Methodology and phantom studies for the development and assessment of motion correction in simultaneous PET-MR

摘要

Simultaneous PET-MR imaging has recently emerged into routine clinical use, with significant role in diagnosis and treatment. The high sensitivity of a PET scanner, combined with the anatomical and functional information given by MR can focus to a molecular level and provide simultaneous information about individual but correlated parameters. The combination of these modalities has also brought new opportunities. One of these is the potential for the MRI to capture and measure patient motion during PET scanning. This motion information, in the form of motion fields can be applied back to the PET and correct any motion artefacts created. For the success of this study many methodological problems have to be resolved. This thesis investigates the methodology for motion correction of PET data by appropriate use of MR motion information, which could in due course become a standard motion correction method for both pre-clinical and human studies. In particular, this study utilized prototype pre-clinical MR-compatible PET inserts. Following the phantom development and during the experimental procedures, various prototype simultaneous MR-compatible PET small-diameter scanners, in conjunction with a 3T clinical MR scanner, were used to validate the phantom and evaluate several different approaches to motion correction. All scanner combinations used were a combination of a removable PET insert operating within a clinical MR scanner so it was necessary to develop robust methods to ensure spatial alignment and temporal synchronization. The data obtained from simultaneous PET and MR acquisition was arranged into gates and reconstructed by deriving motion transformations from the MR data and using this to correct the PET data for the effects of motion using several PET reconstruction approaches. The experimental results obtained with various PET-MR systems demonstrate the feasibility of this approach and the degree of motion correction that can realistically be expected in pre-clinical PETMR investigations.