PET scans can reveal inflammation in the brain, but they’re costly and require injecting a small amount of radioactive material. In a new study, Marco Loggia and colleagues developed an artificial intelligence system that can create “synthetic” [11C]PBR28 (or TSPO) PET images directly from regular MRI scans—no radiation required. The AI learned to predict patterns of brain inflammation that normally show up only in PET scans, producing results that closely matched real images. This technology could make it easier and safer to monitor diseases like multiple sclerosis or Alzheimer’s using equipment already common in hospitals.
Read the Paper
Ferrante M, Inglese M, Brusaferri L, Toschi N, Loggia ML. Generation of synthetic TSPO PET maps from structural MRI images. Front Neuroinform. 2025 Sep 8;19:1633273. doi: 10.3389/fninf.2025.1633273. PMID: 40988654; PMCID: PMC12450709. (PubMed page)