Scientists create realistic brain-wide connection maps through digital modeling

EPFL researchers have developed a powerful method to generate brain-wide, biologically realistic wiring maps of the mouse brain. Their approach bridges experimental data with mathematical and computational modeling to simulate how neurons connect across the entire brain.

The study is published in the journal Nature Communications.

One of neuroscience’s greatest challenges is understanding how the brain is wired. Even with modern imaging tools, it has been a challenge to create detailed maps that show how the brain’s billions of cells (neurons) connect, not just with their local “neighbors” but also to other, more distant cells in the brain.

These maps, known as connectomes, are key to unlocking the secrets of brain function and disease. But traditional brain-mapping techniques only offer a partial view.

Even though experimental datasets are growing, they are still too sparse to reconstruct all the connections that matter, especially those that bridge distant brain regions. As a result, it is still difficult to understand complex cognitive functions or pinpoint the cause of neurological diseases.

A model for research and medical applications

Scientists in the group of Professor Henry Markram at EPFL’s Blue Brain Project have now developed a way to generate digital (“synthetic,” as they’re known in the field), but biologically realistic, brain-wide connection maps.

Led by Lida Kanari, they created detailed digital models of how neurons extend their wiring throughout the brain, taking a step closer to building comprehensive connectomes that can be used for both research and medical applications.

Working with large datasets of biological “axonal reconstructions,” including new data collected with the collaborating team of Professor Hanchuan Peng (Southeast University, China), they used machine learning to group neurons by their wiring patterns.

Remy Petkantchin, the first author of the study, developed a powerful computational method that generates synthetic axons matching these wiring patterns. The method was based on a mathematical model they developed in 2022 to generate digital copies of neurons that mimic how real ones branch out and connect to various brain regions.

The synthetic axons were designed to follow the same paths as their biological counterparts, so the resulting connectomes also reflected how neurons are connected in an actual brain.

Realistic wiring

The synthetic axons match key features of biological ones, such as how they look and where they connect. When they are used to build brain-wide networks, the resulting connectome closely resembles those built from experimental data, including the critical long-range connections that link distant brain areas.

By generating thousands of synthetic axons, the team created a digital mouse brain model with realistic wiring. This means that it is now possible to fill gaps in existing connectome datasets, explore how neurons connect across the brain, and even test ideas about brain organization that would be impossible to study in living animals.

This research opens new possibilities for neuroscience. Digital connectomes can support large-scale brain simulations, guide experiments, and offer new insights into neurological diseases.

Although the study focused on the mouse brain, the same principles could be applied to other species, including humans, as more data becomes available.