Since the early 1930s, electron microscopy has provided unprecedented access to the extraordinarily small alien world, revealing intricate details that are otherwise impossible to distinguish with the conventional light microscope. But to achieve a high resolution over a large area of specimens, the energy of the electron beams must be fixed, which is costly and detrimental to the specimen under observation.
Researchers at A&M University in Texas may have found a new method to improve the quality of low-resolution electron micrographs without compromising the integrity of the sample. By training deep neural networks, a kind of artificial intelligence algorithm, in pairs of images from the same sample, but with different physical resolutions, they have discovered that the details in lower resolution images can be further improved.
“Normally, a high-energy electron beam is passed through the sample in places where higher image resolution is desired. But with our image processing techniques, we can super solve an entire image using only a few larger images. small, high resolution, “said Dr. Yu Ding, Mike and Sugar Barnes Professor in Wm Michael Barnes ’64 Department of Industrial and Systems Engineering. “This method is less destructive as most parts of the sample do not have to be scanned with high-energy electron beam.”
The researchers published their image processing technique on Transactions of Electrical and Electronics Engineers in image processing in June.
Unlike light microscopy where photons, or tiny packets of light, are used to illuminate an object, in electron microscopy, an electron beam is used. Electrons reflected by or passing through the object are then gathered to form an image, called an electron micrograph.
Thus, the energy of the electron beams plays a crucial role in determining the resolution of the images. That is, the higher the energy electrons, the better the resolution. However, the risk of specimen damage also increases, similar to how ultraviolet rays, which are the most energetic relatives of visible light, can damage sensitive materials such as skin.
“There’s always that dilemma for scientists,” Ding said. “To maintain specimen integrity, high-energy electron beams are used sparingly. But if one does not use energy beam, high-resolution or the ability to see on the smallest scale becomes limited.”
But there are ways to get high resolution or super resolution using low resolution images. One method involves using many low-resolution images in essentially the same region. Another method learns common patterns between small image pieces and uses unrelated high-resolution images to enhance existing low-resolution images.
These methods almost exclusively use light natural images instead of electron micrographs. Therefore, they run into problems for electron micrographs that are super solvable as the basic physics for light and electron microscopy is different, Ding explained.
The researchers turned to pairs of low- and high-resolution electron microscopic images for a given sample. Although these types of couples are not very common in public image databases, they are relatively common in research materials and medical images.
For their experiments, Ding and his team first took a low-resolution image of a specimen and then subjected approximately 25% of the area under high-energy electron beam observation to obtain a high-resolution image. The researchers noticed that the information in the pair of high-resolution and low-resolution images are highly correlated. They said this property could be promoted even though the data available may be small.
For their analyzes, Ding and his team used 22 pairs of images of nanoparticle-injected materials. They then divided the high-resolution image and its equivalent area into the low-resolution image into three of three subdivisions. Next, each subconscious pair was used to “self-train” the deep neural networks. After training, their algorithm became known in recognizing image features, such as edges.
When they tested the deep-trained neural network at a new location in the low-resolution image for which there was no high-resolution counterpart, they found that their algorithm could improve features that were difficult to distinguish by up to 50%.
Although their image processing technique shows many promises, Ding noted that it still requires a lot of computing power. In the near future, his team will direct their efforts in developing algorithms that are much faster and can be supported by a smaller computing hardware.
“Our image processing pairing technique reveals details in low-resolution images that were not visible before,” Ding said. “We are all familiar with the magic wand feature in smartphones. It makes the image clearer. What we aim to do in the long run is to provide the research community with a similarly convenient tool to improve micrographs.” electronics. ”
Cryo-electron microscopy: Using a cheap technology to produce high resolution images
Yanjun Qian et al, Effective super-resolution method for paired electron microscopic images, IEEE transactions in image processing (2020). DOI: 10.1109 / TIP.2020.3000964
Provided by Texas A&M University
citation: New super resolution method reveals great details, without having to constantly zoom in (2020, August 13) Retrieved August 13, 2020 from https://techxplore.com/news/2020-08-super-resolution-method -reveals-fine- constantly.html
This document is subject to copyright. Except for any fair action for the purposes of private study or research, no part may be reproduced without our written permission. Content is provided for informational purposes only.