Summary:
UCLA researchers in the department of Electrical and Computer Engineering have developed a deep learning-based device that uses a scalable plasmonic encoder to spectrally reconstruct a variety of complex mixed data from analytical chemistry to astronomy with high spectral resolution.
Background:
Optical sensors have become paramount in the quantitative fields of analytical chemistry, astronomy, and quality control. The light spectrum that these sensors detect is often heterogeneous, containing information about multiple sources. Though there are methods for separating these spectra, they come at the cost of sensitivity. This impairs signal analysis of distant stars as well as detection of trace amounts of molecules. While the overall size of traditional instruments can be increased to overcome this limitation, this comes at a steep physical cost, requires complex instrument calibration, and virtually eliminates the possibility of developing portable instruments. Portable optical sensors that are capable of high spectral resolution would have plentiful applications, including astronomy, forensic science, and industrial applications. Therefore, there is a need for highly sensitive portable optical sensors that can distinguish the optical spectrum produced from different sources.
Innovation:
Using a scalable plasmonic encoder, UCLA researchers have developed a high-resolution spectral analysis device coupled with a deep learning neural network. This technology was capable of high spectral resolution while preserving the signal to noise ratio found in larger, more expensive detectors. The system identified 96.86% of spectral peaks, with a peak localization error of 0.19 nm, peak height error of 7.60%, and peak bandwidth error of 0.18 nm. The proposed optical sensor provides a cost-effective and compact solution with universal applicability.
Potential Applications:
- Astronomy
- Analytical Chemistry
- Manufacturing
- Law Enforcement/Forensics
- Portable Scientific Equipment
- Spectral unmixing
- Hyperspectral detection
Advantages:
- Cost Effective
- Compact, portable
- No Mechanical Scanning components
- Robust to environmental changes
- Highly Sensitive
- High Spectral Resolution
Development-To-Date
A prototype has been developed and extensively tested to successfully demonstrate high accuracy of the proposed device.
Related Papers:
Goncharov, A.; Brown, C.; Ballard, Z.; Fordham, M.; Clemens, A.; Qiu, Y.; Rivenson, Y.; Ozcan, A. “ Deep learning-based spectral reconstruction on a chip using scalable plasmonic encoder” Optical Society of America, 2021.