Abstract
73
Introduction: Aggregated tau proteins are one of the hallmarks of Alzheimer’s disease (AD), and there is a great need for further exploration of how tau accumulates and spreads in early stages of disease. [F-18]MK-6240 is a next generation PET ligand for imaging tau aggregates in AD. Using the PET Simulation Of Realistic Tridimensional Emitting Objects (SORTEO) software package (Reilhac, 2004) we developed a simulation tool to investigate the detection sensitivity of [F-18]MK-6240 used to detect tau aggregates.
Methods: Simulations were performed with the PET-SORTEO package, which uses Monte-Carlo methods to generate realistic PET data with noise given regional time activity curves (TAC) as an input. We identified 12 brain regions seen in the progression of AD and created a digital brain phantom based on the USC Laboratory of Neuroimaging Probabilistic Brain Atlas 40 for use in our studies. Each region had an assigned TAC value for a 20-minute (4x5min frames) scan 70 minutes after injection. Spherical lesions of radius 6mm were also placed bilaterally in the entorhinal cortex region where tau accumulates in early stage AD and full TACs were based on 19 participants with 90-minute acquisitions of [F-18]MK-6240 at the Wisconsin Alzheimer’s Disease Research Center. Using inferior cerebellar grey matter as our reference region with average PET tracer concentration 12% higher than cortical grey matter, we designated 3 activity levels in the entorhinal lesions for separate trials: 2.5x (high), 1.75x (medium) and 1x (low) reference concentrations and completed 20 realizations at each concentration level. These values were chosen to explore the range of tau accumulation seen in our population. The SORTEO generated data was based on the Siemens CTI ECAT HR+ PET scanner (FWHM = 4.1mm, in plane resolution) and reconstructions of our simulated images were completed using our research protocols (FBP, all pass filter, 2.059x2.059x2.425mm3 voxel size) and then summed across the 4 PET time frames. The concentration of the lesions of the reconstructed images were obtained by coregistering the reconstructed image to our digital brain phantom, finding the average signal in the assigned regions and then calculating the SUVR across all cortical grey matter and the lesions respectively compared to the cerebellum.
Results: Using multithreading across 20 different cores each realization took approximately 90 minutes to run. From our simulations of high uptake we found an average SUVR of 1.588±0.178 in the entorhinal cortex compared to an average SUVR of 0.881±0.006 in the cortical grey matter, a difference of nearly 4 standard deviations. For the medium uptake scans, however, the average SUVR had values of 1.283±0.219, only 1.84 standard deviations larger than the average SUVR of the cortical grey matter of 0.880±0.007. The low uptake simulations had an average SUVR of 0.876±0.268, lower than that of the cortical grey matter average of 0.884±0.009, meaning at this concentration the lesions are indistinguishable from cortical grey matter using an ROI volume defined by the digital mask, although these values did have a very high variance.
Conclusions: Our results demonstrate the feasibility of using PET-SORTEO simulations to inform the detection limits of tau PET in AD. Further development of this model is ongoing to incorporate the postmortem concentration and spatial distribution of tau lesions with the measured protein-ligand affinity of [F-18]MK-6240 to provide insight for determining the threshold of in vivo PET detectability for tau pathology.
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.