New imaging method reveals hidden structures in brain wiring maps
A faster electron-tomography technique developed at Harvard Medical School and UC San Diego captured electrical synapses and neuron-glia contact points that standard brain-mapping tools routinely miss.
Researchers at Harvard Medical School, Boston Children's Hospital and the University of California San Diego have developed a faster method for imaging brain tissue at the nanoscale, one that captures fine cellular structures missed by the imaging techniques now used to build "connectomes," the detailed wiring diagrams of how neurons link together. The method and the structures it uncovered were described Nov. 19 in the journal Nature Methods, in a paper titled "Scalable electron tomography for connectomics."
Connectomics is the effort to map every neuron and synapse in a piece of brain tissue, work that researchers hope will eventually explain how circuits give rise to behavior and how they break down in disease. Mapping has typically relied on two imaging approaches: MRI-based scanning, which shows large-scale pathways but not individual cells, and standard electron microscopy, which traces individual neurons and synapses but loses finer detail across large tissue volumes. Both methods tend to miss certain structures altogether, among them electrical synapses, the direct, nearly instantaneous junctions between neurons that are distinct from the chemical synapses most connectome studies track.
A faster way to look
The new technique builds on electron tomography, a method that generates three-dimensional images by photographing a tissue sample from many angles and reconstructing the volume computationally. Electron tomography has existed for years but has been too slow for mapping large brain volumes, because conventional versions require hundreds of images of the same region from different tilt angles.
The team, led by Aaron T. Kuan along with co-senior author Wei-Chung Allen Lee, a neurobiologist at Harvard Medical School and Boston Children's Hospital, found a way around that bottleneck. Their approach used about 50 times fewer images than standard electron tomography, applied to tissue sections roughly 20 times thicker than researchers typically use, while still achieving three-dimensional resolution of about 15 nanometers or finer — fine enough to make out individual synaptic structures. Mark Ellisman's group at UC San Diego, which operates specialized imaging facilities for neuroscience, contributed to the imaging work.
Applied to tissue from the mouse cerebellum, the method turned up structures that are easy for other approaches to miss or misinterpret. The researchers identified electrical synapses, or gap junctions, which allow signals to pass between neurons almost instantly and are thought to help synchronize activity across a network. They also found small protrusions where neurons and neighboring glial cells — the brain's support cells — appear to be in unusually close contact. The imaging suggested one cell type may be absorbing small amounts of material from the other, a process the researchers had not expected to see so clearly.
"We're seeing uncharacterized structures, suggesting that neurons are 'nibbling' their neighboring cells," Lee said, describing the contact points between neurons and glia.
Why the gaps mattered
Connectomics has advanced rapidly in the past several years. Large electron-microscopy projects have produced wiring diagrams of mouse visual cortex and fragments of human cortical tissue, and Nature Methods named EM-based connectomics its 2025 "Method of the Year," reflecting how central the approach has become to neuroscience. But researchers in the field have long acknowledged that standard imaging, built around speed and scale, trades away some structural detail. Electrical synapses and fine glial contacts are small and easy to overlook, yet both are thought to matter for how circuits function, develop and change with age.
Lee's lab has spent years building tools aimed at that problem, including earlier work combining large-scale electron microscopy with imaging of brain activity, an approach the group calls functional connectomics. The new tomography method does not replace existing connectome-mapping pipelines; rather, the authors frame it as a complementary tool that can be layered onto large-volume reconstructions to recover detail that would otherwise be lost, without requiring researchers to redo the entire imaging process at a slower speed.
Who stands to benefit
The immediate beneficiaries are researchers assembling large connectome datasets, who have had to choose between imaging speed and structural completeness. A faster, more detailed tomography pass could let them revisit flagged regions of a sample, such as the HMS release notes, without the time cost of running an entire dataset through traditional tomography. The authors point to broader downstream uses as well. Because electrical synapses and glia-neuron interactions are implicated in neural development, aging and a range of neurological and psychiatric disorders, more complete maps of those structures could inform research into conditions ranging from neurodevelopmental disorders to neurodegenerative disease, though the current paper does not test any disease model directly — the demonstration was done in healthy mouse cerebellar tissue.
Lee framed the work in terms of that longer-term goal. "We hope this work will help us understand how nervous systems are built, what changes with age, and what goes awry" in neurological and psychiatric disorders, he said, according to the Harvard Medical School announcement of the findings.
What comes next
The authors are explicit that their current results are a proof of concept rather than a finished mapping pipeline. Scaling the method to image an entire mammalian brain, the stated long-term ambition of much of the connectomics field, will require substantial additional engineering, particularly around automating the handling of tissue samples, a step that remains labor-intensive even with recent advances in electron microscopy throughput. The Nov. 21 Medical Xpress account of the findings notes that the Harvard and UC San Diego researchers consider full-brain application a long-term goal rather than an immediate next step.
The paper adds to a wave of methods papers aimed at making connectomics faster and cheaper. It reflects a broader strategic shift in the field: rather than simply building bigger, slower microscopes, researchers are increasingly using computation and smarter sampling to extract more information from the imaging passes they can already afford to run, a shift that could determine how quickly whole-brain wiring maps move from an aspirational goal to a routine laboratory tool. The underlying paper is cataloged at its Nature Methods DOI, where the full methodology is available to other labs seeking to adopt the technique.

UK-Built X-ray Telescope Opens Its Eyes for the First Time as Mercury Probe Nears Arrival
New Theory Predicts the Power Fusion Reactors Need to Reach Peak Efficiency

Europa's sodium haze points to a hidden ocean, not Io's volcanoes, astronomers find
