Optical fibers at ~125 µm are a promising material for cellular-level medical imaging within the human body. These fibers could be inserted for endoscopies, or to study organs. But the fibers invariably bend and flex, which degrades images. A research group at the University of Cambridge, led by George S.D. Gordon, has devised a new strategy that can instantly correct such image distortions by using a unique system of stacked reflectors and multiple light wavelengths.

In a perfect setting, light shines through the optical fiber to an object. The light that reflects back is represented in the form of a transmission matrix—data that can then be reconstructed mathematically into an image. However, the fibers are constantly in motion, bending with motion and temperature changes thereby distorting the images.

Current experimental systems that correct image distortions suffer from limitations. Bulky optics at the end of the fiber, for example, while effective, defeat the goal of inserting a very thin fiber easily into a body. 

New systems with corrective optics on the tips of the fibers, rather than bulky lenses, have been gaining popularity. Gordon, who was a member of Cambridge’s VisionLab headed by Sarah Bohndiek, and his colleagues propose a system that uses multiple wavelengths of light to create a calibration image. Their system model consists of fibers with a silica core, encased by a sheath, also made of silica, with a different index of refraction. They are known as step-index multimode silica optical fibers, and are most often used in medical or industrial applications. 

The fiber is tipped with a thin stack of structured metasurface reflectors. They consist of a 1-mm-thick layer of glass, then three commercial 3-mm-thick optical filters, which can be made of doped glass. In between each layer are extremely thin wire-grid polarizers. These are arrays of wires—in this case, made of silver or gold—on a substrate of silica. They reflect light at certain wavelengths. 

Three different wavelengths of light, from a tunable laser, shine down the fiber, and each wavelength reflects from one of the reflectors. The data from the three reflections are used to create a calibration map that can be used to translate the distorted image from the bent fibers into a clear image. This translation can then be used to correct the target image—similar to a reference star used in adaptive optics on telescopes.  To get the target image, a fourth wavelength of light goes through the entire fiber to capture a reflection from the target object. The image is then reconstructed using the calibration map. If the fiber bends again, the calibrations can be redone constantly and periodically to determine new corrections. The research group published their proof-of-concept work in a recent issue of Physical Review X. 

“This is a huge step forward for fiber endoscopy, as it paves the way to a self-calibrating endoscope,” says Dirk Boonzajer Flaes, a researcher at the Leibniz Institute of Photogenic Technology in Thuringia, Germany, who is not involved with the research. 

The Cambridge group’s breakthrough stemmed from solving a theoretical roadblock, in which reflection data contained mathematical ambiguities—similar to not knowing whether the square root of nine is three or negative three—that made it difficult to reconstruct a clear image. Their calculations showed that they could get around that problem by taking several calibration maps at slightly different wavelengths. 

Their calculations showed they need at least three wavelengths for the calibration. One wavelength does not work, and it’s also not clear why this strategy wouldn’t work with only two wavelengths, Gordon says.

Using measurements from real optical fibers, the group performed simulations of image reconstruction. They were able to obtain clear images from their data. 

Now that they have shown their method works, the group is working to construct a prototype system. 

Before this method can be used for real-life applications, Gordon says, they will need to increase imaging speed to deal with biological systems that can change in 50 ms. They will also be encasing the fiber in a material—such as the commercially available cultured diamond screen material PolyDiagnost—that will protect the system from body fluids and temperatures.

The method could be generalized in many ways, Gordon says. Different types of materials should work, including plastic and hollow core fibers. They could also use the method for different applications, including fluorescence microscopy, quantitative phase microscopy, and polarization microscopy. 

“It will be interesting to see how far this idea can be pushed,” Boonzajer Flaes says. 

Read the abstract in Physical Review X.