Researchers have used an AI model to generate a remarkably detailed map of the mouse brain, identifying around 1,300 distinct regions and subregions.
The effort, led by scientists at Seattle’s Allen Institute and UCSF, relies on a “transformer” neural network trained to analyze spatial and molecular data. “I think there are already indications that we can go beyond what we see now,” said Bosiljka Tasic, director of molecular genetics at the Allen Institute for Brain Science. The automated process provides a scalable alternative to manual annotation and may soon enable similar mapping of human brain tissue.
The work strengthens Seattle’s role in AI-driven neuroscience, positioning the region to lead in computational neurobiology.