SLIM technique assists Covid-19 detection

14 Sep 2021

Beckman Institute and partners employ spatial light-interference microscopy to spot virus.

A team including the Beckman Laser Institute and partners from the University of Illinois has developed an optical method for identifying Covid-19 viruses rapidly and without additional labeling.

Published in Light Science & Applications, the technique pairs spatial light-interference microscopy (SLIM) with bespoke deep-learning algorithms optimized for detection and classification of viral particles.

SLIM, developed jointly by the Institute and Illinois universities in 2011, combines two methodologies: phase contrast microscopy, which converts phase shifts in light passing through a specimen into brightness changes; and holography, as the means of recording the phase information generated.

The technique's developers claimed that SLIM "reveals the intrinsic contrast of cell structures and renders quantitative optical path-length maps across the sample. The resulting topographic accuracy is comparable to that of atomic force microscopy."

In the current project's pre-clinical experiment, viruses deposited on a glass slide were imaged using SLIM. Conventional fluorescence imaging of the same fields-of-view provided "ground truth" data for subsequent computational analysis by algorithms able to identify the viruses.

"Although individual viruses are below the diffraction limit of the microscope, the optical path-length information retrieved by SLIM unravels the nanoscale distribution of the refractive index associated with the individual and aggregated viral particles," commented the team.

This phase information can be used to extract specific biophysical information, in particular the parameter of dry mass density that can effectively act as a marker for different types of virus and which the algorithms can assess from the data.

"There two main components in our model," said the project team. "First the dry mass density can report on the differences in the refractive index caused by the protein compositions of the virus. And second, the nanostructure signature of individual viruses and their surface structures are subtle features within the SLIM images, exploited by the neural network."

Fast detection of viruses is needed

To make the test as effective as possible, and also assess the specificity of its neural network analysis, the project applied its technique to samples of SARS-CoV-2 alongside H1N1 virus, a scenario which might well occur in eventual real-world uses of the technique.

An additional pair of viruses, human adenovirus (HAdV) and zika virus (ZIKV), were also introduced, by digitally mixing the different species into the same SLIM image for deep-learning development.

According to the project, the results indicated "a 96 percent success rate for SARS-CoV-2 detection and classification," along with 99 percent for H1N1, 92 percent for HAdV, and 91 percent for ZIKV, collectively representing "a notable success."

The team envisages a clinical use of the technique to analyze samples collected when a patient has exhaled on a glass slide, with the SLIM platform then rapidly distinguishing the viruses from other bacteria, phospholipids and aerosols present. According to Neha Goswami of the Beckman Laser Institute, the key advantages of such a Covid test would be its speed, with a test duration of around one minute; and the absence of any added chemicals or other modifications to the samples.

"We need fast detection of diseases, not only Covid but others," Goswami said. "We can and should put our efforts together, both in terms of optics and AI, to try and find out just how far we can go."