The paper describes the development and application of NextBrain, a new probabilistic histological atlas of the human brain designed to enhance Magnetic Resonance Imaging (MRI) segmentation. Created using custom artificial intelligence-enabled methods to align approximately 10,000 histological sections from five human brain hemispheres, NextBrain offers 333 regions of interest (ROIs) with probabilistic labels across the whole brain. This atlas is paired with a companion Bayesian segmentation tool for automatically segmenting these ROIs in MRI scans at a finer granularity than previous atlases. The authors demonstrate the utility of NextBrain through applications like segmenting ultra-high-resolution ex vivo MRI and performing fine-grained volumetric analysis for Alzheimer’s disease and aging studies, showing significant improvements over existing methods. The raw data, the atlas, and the tools are all publicly released to enable widespread research use.
References:
- Casamitjana A, Mancini M, Robinson E, et al. A probabilistic histological atlas of the human brain for MRI segmentation[J]. Nature, 2025: 1-8.
