skull and incorporate into attenuation maps, and therefore must result in closed standard uptake value (SUV) of brain tissue to those with AC using X-ray CT or ⁶⁸Ge-⁶⁸Ga line source. In this study, the effects of AC by UTE sequence (UTE-AC) were evaluated by comparing with AC by Dixon sequence (Dixon-AC) in the brain PET/MRI studies with ¹⁸F-fluorodeoxyglucose (FDG)

Methods: Twenty three patients examined by PET/MRI ranging from 9 to 71 years (19 males and 4 females, mean age ± SD: 55±19.8 years) were included. All patients were assessed for malignant brain lesions including five patients suspected brain tumor. Used PET/MRI camera was Siemens Biograph mMR. A static PET scan for 5 min were performed 45 min after intravenous infusion of FDG. Both attenuation maps with UTE and Dixon sequence were acquired by MRI simultaneously with PET scanning. Both T1 and T2 weighted images were also acquired. Regions-of-interest (ROI) were located on the cerebellum, pons, thalamus, and parietal, frontal, temporal and occipital cortices referring T1 and T2 weighted images to calculate mean and maximum of standard uptake value (SUV) of FDG in ROI. In five patients, SUVs of eight brain lesions were also measured.

Results: Both average of mean SUV with Dixon-AC and UTE-AC were 2.8-12.9 and 3.5-16.0 respectively. Both maximum SUV with Dixon-AC and UTE-AC were 3.4-15.9 and 4.0-17.6. The mean SUV with UTE-AC were 17.4%, 13.0%, 17.9%, 24.6%, 23.0%, 14.7% and 17.0% higher than those with Dixon-AC for the cerebellum, pons, thalamus, and parietal, frontal, temporal and occipital cortices, respectively. The maximum SUV with UTE-AC were 19.1%, 15.5%, 17.5%, 24.7%, 24.1%, 16.6% and 19.2% higher than those with Dixon-AC for the cerebellum, pons, thalamus, and parietal, frontal, temporal and occipital cortices, respectively. Both mean and maximum SUVs with UTE-AC were significantly higher than those with Dixon-AC. Both average of mean SUV with Dixon-AC and UTE-AC of brain tumors were 9.4 ± 2.5 SD and 12.0 ± 3.3 SD respectively. Both maximum SUV with Dixon-AC and UTE-AC were 11.3± 3.0 SD and 14.1 ± 3.9 SD. The mean and maximum SUV with UTE-AC of brain lesion were 15.0% and 16.6% higher than those with Dixon-AC respectively.

Conclusion: By using the UTE-AC, calculated SUV in all brain regions were higher than those with Dixon-AC. SUVs with UTE-AC can probably be considered to be closed to those with AC using X-ray CT or ⁶⁸Ge-⁶⁸Ga line source as compared to those with Dixon-AC. Thus, we should be aware of underestimation of SUV in brain PET/MRI studies using Dixon-AC. Research Support: none