MLAA-based headphone attenuation estimation in hybrid PET/MR imaging

To enable the use of MR-safe pneumatic headphones for quantitative PET imaging of the brain, we propose an MLAA-based algorithm compensating for headphone attenuation. The proposed approach simultaneously reconstructs attenuation and activity distributions from the PET emission data. Compared to the original MLAA algorithm, the attenuation update is only applied outside the patient body outline where the headphones are assumed to be located. Prior information on expected attenuation coefficients is incorporated into the attenuation update. Neglecting headphone attenuation underestimates the activity distribution, especially in regions embraced by the headphones. Dedicated phantom experiments demonstrate that the proposed MLAA-based approach can significantly improve PET quantification compared to the case where headphone attenuation is neglected. The feasibility of the proposed method for patient data acquired with a clinical PET/MR device is also demonstrated.