Abstract
642
Objectives PET Imaging of tau tangles in the brain is highly promising for monitoring the progression of Alzheimer’s disease and chronic traumatic encephalopathies. However, partial volume effects associated with the limited PET spatial resolution pose a challenge to quantitation. We developed an MR-based prior for partial volume correction in PET and applied it to [18F]T807 imaging in 10 subjects.
Methods Our technique is based on 1) deconvolution of PET data with the spatially variant point spread function (PSF) of the scanner measured in the image space and 2) an MR-based information theoretic framework that minimizes the joint entropy between PET and T1MPRAGE MR. The PSF was measured with 0.5 mm diameter 18F point sources placed at different radial and axial locations in the scanner bore. The proposed joint entropy prior obviates the need for MR segmentation. We validate our approach in a Hoffman phantom and evaluate its performance in 10 normal subjects imaged for 20 min with [18F]T807 by quantifying grey to white matter SUV ratios. Correlations of the conventional and corrected PET images with the MMSE scores of the subjects are computed to further evaluate performance.
Results Joint entropy based deconvolution successfully recovered edges and structural details in the PET images. In the Hoffman phantom, the mean grey-to-white SUV ratio increased from 3.0 to 3.3 (ground truth 4.0). There was a prominent reduction in the standard deviation of the SUV by 19% in the grey matter and 37% in the white matter. In [18F]T807 studies, the correlation coefficient of the MMSE scores with the SUVRs for the whole brain, the temporal lobe, the inferior temporal gyrus, the hippocampus, and the parahippocampal gyrus improved by 32%, 21%, 3%, 4%, and 13% respectively.
Conclusions Accurate measurement of the PET PSF and MR-based priors led to significantly enhanced image contrast and reduction of variability. This translated into improved correlation of [18F]T807 SUVRs with neuropsychological test scores, making this approach promising for quantitative imaging of tauopathies.