A Method for PET Attenuation Correction Using Background Radiation in PET/MRI

Many modern PET scanners feature lutetium-based scintillators, such as LSO, which emit background radiation due to the existence of the natural occurring radioisotope
¹⁷⁶
Lu. It has been shown that the background radiation from LSO can be used to reconstruct an estimate of patient attenuation maps which can then be used in a joint attenuation and activity reconstruction algorithm such as TOF-MLAA to create enhanced attenuation maps. Following this methodology, the current work explores CT-less reconstruction methods that are particularly suitable for hybrid PET/MRI scanners. Anatomical information from MR images can be used to create regions with unknown linear attenuation coefficients. The unknown attenuation coefficients in each region are first estimated using the attenuation maps from the background radiation and then enhanced in a modified TOF-MLAA algorithm which will be referred to as TOF-MLAA-Reg. In the absence of TOF PET/MR data with background radiation, the developed TOF-MLAA-Reg algorithm was examined using pseudo-MR images derived from CT of a PET/CT scanner and compared against PET images from OSEM using CT-based attenuation and scatter correction and TOF-MLAA. Phantom and patient data were used to evaluate the performance of the developed TOF-MLAA-Reg algorithm incorporating background radiation. Results suggest that TOF-MLAA-Reg significantly minimizes quantification error in PET images when compared to TOF-MLAA. Using a segmented anatomical map of the studied patient, TOF-MLAA showed an average organ SUV error of - 14.80% (range: -27.20% to -6.93%) while the proposed TOF-MLAA-Reg algorithm reduced the error down to only 3.32% (range: -8.10% to 5.18%) across various organs. In conclusion, the proposed method can be a solution to quantitative PET imaging in modern TOF PET/MRI scanners with lutetium-based scintillators.