Skip to content

Brain Age Prediction project

About the project

Full title: MRI based brain age prediction in patients with CNS hypersomnolence disorders and healthy controls

Analysis leads: Hilde T. Juvodden, Dag Alnæs, Esten H. Leonardsen, Lars T. Westlye, and Stine Knudsen-Heier

This project represents a NICHY subproject that applies advanced machine learning approaches to neuroimaging data. This study uses pre-trained deep learning models applied to minimally processed T1-weighted MRI scans to assess brain age and brain age gap (BAG) as a holistic marker of brain health. Specifically, we aim to:

  • Apply pre-trained MRI-based brain age models on NICHY datasets and compare brain age and corresponding BAG between narcolepsy type 1, narcolepsy type 2, idiopathic hypersomnia, Kleine-Levin Syndrome, and healthy controls.
  • Characterise the associations between brain age/BAG and individual-level clinical and sleep data (disease onset, duration, medication, H1N1-vaccination status, HLA-DQB1*06:02 status, CSF hypocretin level, disease severity, and sleep characteristics) across all CNS hypersomnolence disorder groups.
  • Evaluate whether global imaging measures derived from brain age prediction may increase sensitivity to case-control differences and clinical characteristics compared to conventional region-of-interest approaches.

This work leverages previously developed convolutional neural network models trained on large and diverse datasets to provide a more holistic perspective on brain health in CNS hypersomnolence disorders, with potential to reveal subtle brain changes and inform understanding of neurobiological mechanisms underlying these severe neurological conditions.