Researchers have completed the most advanced brain map yet, an insect map, a landmark achievement in neuroscience that brings scientists closer to a true understanding of the mechanism of thinking.
The international team led by Johns Hopkins University and the University of Cambridge has produced a stunningly detailed diagram that tracks every neural connection in the brain of a larval fruit fly, an original scientific model with brains similar to humans.
The work, which is likely to underpin future brain research and inspire new constructs for machine learning, appears in the journal today Sciences.
“If we want to understand who we are and how we think, part of that is understanding the mechanism of thinking,” said senior author Joshua T. Vogelstein, a biomedical engineer at Johns Hopkins University who specializes in data-driven projects including connections, in the study. of nervous system connections. “The key to this is figuring out how neurons communicate with each other.”
The first attempt at mapping the brain – a 14-year study of roundworms that began in the 1970s, resulted in a partial map and a Nobel Prize. Since then, the partial neural network has been mapped into many systems, including flies, mice, and even humans, but these reconstructions usually represent only a small part of the overall brain. Universal neural networks have only been created for several small species with a few hundred to a few thousand neurons in their bodies – roundworms, sea squirt larvae, and sea squirt larvae.
This team’s neural network for a small fruit fly, Drosophila melanogaster Larva, is the most complete and most extensive map of an entire insect brain ever produced. It contains 3,016 neurons and each connection between them: 548,000.
“It’s been 50 years and this is the first brain neural network,” Vogelstein said. “It’s science in the sand that we can do this.” “Everything works so far.”
Mapping entire brains is challenging and extremely time consuming, even with the best of modern technology. Getting a complete picture at the cellular level of the brain requires dissecting the brain into hundreds or thousands of individual tissue samples, all of which must be imaged using electron microscopes before the painstaking process of reconstructing all those pieces, neuron by neuron, into whole parts. , an accurate picture of the brain. It took more than a decade to do this with the tiny fruit fly. The mouse brain is estimated to be a million times larger than that of a small fruit fly, which means that the chance of mapping anything close to the human brain is not likely in the near future, perhaps even in our lifetime.
The team purposefully chose a fruit fly larva, because the insect shares much of its basic biology with humans, including a comparable genetic basis. It also has rich learning and decision-making behaviours, making it a useful living model in neuroscience. For practical purposes, his relatively compact brain could be imaged and its circuits reconstructed in a reasonable time frame.
However, it took Cambridge University and Johns Hopkins 12 years to complete the work. Imaging alone took about a day per neuron.
Cambridge researchers created high-resolution images of the brain and manually studied them to find individual neurons, meticulously tracking each one and correlating their synaptic connections.
Cambridge handed the data over to Johns Hopkins University, where the team spent more than three years using the original code they had created to analyze brain connectivity. The Johns Hopkins team developed techniques to find groups of neurons based on shared connectivity patterns, and then analyzed how information might propagate through the brain.
In the end, the entire team charted every neuron and every connection, classifying each neuron according to the role it plays in the brain. They found that the busiest brain circuits are those leading to neurons at and away from the learning center.
The methods developed by Johns Hopkins are applicable to any brain connectivity project, and their code is available to anyone trying to map the brain of a larger animal, Vogelstein said, adding that despite the challenges, scientists are expected to tackle the mouse, perhaps within the next decade. Other teams are already working on a map of the brain of adult fruit flies. Co-first author Benjamin Pedigo, a Johns Hopkins PhD candidate in biomedical engineering, speculates that the team’s code could help reveal important comparisons between connections in the adult and larval brains. As neural networks are created for more caterpillars and from other related species, Pedigo expects that their analysis techniques could lead to a better understanding of differences in brain wiring.
The work of Drosophila larvae showed circuit features that are strikingly reminiscent of prominent and robust machine learning structures. The team expects that continued study will reveal more computational principles and possibly inspire new AI systems.
“What we’ve learned about the code for fruit flies will have implications for the code for humans,” Vogelstein said. “This is what we want to understand – how to write a program that leads to the human brain network.”
video: https://youtu.be/NXr0ZdoYgRw
The authors are: Michael Winding, Christopher L. Barnes, Heather G. Batsolic, Youngser Park, Tom Kazimierz, Akira Fushiki, Ingrid V. Andrade, Avinash Kandelwal, Javier Valdes Aleman, Feng Li, Nadine Randel, Elizabeth Barsotti, Anna Correa, Richard D. Vitter Volker Hartenstein, Carrie E. Preby, Albert Cardona, Marta Zlatek.
