Hyperspectral imaging and metasurfaces combined for novel biosensing platform

03 Apr 2019

EPFL project could assist in personalized medicine and at-home disease detection.

The concept of at-home medicine relies on effective ways to identify diseases without needing the initial services of the healthcare industry, and in turn requires the development of new platforms to image and analyze biological species.

A project at the Swiss research institute EPFL in Lausanne has now developed an optical sensor that could be an important step along this road, allowing the straightforward identification of undesirable biomarkers in blood or saliva. The work was published in Nature Photonics.

The device, designed at EPFL’s BioNanoPhotonic Systems Laboratory (BIOS), involves the use of an optical chip coated in a metamaterial, a class of optically active material in which carefully designed surface nanostructures influence and control the behavior of light falling on the chip.

This optical component, combined with a standard CMOS camera and subsequent image analysis, is able to count biomolecules one by one in a sample and determine their location, according to the project team.

EPFL employed a specific type of metamaterial termed an all-dielectric asymmetric metasurface, in which the symmetry of the arrays of nano-scale units is deliberately broken. These surfaces are known to be capable of exhibiting high-quality optical resonances, arising from the behavior of the quantum states present in the material.

In this case, the nanostructures are able to squeeze certain frequencies of light into extremely small volumes, creating ultra-sensitive optical hotspots. When light shines on the metasurface and hits a biomarker molecule at one of these hotspots, the molecule can be detected immediately, thanks to the change in wavelength it brings about.

Personalized medicine

A second key aspect is the platform's use of hyperspectral illumination, to maximize the sensing efficiency and sensitivity of the device. CMOS images recorded at different wavelengths and passed through image analysis algorithms allowed the project to precisely count and analyze specific molecules in a sample.

"We combine dielectric metasurfaces and hyperspectral imaging to develop an ultra-sensitive label-free analytical platform for biosensing," commented the project team in its published paper. "Our technique can acquire spatially resolved spectra from millions of image pixels and use smart data processing tools to extract high-throughput digital sensing information at the unprecedented level of less than three molecules per square micron."

The EPFL platform proved capable of retrieving spectral data from a single image without the use of spectrometers, potentially a significant step towards eventual portable diagnostic applications. It also extended the known capabilities of dielectric metasurfaces as tools to analyse two-dimensional layers of biological entities at atomic-layer thicknesses, according to the team.

A modified version of the same platform in which the metasurfaces are designed to resonate at different wavelengths in different areas is also being investigated —a potentially simpler technique, but less precise in locating the molecules.

"Optical sensors could play a major role in addressing future challenges, particularly in personalized medicine," commented Hatice Altug, head of the EFPL BIOS lab.