Machine learning helps some of the best microscopes to see better, work faster, and process more data
From:
European Molecular Biology Laboratory
May 9, 2021 -- Collaboration between
deep learning experts and microscopy experts leads to an significantly improved
data-intensive light-field microscopy method by using AI and ground-truthing it
with light-sheet microscopy. The result is the power of light-field microscopy
available to biologists in near real time vs. days or weeks, AND the expansion
of biologists' ability to use this microscopy for many things more things
requiring the most detailed observation.
To observe the swift neuronal signals in
a fish brain, scientists have started to use a technique called light-field
microscopy, which makes it possible to image such fast biological processes in
3D. But the images are often lacking in quality, and it takes hours or days for
massive amounts of data to be converted into 3D volumes and movies.
Now, EMBL scientists have combined
artificial intelligence (AI) algorithms with two cutting-edge microscopy
techniques -- an advance that shortens the time for image processing from days
to mere seconds, while ensuring that the resulting images are crisp and
accurate. The findings are published in Nature Methods.
"Ultimately, we were able to take
'the best of both worlds' in this approach," says Nils Wagner, one of the
paper's two lead authors and now a PhD student at the Technical University of
Munich. "AI enabled us to combine different microscopy techniques, so that
we could image as fast as light-field microscopy allows and get close to the
image resolution of light-sheet microscopy."
Although light-sheet microscopy and
light-field microscopy sound similar, these techniques have different
advantages and challenges. Light-field microscopy captures large 3D images that
allow researchers to track and measure remarkably fine movements, such as a
fish larva's beating heart, at very high speeds. But this technique produces
massive amounts of data, which can take days to process, and the final images
usually lack resolution.
Light-sheet microscopy homes in on a
single 2D plane of a given sample at one time, so researchers can image samples
at higher resolution. Compared with light-field microscopy, light-sheet
microscopy produces images that are quicker to process, but the data are not as
comprehensive, since they only capture information from a single 2D plane at a
time.
To take advantage of the benefits of
each technique, EMBL researchers developed an approach that uses light-field
microscopy to image large 3D samples and light-sheet microscopy to train the AI
algorithms, which then create an accurate 3D picture of the sample.
"If you build algorithms that
produce an image, you need to check that these algorithms are constructing the
right image," explains Anna Kreshuk, the EMBL group leader whose team
brought machine learning expertise to the project. In the new study, the
researchers used light-sheet microscopy to make sure the AI algorithms were
working, Anna says. "This makes our research stand out from what has been
done in the past."
Robert Prevedel, the EMBL group leader
whose group contributed the novel hybrid microscopy platform, notes that the
real bottleneck in building better microscopes often isn't optics technology,
but computation. That's why, back in 2018, he and Anna decided to join forces.
"Our method will be really key for people who want to study how brains
compute. Our method can image an entire brain of a fish larva, in real
time," Robert says.
He and Anna say this approach could
potentially be modified to work with different types of microscopes too,
eventually allowing biologists to look at dozens of different specimens and see
much more, much faster. For example, it could help to find genes that are
involved in heart development, or could measure the activity of thousands of
neurons at the same time.
Next, the researchers plan to explore
whether the method can be applied to larger species, including mammals.
No comments:
Post a Comment