ORIGINAL RESEARCH article
Front. Neuroinform.
Volume 19 - 2025 | doi: 10.3389/fninf.2025.1633273
This article is part of the Research TopicMultimodal Brain Data Integration and Computational ModelingView all 3 articles
Generation of synthetic TSPO PET maps from structural MRI images
Provisionally accepted- 1University of Rome Tor Vergata, Roma, Italy
- 2Imperial College London, London, United Kingdom
- 3London South Bank University, London, United Kingdom
- 4Massachusetts General Hospital Athinoula A Martinos Center for Biomedical Imaging, Charlestown, United States
- 5Harvard Medical School, Boston, United States
Select one of your emails
You have multiple emails registered with Frontiers:
Notify me on publication
Please enter your email address:
If you already have an account, please login
You don't have a Frontiers account ? You can register here
Introduction: Neuroinflammation, a pathophysiological process involved in numerous disorders, is typically imaged using [ 11 C]PBR28 (or TSPO) PET. However, this technique is limited by high costs and ionizing radiation, restricting its widespread clinical use. MRI, a more accessible alternative, is commonly used for structural or functional imaging, but when used using traditional approaches has limited sensitivity to specific molecular processes. This study aims to develop a deep learning model to generate TSPO PET images from structural MRI data collected in human subjects.Methods: A total of 204 scans, from participants with knee osteoarthritis (n = 15 scanned once, 15 scanned twice, 14 scanned three times), back pain (n = 40 scanned twice, 3 scanned three times), and healthy controls (n=28, scanned once), underwent simultaneous 3T MRI and [ 11 C]PBR28 TSPO PET scans. A 3D U-Net model was trained on 80% of these PET-MRI pairs and validated using 5-fold cross-validation. The model's accuracy in reconstructed PET from MRI only was assessed using various intensity and noise metrics.Results: The model achieved a low voxel-wise mean squared error (0.0033 ± 0.0010) across all folds and a median contrast-to-noise ratio of 0.0640 ± 0.2500 when comparing true to reconstructed PET images. The synthesized PET images accurately replicated the spatial patterns observed in the original PET data. Additionally, the reconstruction accuracy was maintained even after spatial normalization. Discussion: This study demonstrates that deep learning can accurately synthesize TSPO PET images from conventional, T1-weighted MRI. This approach could enable low-cost, noninvasive neuroinflammation imaging, expanding the clinical applicability of this imaging method.
Keywords: Synthetic PET, U-net, TSPO, Neuroinflammation, Separable convolutions
Received: 22 May 2025; Accepted: 07 Aug 2025.
Copyright: © 2025 Ferrante, Inglese, Brusaferri, Toschi and Loggia. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
* Correspondence: Marianna Inglese, University of Rome Tor Vergata, Roma, Italy
Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.