Improved brain PET quantification using partial volume correction techniques.
Doctoral thesis, UCL (University College London).
Positron emission tomography (PET) suffers from a degradation in quantitative accuracy due to a phenomenon known as the partial volume effect (PVE). The effects are due to the limited spatial resolution of the scanner. Methods that correct for PVEs are known as partial volume correction (PVC) techniques and are either data-driven or make use of anatomical information from other modalities such as magnetic resonance (MR) imaging. This thesis reports investigations into PVC techniques for improving the quantification of brain amyloid PET tracers. These tracers image amyloid plaque aggregation in-vivo, which is a pathological hallmark of Alzheimer’s disease. An extension to existing anatomy-based PVC methods is reported. Region-based voxelwise (RBV) correction has been shown to reduce PVE-induced regional bias and variance when compared to commonly applied PVC techniques. This has been proven in phantom studies and observed in clinical data. In addition, RBV has been used to demonstrate that white matter variability exists in two different amyloid tracers. This finding has implications for the application of PVC in amyloid imaging and also how scans should be normalised. Alternative reference regions were investigated in two amyloid PET tracers. The brain stem, in combination with PVC, was found to result in the strongest agreement between tracers. Anatomy-based PVC techniques rely on parcellations of structural images. These parcellations are not necessarily representative of the PET data. A further extension to RBV is proposed which iteratively modifies the parcellations to find an optimal PVC in terms of the observed PET data. This novel technique reduces quantification errors due to PET-MR mismatch and has the potential to provide an additional parameter of ‘functional volume change’ in longitudinal studies.
|Title:||Improved brain PET quantification using partial volume correction techniques|
|Open access status:||An open access version is available from UCL Discovery|
|UCL classification:||UCL > School of Life and Medical Sciences > Faculty of Medical Sciences > Medicine (Division of)|
Archive Staff Only