Researchers have developed a new imaging and virtual reconstruction technology called LIONESS, which offers high-resolution imaging of live brain tissue, visualizing it in real-time 3D nanoscale detail. LIONESS integrates advanced optics, artificial intelligence and a collaborative interdisciplinary approach, overcoming the limitations of previous imaging methods and paving the way for a better understanding of the dynamics and complexity of brain tissue.
Collaborative efforts at ISTA yield an unprecedented “live” view of the brain’s complexity.
The human brain, with its intricate network of some 86 billion neurons, is one of the most complex specimens scientists have ever encountered. It contains an enormous, yet currently unmatched, wealth of information, at the cutting edge of computing equipment.
Understanding this level of complexity is challenging, requiring us to use advanced technologies that can decode the complex interactions that occur in the brain at a microscopic level. Thus, imaging emerges as an important tool in the field of neuroscience.
A new imaging and virtual reconstruction technology developed by Johann Danzel’s group at ISTA is a major leap forward in imaging brain activity and is aptly named LIONESS – Live Information Optimized Nanoscopy Enabling Saturated Segmentation. LIONESS is a pipeline for imaging, reconstructing and analyzing living brain tissue with comprehensiveness and spatial resolution.
LIONESS describes the complexity of dense brain tissue. a: Complex neuronal environment b: LIONESS can image and reconstruct a pattern illustrating many dynamic structures and functions in live brain tissue. Credit: Johan Danzel
“With LIONESS, for the first time, comprehensive, dense reconstruction of living brain tissue is possible. By imaging the tissue multiple times, LIONESS allows us to observe and measure dynamic cellular biology in the brain,” says first author Philip Welicki. “The output is a reconstructed image of the cellular arrangement in three dimensions, with time forming the fourth dimension, as samples last for minutes, hours or can be filmed in days,” he adds.
Collaboration and AI the key
The strength of LIONESS lies in the sophisticated optics and the two layers of deep learning – a method of artificial intelligence – that form its core: the first enhances image quality and the second identifies different cellular structures in a dense neuronal environment.
This pipeline is the result of collaboration between the Danzel Group, the Bickel Group, the Jonas Group, the Novarino Group and the Scientific Service Units of ISTA, as well as other international collaborators. “Our vision is to bring together a dynamic group of scientists with unique combined expertise across disciplinary boundaries who work together to bridge the technology gap in analyzing brain tissue,” says ISTA’s Johan Danzl.
Pipeline for direct brain tissue reconstruction. Acquisition of microscopy with optimized laser focus – Image processing (DL) – Segmentation (DL) – 3D visual analysis. Credit: Johan Danzel
Overcoming obstacles
Previously, it was possible to reconstruct brain tissue using electron microscopy. This method creates a sample image based on its interaction with electrons. Despite the ability to capture images at a resolution of a few nanometers—a millionth of a millimeter—electron microscopy requires fixing a sample in a biological state, which must be physically sectioned to obtain 3D information. Hence no dynamic information can be obtained.
Another previously known technique of light microscopy allows the observation of living systems and the recording of intact tissues by “optically” cutting them without physically placing them. However, light microscopy is severely hampered in its resolving power by the properties of the light waves it uses to produce images. Its best-case resolution is a few hundred nanometers, too coarse to capture important cellular details in brain tissue.
Scientists can break this resolution barrier by using super-resolution light microscopy. Recent work in this area called SUSHI (Super-Resolution Shadow Imaging) has demonstrated that applying dye molecules to the free space around cells and using the Nobel Prize-winning super-resolution technique STED (Stimulated Emission Depletion) microscopy reveals super-resolution ‘shadows’. . ‘ of all cellular structures and thus makes it visible in tissue.
LIONESS can image and reconstruct the sample in a way that directly illustrates the many dynamic structures and functions within the brain tissue. Credit: Julia Ludchik ISTA
However, it has been impossible to image entire volumes of brain tissue with resolution enhancements that match the complex 3D architecture of brain tissue. This is because increased resolution also places a higher load of imaging light on the specimen, which can damage or ‘fry’ fine, living tissue.
Herein lies the feat of LIONESS, which, according to the authors, was developed for “fast and gentle” imaging conditions, thus keeping the sample alive. The technique does this while providing isotropic super-resolution—that is, it is equally good in all three spatial dimensions—which allows visualization of cellular components of tissues in 3D. Nanoscale Fix details.
LIONESS collects as little information as necessary from the sample during the imaging step. This is the first deep learning step to fill in additional information on the structure of brain tissue in a process called image restoration. In this innovative way, it achieves a resolution of about 130 nanometers, fine enough to image living brain tissue in real-time. Together, these steps allow for another step in deep learning, this time to interpret highly complex imaging data and identify neuronal structures in an automated manner.
ISTA scientist Johann Danzel in his laboratory at the Austrian Institute of Science and Technology. Credit: Nadine Poncioni | ISTA
Homing in
“The interdisciplinary approach allowed us to break the bounds of interconnectedness to resolve force and light exposure in living systems, realize complex 3D data, and link the cellular architecture of tissues with molecular and functional measurements,” says Danzel.
For the virtual reconstruction, Danzel and Welicki teamed up with visual computing experts: the Bickel group at ISTA and a group led by Hanspeter Pfister at Harvard University, who contributed their expertise in automatic segmentation—the process of automatically identifying cellular structures in tissues—and visualization, ISTA’s image analysis staff scientist Christoph With further support from Sommer. For sophisticated labeling strategies, neuroscientists and chemists from Edinburgh, Berlin and ISTA contributed.
As a result, it became possible to perform functional measurements, i.e. to read cellular structures with biological signaling activity in the same living neuronal circuit. This was done by imaging calcium ion flux in cells and measuring cellular electrical activity in collaboration with the Jonas group at ISTA. Novarino’s group contributed human cerebral organoids, often nicknamed mini-brains that mimic human brain development. The authors underline that all this is facilitated by expert support from ISTA’s high-quality scientific service units.
Brain structure and activity are highly dynamic; As the brain learns and performs new tasks, its structure evolves. This aspect of the brain is referred to as “plasticity”. Therefore, observing changes in brain tissue structure is essential to unravel the mysteries behind its plasticity. A new tool developed at ISTA shows the potential to understand the functional architecture of brain tissue and potentially other organs by revealing subcellular structures and capturing how they may change over time.
References: Philip Velicky, Eder Miguel, Julia M. Michalska, Julia Ludchik, Donglai Wei, Judy Lin, Jake F. “Dense 4D Nanoscale Reconstruction of Living Brain Tissue” by Watson, Jakob Troidal, Johanna Beyer, Yoav Ben-Simon, Christop Sommer, Wiebke Jahr, Alban Cenameri, Johannes Broichhagen, Seth GN Grant, Peter Jonas, Gaia Novarino, Hanspeter Pfister, Bernd Bickel and Johann G. Danzl, 10 July 2023, Nature’s method.
DOI: 10.1038/s41592-023-01936-6
The study was funded by the Austrian Science Fund, Gesellschaft für Forschungsförderung NÖ (NFB), H2020 Marie Skłodowska-Curie Actions, H2020 European Research Council, the Human Frontier Science Program, the Simons Foundation, the Wellcome Trust and the National Science Foundation. .