In a major leap for neuroscience, researchers have created the first functional connectome of the mouse visual cortex at cubic millimeter scale, linking neural activity to synaptic wiring across an entire brain region.
Study: Functional connectomics spanning multiple areas of mouse visual cortex. Image Credit: r.classen/Shutterstock.com
Detailed in a recent Nature article, the Machine Intelligence from Cortical Networks (MICrONS) dataset combines dense calcium imaging of approximately 75,000 neurons with electron microscopy (EM) reconstructions of more than 200,000 cells and over half a billion synapses. The dataset includes full dendritic and axonal reconstructions, along with tools for analyzing cell-type-specific connectivity, synaptic architecture, and functional responses to visual stimuli.
As an open-access resource, MICrONS offers a detailed view into how the brain processes visual information, highlighting principles of neuronal invariance and the complex relationship between structure and function in cortical circuits.
Background
Understanding how the brain works means mapping how patterns of neural activity emerge from synaptic connections—a challenge famously laid out by Francis Crick in 1979. While earlier techniques like electrophysiology, calcium imaging, and viral tracing helped illuminate pieces of this puzzle, they offered only partial views of the circuitry involved.
Recent efforts to combine EM with functional imaging have succeeded in smaller systems, such as retinal tissue or individual cortical columns, but extending these methods to larger brain volumes has remained a key barrier.
The MICrONS project addresses that limitation by integrating dense calcium imaging with full EM reconstruction within a shared volume of mouse visual cortex. This massive, multimodal dataset enables researchers to directly link neural activity with detailed morphological and connectivity data. Beyond simply making the data public, the project includes tools for analyzing structure-function relationships at scale, offering a platform for testing long-standing hypotheses in visual neuroscience.
Mapping Mouse Visual Cortex at Scale
The dataset pairs in vivo two-photon calcium imaging of excitatory neurons across four mouse visual areas with ex vivo EM reconstruction of the same 1.2 mm3 tissue volume—capturing over 200,000 cells and 500 million synapses. Neural responses to both naturalistic and controlled visual stimuli were recorded using a 2P random access mesoscope (2P-RAM), while transmission EM achieved 4-nanometer resolution across 27,000 serial sections.
What sets this study apart is the precise alignment of functional data with fully reconstructed cell morphologies and synaptic connections. Vascular labeling and advanced computational alignment allowed researchers to co-register activity and anatomy with high fidelity. Automated segmentation reached 96 % precision in identifying cellular processes and synapses, which was further refined through extensive manual proofreading. Meanwhile, machine learning classified over 144,000 nuclei into neuronal and non-neuronal categories.
In total, nearly 2 petabytes of raw EM data, calcium imaging, and derived datasets—such as 3D neuron reconstructions and annotated synapses—have been made freely available for the scientific community.
This integration bridges scales, from single-cell tuning properties to circuit-level connectivity. Early analyses using the dataset have revealed insights into how different neuron types connect and transmit information across cortical layers. Moreover, the technical foundation developed here offers a model for applying large-scale connectomics methods to other regions of the brain and even other species.
Proofreading at Petabyte Scale
A key achievement of the MICrONS project was the integration of rigorous proofreading workflows into this massive dataset. Using the ChunkedGraph system, researchers made over one million collaborative edits across 1433 neurons, achieving 99 % accuracy in dendritic input identification and reconstructing axons up to 32.3 mm in length. Automated tools like NEURD (neural de-composition) handled 164,000 merge corrections, while manual proofreading focused on extending long axonal projections—at a rate of 400–600 edits per researcher per week.
To link function and structure, a three-phase co-registration process matched 75,909 imaged neurons to their EM counterparts using vascular landmarks and machine learning, achieving 90 % precision in automated matching. The final public dataset includes 337 million synapses, detailed cell-type labels, and functional activity data, all accessible via platforms like Neuroglancer and the Connectome Annotation Versioning Engine (CAVE).
Analyses of the proofread neurons revealed distinctive connectivity patterns across layers—for instance, layer 3 pyramidal cells showed selective targeting of inhibitory subtypes. In one remarkable case, a single neuron’s 1400+ outputs were traced to postsynaptic partners with known functional responses, something previously impossible at this scale.
The Virtual Observatory of the Cortex project continues to build on this work, expanding and refining the dataset, which now includes the most extensive axonal reconstructions ever produced in neocortical EM data. The MICrONS framework not only sets a new technical bar but also enables collaborative, community-driven discovery in brain science.
Conclusion
The MICrONS dataset delivers a major advance in connectomics, offering the first complete view of how activity and wiring come together at millimeter scale in the brain. By integrating calcium imaging of 75,000 neurons with EM reconstructions of over 200,000 cells and half a billion synapses, the project brings unprecedented clarity to questions about neural computation, connectivity, and visual processing.
With high-precision segmentation (96 %), expertly proofread reconstructions, and machine-learning-guided co-registration (90 %), the dataset provides a robust platform for understanding cortical microcircuits. While challenges remain, particularly in extending axonal coverage and expanding functional sampling, the current resource already supports detailed exploration of connectivity rules and information flow.
Looking ahead, future iterations will incorporate transcriptomic data and further refine structural accuracy. As the first mm3-scale functional connectome, MICrONS not only delivers new insights into the visual cortex but also lays a foundation for mapping the brain at systems level—one circuit at a time.
Journal Reference
Bae et al., (2025). Functional connectomics spanning multiple areas of mouse visual cortex. Nature, 640(8058), 435–447. DOI:10.1038/s41586-025-08790-w. https://www.nature.com/articles/s41586-025-08790-w
Disclaimer: The views expressed here are those of the author expressed in their private capacity and do not necessarily represent the views of AZoM.com Limited T/A AZoNetwork the owner and operator of this website. This disclaimer forms part of the Terms and conditions of use of this website.