Simultaneous partial volume correction and denoising of brain PET images, using transformers and transfer learning
Sanaz Kaviani, Amirhossein Sanaat, Mersede Mokri, Habib Zaidi, Jean-François Carrier · 2026
Read Simultaneous partial volume correction and denoising of brain PET images, using transformers and transfer learning by Sanaz Kaviani, Amirhossein Sanaat, Mersede Mokri, Habib Zaidi, Jean-François Carrier (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13473042.txt.
Description
Sanaz Kaviani and four colleagues presented in 2026 a computer method that cleans up brain PET scans and corrects their measurements at the same time. PET maps activity in the brain, but it loses quality when dose or scan time is cut. Small structures then blur into neighbours, an effect called partial volume error. Older correction methods need a matching MRI scan and fail when counts are low.
Their model, built on a transformer design called Swin UNETR, was trained on phantom models and real patient scans with two tracers. It was then tested on scans kept at only 10 percent of full counts. It beat uncorrected low-count images on every measure, raising image similarity by about 2 percent and signal quality by about 11 percent overall. Because it needs only the PET scan itself at run time, it could spare patients an extra MRI and still give numbers clinicians can trust.
Catalog record
| CREATOR | Sanaz Kaviani, Amirhossein Sanaat, Mersede Mokri, Habib Zaidi, Jean-François Carrier |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 58 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |