Converting 2D Images into 3D Using Deep Learning

A research team at the Univer­sity of California, Los Angeles, has devised a technique that extends the capa­bilities of fluores­cence micro­scopy, which allows scientists to precisely label parts of living cells and tissue with dyes that glow under special lighting. The researchers use arti­ficial intelli­gence to turn two-dimen­sional images into stacks of virtual three-dimensional slices showing activity inside organisms. Now, the scientists reported that their framework, called “Deep-Z,” was able to fix errors or aberra­tions in images, such as when a sample is tilted or curved. Further, they demons­trated that the system could take 2D images from one type of microscope and virtually create 3D images of the sample as if they were obtained by another, more advanced microscope.

This illustration represents Deep-Z, an artificial intelligence-based framework that can digitally refocus a 2D fluorescence microscope image (at bottom) to produce 3D slices. (Source: Ozcan Lab, UCLA)

“This is a very powerful new method that is enabled by deep learning to perform 3D imaging of live specimens, with the least exposure to light, which can be toxic to samples,” said Aydogan Ozcan, UCLA chancellor’s professor of electrical and computer engi­neering. In addition to sparing specimens from poten­tially damaging doses of light, this system could offer biologists and life science researchers a new tool for 3D imaging that is simpler, faster and much less expensive than current methods. The oppor­tunity to correct for aberra­tions may allow scientists studying live organisms to collect data from images that otherwise would be unusable. Investi­gators could also gain virtual access to expensive and complicated equipment.

This research builds on an earlier technique Ozcan and his colleagues developed that allowed them to render 2D fluores­cence microscope images in super-resolution. Both techniques advance micro­scopy by relying upon deep learning – using data to train a neural network, a computer system inspired by the human brain. Deep-Z was taught using experimental images from a scanning fluores­cence micro­scope, which takes pictures focused at multiple depths to achieve 3D imaging of samples. In thousands of training runs, the neural network learned how to take a 2D image and infer accurate 3D slices at different depths within a sample. Then, the framework was tested blindly – fed with images that were not part of its training, with the virtual images compared to the actual 3D slices obtained from a scanning micro­scope, providing an excellent match.

Ozcan and his colleagues applied Deep-Z to images of C. elegans, a roundworm that is a common model in neuro­science because of its simple and well-understood nervous system. Converting a 2D movie of a worm to 3D, frame by frame, the researchers were able to track the activity of individual neurons within the worm body. And starting with one or two 2D images of C. elegans taken at different depths, Deep-Z produced virtual 3D images that allowed the team to identify individual neurons within the worm, matching a scanning microscope’s 3D output, except with much less light exposure to the living organism.

The researchers also found that Deep-Z could produce 3D images from 2D surfaces where samples were tilted or curved – even though the neural network was trained only with 3D slices that were perfectly parallel to the surface of the sample. “This feature was actually very sur­prising,” said Yichen Wu, a UCLA graduate student. “With it, you can see through curvature or other complex topology that is very challen­ging to image.”

In other experiments, Deep-Z was trained with images from two types of fluores­cence micro­scopes: wide-field, which exposes the entire sample to a light source; and confocal, which uses a laser to scan a sample part by part. Ozcan and his team showed that their framework could then use 2D wide-field micro­scope images of samples to produce 3D images nearly identical to ones taken with a confocal microscope. This conversion is valuable because the confocal micro­scope creates images that are sharper, with more contrast, compared to the wide field. On the other hand, the wide-field micro­scope captures images at less expense and with fewer technical requirements.

“This is a platform that is generally applicable to various pairs of micro­scopes, not just the wide-field-to-confocal conversion,” said Yair Rivenson, UCLA assistant adjunct professor of electrical and computer engi­neering. “Every micro­scope has its own advantages and disad­vantages. With this framework, you can get the best of both worlds by using AI to connect different types of micro­scopes digitally.” (Source: UCLA)

Reference: Y. Wu et al.: Three-dimensional virtual refocusing of fluorescence microscopy images using deep learning, Nat. Meth., online 4 November 2019; DOI: 10.1038/s41592-019-0622-5

Link: California NanoSystems Institute, UCLA University of California, Los Angeles, USA

Speak Your Mind