TOF PET/MR attenuation correction of flexible MR-hardware utilizing MLAA and template alignment

Abstract

Accurate attenuation correction (AC) of hardware components used during Magnetic Resonance Imaging (MR) remains a challenge in PET/MR, as these components are not visible in MR-derived attenuation maps. CT-based template μ-maps are widely used for stationary hardware such as the patient table, but flexible MR hardware is currently not corrected for in clinical protocols. Previously, the influence of such MR hardware (MR-HW) on PET quantification was investigated using phantom [1] and patient [2] measurements performed on the Siemens Biograph mMR. These initial studies explored the simultaneous estimation of attenuation and activity (MLAA) [3] together with registration of a CT-based template μ-map on non-time of flight data to correct for the coil attenuation, similar to [4]. It has been shown that time-of-flight (TOF) information can stabilize the joint estimation problem for patient attenuation correction, eliminating crosstalk artefacts in the presence of errors in initial attenuation estimates [5].

Building on these findings, the current work evaluates AC strategies for MR hardware using a state-of-the-art TOF PET/CT system (Biograph Vision 600, Siemens Healthineers) to explore how TOF information influences the robustness and accuracy of MRAC for recent PET/MR systems. Measurements of a NEMA body phantom—with and without an MR-hardware coil placed on top—were acquired using the TOF PET/CT system. MLAA, initialized with a phantom (and patient table) μ-map only, was employed to reconstruct the attenuation distribution outside the phantom, i.e., the coil attenuation. These results were compared to the known coil position obtained from a dedicated CT acquisition of the setup. Additionally, a CT scan of the coil in a different position was converted into a PET μ-map and registered to the MLAA output to assess registration-based correction. The feasibility and accuracy of the registration and the impact of the different MR-hardware attenuation correction strategies on the reconstructed PET images were evaluated.

Please click on the 'PDF' for the full abstract!