AI-enabled realtime dual-comb molecular fingerprint imaging

Dual-laser frequency combs and artificial intelligence enable hyperspectral imaging of molecular absorption signatures in realtime. This novel technique may open new opportunities for environmental monitoring, process control and bio-imaging.

Hyperspectral images contain detailed spectral information for each pixel of an image. The spectral resolution is often limited by dispersive optical elements or filters. To overcome this limitation, we use dual-laser frequency combs as active illumination sources and an infrared detector array to perform hyperspectral imaging with ultra-high spectral resolution. This approach can resolve the fine absorption features of gas molecules. Although large amounts of raw data are generated, artificial intelligence (similar to what is used in image recognition) can analyze and interpret the recorded data in realtime. Providing molecule specific images of gases, this new technique may prove useful in environmental and industrial monitoring as well as bio-imaging applications.

Hyperspectral_3.png
Select frames from a movie of a molecular gas jet recorded and processed with a framerate 10 Hz.
 

Reference:

AI-enabled real-time dual-comb molecular fingerprint imaging; Thibault Voumard, Thibault Wildi, Victor Brasch, Raúl Gutiérrez Álvarez, Germán Vergara Ogando, Tobias Herr; Optics Letters, 2020; DOI: 10.1364/OL.410762