The effect of including bone in Dixon-based attenuation correction for18F-fluciclovine PET/MRI of prostate cancer

Abstract

The objective of this study was to evaluate the effect of including bone in Dixon-based attenuation correction for18F-fluciclovine PET/MRI of primary and recurrent prostate cancer. Methods:18F-fluciclovine PET data from 2 PET/MRI studies—one for staging of high-risk prostate cancer (28 patients) and one for diagnosis of recurrent prostate cancer (81 patients)—were reconstructed with a 4-compartment (reference) and 5-compartment attenuation map. In the latter, continuous linear attenuation coefficients for bone were included by coregistration with an atlas. The SUVmax and mean 50% isocontour SUV (SUViso) of primary, locally recurrent, and metastatic lesions were compared between the 2 reconstruction methods using linear mixed-effects models. In addition, mean SUVs were obtained from bone marrow in the third lumbar vertebra (L3) to investigate the effect of including bone attenuation on lesion–to–bone marrow SUV ratios (SUVRmax and SUVRiso; recurrence study only). The 5-compartment attenuation maps were visually compared with the in-phase Dixon MR images for evaluation of bone registration errors near the lesions. P values of less than 0.05 were considered significant. Results: Sixty-two lesions from 39 patients were evaluated. Bone registration errors were found near 19 (31%) of these lesions. In the remaining 8 primary prostate tumors, 7 locally recurrent lesions, and 28 lymph node metastases without bone registration errors, use of the 5-compartment attenuation map was associated with small but significant increases in SUVmax (2.5%; 95% confidence interval [CI], 2.0%–3.0%; P, 0.001) and SUViso (2.5%; 95% CI, 1.9%–3.0%; P, 0.001), but not SUVRmax (0.2%; 95% CI, -0.5%–0.9%; P 5 0.604) and SUVRiso (0.2%; 95% CI -0.6%–1.0%; P 5 0.581), in comparison to the 4-compartment attenuation map. Conclusion: The investigated method for atlas-based inclusion of bone in18F-fluciclovine PET/MRI attenuation correction has only a small effect on the SUVs of soft-tissue prostate cancer lesions, and no effect on their lesion–to–bone marrow SUVRs when using signal from L3 as a reference. The attenuation maps should always be checked for registration artifacts for lesions in or close to the bones.