medRxivJClub
Multi-BOUNTI: Multi-lobe Brain vOlUmetry and segmeNtation for feTal and neonatal MRI
medRxiv · · Preprint · Open access
Uus, Fukami-Gartner + 27 more
Abstract ↗AI summary
The abstract is read at the publisher; the summary is JClub's.
Multi-BOUNTI enables automated 43-label multi-lobe segmentation of fetal and neonatal T2w brain MRI, achieving fast (<10 min/case) and accurate agreement with refined labels.
- Why it matters: Perinatal brain volumetry is constrained by inconsistent multi-region segmentation methods that work across fetal and neonatal scans.
- What they did: A 3D Attention U-Net was trained on MRI spanning 21–44 weeks gestational/postmenstrual age using a dedicated 43-label parcellation protocol, then evaluated on independent datasets.
- The result: Using 267 fetal and 593 neonatal Human Connectome Project datasets, the pipeline derived normative 21–44 weeks GA/PMA growth models and produced subject-specific centiles and z-scores for volumetric reporting.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.
- Open access