Effect of MR truncation compensation on quantitative PET image reconstruction for whole-body PET/MR

Truncation compensation (TC) is critical in MR based attenuation correction of quantitative PET image reconstruction in clinical whole-body PET/MR systems. The purpose of this study is to quantitatively evaluate the impact of MR image truncation on PET image reconstruction and the performance of a newly developed TC technique, using Monte-Carlo (MC) simulation. The NCAT phantom was used to create voxelized activities for various organs and hot lesions, and the corresponding attenuation maps (AMs), for a typical FDG patient study. The PURE simulator, which models the Philips Gemini TF PET system, was used to generate realistic list mode data. The full AM with arms down was cropped to create the truncated AM based on the MR field of view. The TC technique was applied on the truncated AM and the corresponding non-attenuation corrected reconstructed image to create a truncation-compensated AM. The list mode data was reconstructed using the above three AMs, and lesion activities were extracted from the reconstructed images. Using the lesion activities from the full AM as the ground truth, errors of the estimated lesion activities from the truncated AM and the truncation-compensated AM were calculated. The study demonstrated that truncation of the MR image based AM considerably affected lesion activity estimation from reconstructed PET images. Our TC technique effectively reduced errors of lesion activity estimation caused by the truncation.