Benchmarking 3D generative autoencoders for pseudo-healthy reconstruction of brain 18F-fluorodeoxyglucose positron emission tomography
Even if no model clearly outperformed all the others, the benchmark allowed identifying a few models that perform slightly better than the vanilla VAE. It further showed that many VAE-based models can generalize to the detection of anomalies of various intensities, shapes, and locations in 3D brain FDG PET.
Authors
- Olivier Colliot (ORCID: https://orcid.org/0000-0002-9836-654X)
- Maëlys Solal (ORCID: https://orcid.org/0000-0001-9333-9238)
- Ninon Burgos (ORCID: https://orcid.org/0000-0002-4668-2006)
- Ravi Hassanaly (ORCID: https://orcid.org/0009-0009-1304-5906)
- for the Alzheimer’s Disease Neuroimaging Initiative
Institutions
- Centre National de la Recherche Scientifique (FR)
- Inserm (FR)
- Sorbonne Université (FR)
- Assistance Publique – Hôpitaux de Paris (FR)
- Pitié-Salpêtrière Hospital (FR)
- Institut du Cerveau (FR)
Publication Details
- Journal
- Journal of Medical Imaging
- Published
- 2025-10-16
- DOI
- https://doi.org/10.1117/1.jmi.12.5.054005
- Primary Topic
- Medical Imaging Techniques and Applications
- Type
- article
- Field-Weighted Citation Impact
- 0.00