Â鶹ÒùÔº

Astronomy trick enables researchers to capture high-speed, 4D videos of moving organisms

Biomedical engineers at Duke University have developed a computational imaging system that borrows techniques from astronomy to reconstruct 4D videos of freely moving small model organisms, like zebrafish and fruit fly larvae. By using a concave mirror and an array of sensors, researchers were able to rapidly capture the unrestrained movements of animals from dozens of synchronized viewpoints.

This imaging tool will enable scientists to conduct more precise behavioral studies of model organisms. The research was May 5 in the journal Optica.

Behavioral studies of animals have revolutionized both basic science and medical research by illustrating connections between brain activity, the animal's environment and physical responses. However, the ability to capture detailed, 3D videos of these behaviors has proven to be a challenge.

Researchers can use multiple cameras to capture an animal's movement over a larger area, but shrinking this down to microscopically observe smaller organisms at cellular resolution is not straightforward.

Typically, microscopes scan a lens or a beam to capture 3D imagery, which makes it more likely for motion artifacts to appear in the recording, especially with rapidly moving organisms. Scientists will sometimes constrain or sedate organisms so they can see them clearly under the microscope, but this limits the very behaviors they're trying to study.

To overcome these obstacles, Roarke Horstmeyer, an assistant professor of biomedical engineering, and his lab developed reflective Fourier light field computed tomography, or ReFLeCT. By combining an array of 54 cameras, a parabolic mirror and new computational algorithms, the team was able to reconstruct fast, 4D videos of zebrafish and fruit fly larvae.

To capture a video, a small tube containing the sample organism hangs at a fixed point in the middle of the concave mirror, which resembles a bowl with a reflective interior. A hole at the bottom of the bowl allows a blue LED light to shine into the tube to illuminate the sample, which is tagged with a fluorescent dye.

An array of cameras is positioned in a grid just above the bowl. Each camera captures a different angle of the animal's reflection off the mirror, which a computational algorithm then uses to reconstruct a 3D image. These images are spliced together to create a 4D video, where the researchers can view the movements of the animal over a set period of time.

"Astronomers use concave mirrors to capture their own images, but I'd never heard about it being used in this context," said Horstmeyer. "With other 3D techniques, you'll often have to move your focus as your model moves, but we don't have to move anything because the cameras are synchronously capturing data from almost every perspective.

"The setup ensures that we can create accurate tomographic 3D reconstructions with almost no motion blur and ensure that the animal doesn't leave the array's field of view."

To test their setup, Horstmeyer and his team imaged both zebrafish and fruit fly larvae, two common animal models in behavioral studies. Although the resolution of their images was strongest at the bottom of the sample tube, they were still able to capture and create 3D videos of the samples of up to 120 volumes per second—tens of times faster than most other volumetric microscopy techniques that operate in this regime.

This improved speed and resolution enabled the team to observe zebrafish behaviors that previously hadn't been visible with traditional tools. "We saw that the fruit fly larvae were making wiggle-like movements related to peristalsis, which is the same thing that our throat does when we swallow," said Horstmeyer. "It showed that we are able to capture these very small, precise dynamic movements."

The team already has plans to improve their system and create new collaborations with scientists both inside and outside of Duke.

"This tool has gotten biologists excited because they can see these 3D movements and study them at high speed, so it's been really satisfying to see them eager to use this tool," he said. "I think it would be really exciting to be able to use this in neuro-imaging studies as well, so we're already looking at ways to upgrade the system for expanded uses."