CRBC News
Politics
Conflict
Security
Economy
Science
Technology
Environment
Health
Culture
Society
Lifestyle
Science

Supercomputer Builds a Highly Detailed 3D Virtual Mouse Brain—A Breakthrough for Neurological Research

07:39 AM, 12/06/2025
Supercomputer Builds a Highly Detailed 3D Virtual Mouse Brain—A Breakthrough for Neurological Research

The Allen Institute and the University of Electro-Communications used Japan’s Fugaku supercomputer to create a detailed 3D simulation of a mouse cortex containing ~9 million neurons and ~26 billion synapses across 86 regions. The dynamic model runs at quadrillions of calculations per second and lets researchers observe how neural activity and diseases like Alzheimer’s or seizures spread through the brain. Early results include insights into brain-wave synchronization and hemisphere interactions, and the team aims to scale toward whole-brain and, ultimately, human-brain simulations.

Supercomputer Builds a Highly Detailed Virtual Mouse Cortex

Researchers at the Allen Institute (US) and the University of Electro-Communications (Japan) have created one of the most comprehensive simulations of a mouse cortex to date. The 3D, dynamic model recreates neural activity in unprecedented detail and is already being used to study brain-wave dynamics, hemisphere interactions, and disease processes such as Alzheimer’s and seizure propagation.

Scale and Performance

The simulation contains roughly 9 million neurons and about 26 billion synapses distributed across 86 interconnected cortical regions. The model runs on Japan’s Fugaku supercomputer and can process activity at a throughput measured in quadrillions of calculations per second, enabled by new software optimizations that reduce unnecessary computations while preserving biological fidelity.

Why This Matters

While a full real mouse brain contains on the order of 70 million neurons, this virtual cortex captures core organizational features that are shared between rodent and human brains. Because the model is three-dimensional and dynamic, researchers can visualize how simulated neurons fire and how activity spreads across regions—allowing experiments on seizure spread, attention-related brain waves, and other network-level phenomena without repeated invasive scans of living animals.

"This shows the door is open. We can run these kinds of brain simulations effectively with enough computing power," said Anton Arkhipov, computational neuroscientist at the Allen Institute. "It’s a technical milestone giving us confidence that much larger models are not only possible, but achievable with precision and scale."

Computer scientist Tadashi Yamazaki of the University of Electro-Communications highlighted Fugaku’s role:

"Fugaku is used for research across many computational science fields. On this occasion, we utilized Fugaku for a neural circuit simulation."

Current Findings and Future Goals

The team reports early discoveries about brain-wave synchronization and interhemispheric coordination in the simulated cortex. Their long-term vision is to scale these methods from single brain areas to whole-brain models and, eventually, biologically detailed human-brain simulations using comprehensive cellular and anatomical data.

The project was presented at the SC25 supercomputing conference and the team has made supporting materials available to the scientific community online.

Similar Articles

Trending