The challenge of measuring tissue topography with a bright field endoscope has been of longstanding interest in endoscopic imaging. Most commercial endoscopes have an illumination pathway separate from a collection channel, which measures the light that reflects from the tissue to create an image at the endoscope proximal end. Since most endoscopes are composed of thousands of tiny optical fibers, each one of them acting as an image pixel, the only information that makes it through the endoscope is the bright or dark level of the backreflection. Could there be a way to couple other optical properties through the fiber as well? Since light in a wave, it has a wavefront shape that is modulated on passing through optically dense medium, much like ripples on a lake. Polarization and color are other degrees of freedom that can be used to communicate information and generate contrast. Could we improve conventional endoscopes with a minimal addon that utilizes these extra degrees of freedom in light – one that is low cost, compact, and easily translatable to clinic ?
When MEMS endoscope inventor Eric Seibel, metaoptics pioneer Arka Majumdar, and post doctorate fellow Aamod Shanker got together at the Unviersity of Washingtion in 2023, scratching their heads together over a few zoom meetings, they figured out an ingenous solution to use metamaterials to improve endoscopy.
Arka's lab was by the time already successfully making Fresnel zone plates on Silicon Nitride (a highly refractive transparent material), with the increasingly narrow radial rings of a traditional Fresnel zone plate replaced by vertical pillars with shrinking diameters in these new “meta-lenses” . These metalenses were being optimized to mitigate the color dependent focus seen in traditional Fresnel lenses, with the color splitting along focus considered a nuisance. Eric on the other hand was a couple of decades deep into creating micro mechanical actuators to scan the illumination at the fiber tip, and build an image with endoscopes. They were together looking to minimize the rigid tip length and remove the moving parts out of the system while enhancing contrast - using the miniature optical element, the metaoptic.
Aamod with his background in ophthalmic phase imaging entered the conversation, and together the three put the puzzle pieces together to arrive at a potential solution- using the color splitting of a metalens to scan the focal depth of the endoscope! Instead of mechanical motion, now the spatial information would be encoded in the color channels - each color focusing at a different depth, hence imaging various volumetric cross-sections in its absorption signature. Though similar to optical coherence tomography, where the colours in a broadband source provide depth information in the eye, the method promised brightfield 3D imaging at video rate without a mechanical scanning mirror.
Aamod's experience in quantitative phase imaging helped him resolve another piece of the puzzle- how to represent the 3D data captured by the metalens enabled endoscope? Since they had a color camera at hand to measure the light after it had bounced back through the endoscope, they decided to start with three colors - red, green and blue, built into any color camera. Now, even though the optical fiber itself scrambles the wavefront phase as it passes through, the relationship between the intensity at three depths (encoded into the color channels) has a deterministic correlation based on the wavefront shape before the light entered the endoscope!
There it was - an algorithm took form, where the three color channels of the camera at the far end of the optical fiber would give three images at different depths. Using a inverse algorithm (called the transport of intensity method), they were able to combine the three slightly defocused images (by calculating a derivative with propagation z) to reconstruct the wavefront shape at the sample end of the fiber... hence imaging the quantitative phase of light scattered by the tissue. QPI is a really wonderful biological metric, since the wavefront shape is determined both by the optical density and the thickness of the cells being measured, for instance. Hence there are two types of information embedded in the quantitative phase image - about the internal states as the optical density, and about external topography as the tissue thickness.
Once the idea took fruit, the authors came together with other post-docs and students in the lab to bring the prototype to completion - acquiring the first metalens enabled phase images in endoscopy. The differential nature of the algorithm further meant that much of the common mode fuzzy noise typical of in-vivo imaging automatically disappeared.
After this initial success, the authors hope to extend the volumetric imaging principle with three colors to the hyperspectral case - generating complete tomographic datasets using a white light source and multiple wavelengths packed into its chromatic bandwidrh. As metalenses continue to develop, this volumetric or phase imaging approach will be one of many in the endoscopist´s toolkit - including also wider field of view capabilities , polarization contrast imaging, angular momentum imaging and synthetic aperture high resolution imaging among others. As technologists, we continue to create new methods that may let the clinician and the molecular biologist peek into cellular worlds that have eluded us thus far. Stay tuned!