Artificial vision for automated monitoring of coronavirus development in high risk environments

Smart hyperspectral cameras that reveal details invisible to the human eye, will be used in a new research project for monitoring Covid-19 infectiousness and support preventing transmission events in hospitals.

Sist oppdatert: May 7, 2020
Published May 7, 2020
02 D6 D7 B6 91 D0 454 C A767 24 B385 E0595 F

The NORCE-led project aims to develop monitoring capabilities of COVID-19 in high-risk environments e.g. hospitals.

Preventing virus spread in such environments is extremely important, to maintain a secure workplace and not least to safeguard the patient’s health.

The project has now received a 4.15 million grant from the Research Council of Norway’s Covid-19 emergency call.

International institutes and Norwegian partners

New analytical tools, artificial intelligence, and machine learning will be used, taking humans from the loop, to automatically dig into large spectral bands for fast mapping of descriptors and pattern detection and recognition of infected areas.

NORCE collaborates with international pioneering institutes from Europe and USA in the field and also including the Norwegian partners such as the University of Bergen, Haukeland University Hospital, and NEO.

NEO is a research and development company with specialized expertise in photonics.

Hyperspectral cameras

Unlike conventional RGB cameras that mostly "see" colors of the visual light in three bands (red, green and blue) similar to the human eye - red green and blue, hyperspectral cameras can divide the spectrum into few hundreds of bands.

This means fine wavelength resolution within and beyond the visual range. The technique is very useful for image mining where the spectrum for each pixel in the image can be analyzed.

The purpose is to find descriptors of material properties and changes that can uncover knowledge on the Covid-19 virus and give indicators about contaminated areas.

Looking for certain chemical compounds

- It is important to stress that this research is not intended to detect SARS-CoV-2 viruses since they are small on a nanometer scale. Detection of viral infection is a prohibitive task at the time, and this research is not intended for disease diagnostics. However, spectroscopy has shown success is the detection of compounds and analysis of chemical properties,

COVID-19 virus transmission over areas and surfaces is not completely known, so epidemiological transmission events cannot always be traced back. And for monitoring high-risk areas, there is currently no available technologies.

The researchers work on establishing a baseline sensor system that has the potential to support the imaging of the chemical properties of a room in a hospital by using a large number of spectral bands.

COVID-19 infectiousness and transmission events

- We are looking forward to this initiative on identifying spectra that can provide a better understanding of the COVID-19 epidemiological infectiousness and transmission events.

In that case, such substances can be visualized by the hyperspectral camera and captured as something distinctive by an advanced analysis tool, says Belbachir at NORCE.

Belbachir heads NORCE’s research group focusing on Smart Instrumentation and Industrial Testing.

He underscores that one of the major challenges for the researchers in this project is to find descriptors that characterize the chemical substances associated with the presence of the Covid-19 virus.

- This is an exciting project and an important challenge. The overall goal of this initiative is to establish foundational knowledge on monitoring of epidemiological transmission and infectiousness in a high-risk environment such as healthcare organizations.

- Recognizing visual descriptors that characterize infected areas with the presence of viruses will be further investigated by partners having medical expertise within virology and microbiology, Belbachir says.