Data-Driven Motion Correction Refines Synaptic PET Imaging Accuracy in Alzheimer’s Disease

A clinical neuroimaging study has demonstrated that applying high-temporal-resolution data-driven motion correction during reconstruction significantly improves image contrast and quantitative accuracy in positron emission tomography scans targeting synaptic density. The investigation, published in EJNMMI Research by lead author Charles Carron and colleagues, analyzed how participant movement and brain lesions influence quantitative measurements in individuals with Alzheimer’s disease and cognitively unimpaired volunteers.

The study focused on the radiotracer 18F-SynVesT-1, which targets synaptic vesicle protein 2A as a surrogate biomarker for synaptic density throughout the brain. While both regions have previously been validated, head movement and structural changes in white matter can distort tracer measurements across target and reference tissues.

To evaluate these confounding factors, the researchers reconstructed 18F-SynVesT-1 PET images obtained between 60 and 90 minutes post-injection from a cohort of 40 healthy volunteers and 31 patients diagnosed with Alzheimer’s disease. The team evaluated three distinct motion-handling approaches: uncorrected reconstructions, a conventional five-minute frame-based post-reconstruction coregistration and averaging method, and an advanced high-temporal-resolution data-driven motion correction protocol applied directly during the image reconstruction process.

To account for structural tissue abnormalities, white matter hyperintensities were identified and segmented using T2-weighted fluid-attenuated inversion recovery magnetic resonance imaging. Investigators systematically excluded these hyperintensities from the centrum semiovale volumes of interest to generate lesion-corrected reference regions.

The data-driven reconstruction approach improved visual sharpness and effectively reduced artifacts caused by head movements and attenuation mismatches. Statistically, this correction increased standardized uptake values in high-binding cortical regions by a median of 0.3 to 1.0 percent (p ≤ 0.001) while decreasing values in the low-binding centrum semiovale by a median of 0.5 percent (p = 0.001). In contrast, post-reconstruction frame coregistration and averaging improved image contrast only in acquisitions characterized by severe participant motion.

The cerebellum exhibited notable stability, emerging as one of the least sensitive regions to motion correction, with a median uptake increase of 0.4 percent (p < 0.001) under data-driven correction. Furthermore, white matter hyperintensities influenced centrum semiovale uptake only when lesions occupied at least 25 percent of the regional volume, whereas cerebellar uptake remained unaffected regardless of lesion burden. Although individual patient z-scores showed substantial divergence between reference regions—such as a mean absolute difference of 1.00 ± 0.82 in the precuneus—ratios normalized to the centrum semiovale yielded slightly superior differentiation between Alzheimer’s patients and controls once motion and lesions were controlled.

The researchers concluded that the whole cerebellum provides a more robust and resilient reference region in clinical datasets marked by significant participant movement or substantial white matter pathology. Conversely, the centrum semiovale may offer superior group differentiation in carefully controlled datasets where head motion and white matter hyperintensities are rigorously corrected. The clinical trial associated with this research was registered on ClinicalTrials.gov under the identifier NCT05384353.

Subject of Research: Quantitative reference region selection and motion correction in synaptic PET imaging for Alzheimer's disease

Article Title: Effects of motion and white matter hyperintensities on reference region based ¹⁸F-SynVesT-1 PET quantification in Alzheimer’s disease and healthy volunteers

Article References: Carron, C., Noë, S., Vanderlinden, G., Koole, M., Schramm, G., & Van Laere, K. (2026). Effects of motion and white matter hyperintensities on reference region based ¹⁸F-SynVesT-1 PET quantification in Alzheimer’s disease and healthy volunteers. EJNMMI Research, 16(1), Article 153. https://doi.org/10.1186/s13550-026-01496-8

Image Credits: AI Generated

DOI: 10.1186/s13550-026-01496-8

Keywords: 18F-SynVesT-1, positron emission tomography, synaptic vesicle protein 2A, synaptic density, Alzheimer's disease, centrum semiovale, cerebellum, pseudo-reference regions, data-driven motion correction, white matter hyperintensities, standardized uptake value, standardized uptake value ratio, partial volume correction, T2-FLAIR imaging, attenuation correction, precuneus, clinical trial NCT05384353