Sector(s): Electronics, Sensors & Photonics, Aerospace, Aviation & Transport, Chemical, Healthcare & Pharmaceuticals, Other
About Opportunity:
Implementation of a Raman spectral decomposition technique that allows effective identification of complex mixtures. The computationally and memory efficient software enables new functionality to be added to portable hand-held devices.
The Challenge
Raman spectroscopy is an established method for identifying unknown materials across various sectors. Conventional analysis methods are based on comparing the measured spectrum with a reference spectral library of known chemicals to find the best match. While effective for identifying a single spectrum from a library, a sample composed of a mixture of different chemicals provides a greater challenge.
Technology
Edinburgh researchers have developed a Raman spectral decomposition technique based on a new fast sparse approximation method. Inputting a set of reference spectra and an unknown mixture yields the identity of mixture elements and their contribution percentages. It also has the capability of detecting cases where the mixture has a spectrum outside the reference library. The method is highly computationally and memory efficient, which means that it can run on a low power real-time platform. Implemented as a hardware independent C package, which can handle a given library and input spectrum, the technology enables use with hand-held devices. This provides a portable, non-invasive approach for identification of real-life mixtures of chemical substances.
Exemplification Data
A hardware independent C version of the mixture-matching algorithm has been prepared. Performance has been successfully demonstrated in the identification of real mixtures in different measurement scenarios, including where components are close to noise level.
Development Status: Prototype Development
IP Status:
Software package
Publications: A Sparse Regularized Model for Raman Spectral Analysis, Wu et al, Sensor Signal Processing for Defence, Edinburgh, 2014
The University of Edinburgh is seeking potential industry partners to license this technology.