Daniel Berger is usually found peering into the microscopic world of neurons, synapses, and brain circuitry. Using cutting-edge imaging techniques and computational methods, Berger, Research Scientist within the lab of MCB’s Jeff Lichtman is helping to construct detailed 3D maps of brain tissue that could revolutionize our understanding of how the brain processes information and generates intelligence.
Originally from Karlsruhe, Germany, Berger’s path to neuroscience was not direct. He completed his undergraduate degree in computer science in his hometown before moving to Tübingen to pursue a PhD at the Max Planck Institute. There, he studied human psychophysics–using behavioral experiments to understand how humans perceive and interact with their environment.
“Essentially, human psychophysics is a way to do experiments to figure out how humans perceive their environment or how their brain works at a very high level,” Berger explains. His PhD work involved putting subjects in elaborate setups, like a robotic hexapod platform, to study how people integrate different sensory inputs to estimate their orientation and movement in space.
Computer Science to Connectomics to Brain Imaging
While this work provided insights into human perception, Berger became increasingly curious about the biological underpinnings of these cognitive processes. “I became interested in not only how you can make a model of how humans perceive something, but how is that actually implemented. What is the actual biological foundation of this processing?” he says.
This curiosity led Berger to MIT in 2009 for a postdoc position, where he entered the emerging field of connectomics—the study of comprehensive maps of neural connections in the brain. He joined Sebastian Seung’s lab (now at Princeton) and collaborated closely with Lichtman’s lab to develop computational methods for analyzing high-resolution brain imaging data.
Together, the labs developed a technique involving extremely thin slices of brain tissue – as thin as 30 nanometers – and imaging them with electron microscopes to reveal the intricate structures of neurons and their connections. Berger helped develop algorithms to align and stack thousands of these 2D image slices to reconstruct 3D volumes of brain tissue including its circuitry.
“We can see the machinery, we can see vesicles with neurotransmitter in it, we can see where the contact is and which side sends a signal to which other side,” Berger explains, describing the level of detail visible in their reconstructions. “We can actually see the connection, where the plug is plugged in, where they are touching and have a functional contact where one cell talks to another.”
Building the Largest 3D Brain Reconstruction
In 2014, Berger officially joined Jeff Lichtman’s MCB lab, rising to the position of Research Scientist in January. Over the years, he has played a key role in dramatically scaling up the size and resolution of their brain tissue reconstructions.
“Daniel is a one-of-a-kind scientist with enormous talent and experience, and I am so pleased he is a research scientist in my group,” says Lichtman. “He is an expert in the new field of connectomics and has an unusual gift in rendering and analyzing serial section electron microscopy data. His renderings are beautiful, and he has the keenest eye I know for finding interesting features in the tangled web of wires that comprise a brain.”
The Lichtman Lab’s latest dataset, of which Berger is co-first author, captures incredible details of a cubic millimeter of human brain tissue. The work was originally published in Science in May and featured prominently in many other outlets, including MCB News, the Harvard Gazette, the Boston Globe, WIRED, among others. It is the largest 3D brain reconstruction ever made and about 8,000 times larger than their first published dataset in Cell from 2015. This monumental undertaking required collaboration with Google to leverage their data processing capabilities.
The wealth of data in these reconstructions is staggering. The cubic millimeter sample contains about 57,000 cells and all their intricate branches and connections. Berger and his colleagues are now analyzing this data to uncover the underlying principles of how neurons connect and form functional circuits.
One of Berger’s current interests is a subtype of inhibitory neurons called chandelier cells. “I’ve been working to make models of these cells and trace how they are connected, particularly to pyramidal neurons in the same tissue, the main neuron type in the brain,” he explains. These neurons form connections with the axons of other neurons, potentially acting as gatekeepers that control the output of neural circuits. But that task is a literal search for a needle in a haystack since there are only about 120 chandelier cells in the 57,000 cells – or about 0.21 percent of all the cells in the current dataset. By mapping the connections between chandelier cells in detail, Berger hopes to gain insight into how these cells contribute to information processing in the brain.
“Unfailingly” Generous with Colleagues
Known to be humble and generous with colleagues, Berger sees himself in a tandem role at MCB. “My work is essentially helping other people and doing my own research,” he says. Because of his decade-long experience scrutinizing electron-microscopic brain images and training as a computer scientist, Berger has become an expert in creating stunning visualizations of brain tissue—not just for his own work but also for his colleagues. He is the go-to person for programming and automating the lab’s microscopes, including writing the software to do proofreading.
“Daniel is unfailingly generous with his code and hands-on mentorship of anyone who wants to use it,” says Physics Professor Aravinthan Samuel. He is the sole author of VAST, a manual volumetric annotation software for connectomics that Samuel explains is the best in the field. “Daniel’s software continues to be a widely used engine that he makes readily available,” Samuel says, adding that hundreds of labs, possibly thousands of people (including him), have used Berger’s code.
“The best way to learn about the brain is to delve into its anatomy, and Daniel’s software is still the best way to do it,” Samuel adds. “I often see Daniel teaching summer students, visiting students, really anyone with an interest in the brain, how to analyze synaptic connectivity and wiring with his software.”
Merging Science with Art
Berger is widely known for helping his colleagues prepare stunning presentations – making 3D images out of their data for publication in papers. As an example, he is currently making images for a project on the cerebellum run by another postdoc in the lab, Nagaraju Dhanyasi. “I’m creating those images because I know how to use 3D rendering software –making the model, setting up the scene with lights, and then making nice pictures out of that”, he adds.
His 3D renderings, which transform microscope data into colorful, intricate landscapes of neurons and cellular structures, have been featured in scientific publications, books, and even at the Museum of Natural History in New York. His visualizations have helped bring the beauty and complexity of the brain to both scientific and public audiences.
Berger looking at one of his images at MIT Museum in 2018
On a purely creative front, Berger explores generative AI to make a new generation of striking images. He also enjoys making songs using a generative AI music tool that creates new music out of a short description.
Because Daniel has an artist’s eye, many of the most beautiful movies and images that emerged from the Lichtman lab were created by Berger. “He could have a high-end career in studio animation, but is very happy and content to be a teacher and communicator, not by being the guy in the limelight, but by being the one who gives the guy in the spotlight all the tools he needs to succeed,” says Samuel.
Looking ahead, Berger and his colleagues are pushing towards even more ambitious goals. “Jeff Lichtman has a grant to get towards imaging the whole mouse brain, which is about 1000 cubic millimeters, so 1000 times larger than our human data”, he explains.
As connectomics continues to advance, Berger remains driven by the fundamental question that first drew him to neuroscience: How does the brain implement human-level intelligence through its biological circuitry?
“It’s a big circuit that may have some regularities to it. Like a computer chip has regularities to it, it wouldn’t work if it was randomly connected,” Berger muses. “So there has to be some structure that we are trying to get at by using these large datasets to figure out how biology manages to implement human-level intelligence in a bunch of cells that are connected.”
Through his unique combination of computer science expertise and neuroscience insight, Berger is helping to unravel one of science’s greatest mysteries – the intricate wiring of the human brain. His work not only advances our fundamental understanding of brain structure and function but also lays the groundwork for future breakthroughs in artificial intelligence, brain-machine interfaces, and the treatment of neurological disorders.
As imaging techniques and computational methods continue to evolve, Berger and his colleagues are at the forefront of a revolution in brain mapping, peering ever deeper into the neural forests that give rise to our thoughts, memories, and consciousness.
