Abstract
241996
Introduction: Positron emission tomography (PET) is used to measure accumulation of amyloid beta (Aβ) plaques in the brain, as an indicator of Alzheimer’s disease1. Aβ burden is summarised by the standardized uptake value ratio (SUVR),2 which depends on the radiotracer, scanner hardware and reconstruction algorithm. This is partially owing to changes in spatial resolution altering the partial volume effect (PVE). Furthermore, we have established that spatial resolution differentially impacts Aβ- and Aβ+ subjects.3,4 These dependencies, in conjunction with significant performance variation across scanners, represent a problem for multi-site Aβ-PET studies; the same patient imaged on different PET scanners can exhibit Aβ variability.
We hypothesised that multi-scanner Aβ harmonization can be achieved by changing reconstruction parameters to match spatial resolution across scanners. We interrogated this hypothesis using a three-camera comparison dataset of [18F]-NAV4694 radiotracer, where each participant was scanned on two scanners. Raw PET data and vendor-supplied reconstruction toolboxes enabled comparison of a broad range of reconstruction settings and spatial resolutions.
Methods: Three scanners were used: Siemens Biograph Vision 600; Siemens Biograph mCT; Philips Gemini TF64. Spatial resolution (FWHM) of each reconstruction configuration was estimated using a barrel phantom.4 Three pairwise comparisons were done: i) Vision-mCT; ii) mCT-Philips; iii) Vision-Philips. The same subjects were scanned for Aβ twice on a selected pair of scanners. All subjects were injected with [18F]-NAV4694 radiotracer 50 minutes prior to 20 minutes of continuous scanning.
SUVR values were computed across a total of 72 brain regions.5 Reconstructed images were compared using the set of regional SUVRs, with statistical similarity assessed using within subject paired t-tests. Root mean-squared-differences (RMSD) between paired global SUVR values, was also calculated.
Vision: reconstructed with ordered subset expectation maximization (OSEM) and time-of-flight (TOF). Twelve iterations,i ∊ [1,42], subsets, s=5, Gaussian smoothing from 2-8 mm. Philips: reconstructed with a single configuration, the line-of-response row-action maximum likelihood (LOR RAMLA). mCT: reconstructed using OSEM with TOF, i ∊ [2,12], s=21, Gaussian smoothing 1-13 mm.
Results: Vision-mCT harmonization
Vision reconstruction with spatial resolution of 4.1mm gave maximal statistical similarity (p=0.26) between regional SUVRs (Figure 1A), and minimised RMSD (Figure 1B). Paired global SUVRs calculated using the optimally harmonized reconstruction pair were closely matched (Figure 1C).
Representative brain slices from a subject scanned on mCT and Vision are shown (Figure 1D). Statistically matched mCT (Figure 1D, FWHM=4.3mm, 1st row) and Vision (FWHM=4.1mm, 2nd row) reconstruction configurations are less visually similar than the Vision reconstruction configuration with FWHM=3.5 mm (3rd row).
mCT-Philips harmonization
The mCT reconstruction configuration with FWHM=9.35mm resulted in maximal similarity (p=0.91) (Figure 2A). Global SUVRs for subject pairs of harmonized reconstruction configurations were well matched (Figure 2C).
Vision-Philips harmonization
The Vision reconstruction configuration with FWHM=9.1mm resulted in maximal similarity (p=0.94) (Figure 2D). Global SUVRs for subject pairs of harmonized reconstruction configurations were well matched (Figure2F).
Conclusions: Using Aβ-PET data from three sets of participants each scanned on a pair of PET scanners, we conclude that the process of matching spatial resolution measured by a barrel phantom can be used to minimize the Aβ-PET quantitation differences between the scanners. Visual matching of Aβ-PET images across scanners does not robustly harmonize Aβ-PET quantitation. Our promising technique can support multi-centre trials and enable PET camera updates during longitudinal studies.
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.