Can Deep Learning Detect Brain PET Anomalies from Healthy Variability?

Original title: Leveraging healthy population variability in deep learning unsupervised anomaly detection in brain FDG PET

Authors: Maëlys Solal (ARAMIS), Ravi Hassanaly (ARAMIS), Ninon Burgos (ARAMIS)

The article explores detecting brain anomalies in neuroimaging without labels. Current methods use a subject-specific model of a “healthy” brain, comparing individual images to this ideal. Yet, this approach faces challenges due to imperfect reconstructions and unclear thresholds.

Original article: https://arxiv.org/abs/2311.12081