MRI-based attenuation correction of PET images in clinical PET/MR

Based on our previous approach we seek to develop a robust algorithm for AC in PET/MR brain images and extend this atlas-based approach to whole body.
Given the access to a prototype PET/MR system at the University Hospital Tübingen we intend to validate and fine-tune our MR-based attenuation correction with patient data.
Attenuation correction (AC) is mandatory for quantitative Positron Emission Tomography (PET) imaging. Without attenuation correction activity concentrations on PET images are biased depending on the tissue density.
The lack of conventional X-ray transmission sources in PET/MR systems requires alternative approaches for attenuation correction as a pre-requisite for quantitative PET in combined PET/MR. MR image values do not correlate with attenuation coefficients of the tissue, and therefore can not be directly used for PET-AC.
We apply an approach based on a database of coregistered PET/CT and MRI datasets with pattern recognition methods to generate attenuation maps from MR data.
For a patient subject to MR-AC, atlas MR datasets are co-registered to the MR image volume of the PET/MR patient. The resulting transformations are applied to the corresponding CT datasets from the database. Subsequently, a pattern recognition approach is used to match the MR image of the patient with the appropriate CT information from that MR-CT datasets that best match the patient information. This voxel-based approach can merge partial sub-volumes from independent data sets into a single CT-volume that is used for MR-AC of the patients. The generated "Pseudo CT" is then converted to an attenuation map using a piecewise linear transformation.