Covid-19: Cytometry and pedestrian detection assist pandemic response

12 May 2021

Photonics-based improvements in diagnosis and social distance monitoring offer new tools to combat the disease.

An analysis using a commercial cytometry platform has produced new insights into the different responses caused by Covid-19 and non-SARS-CoV-2 pneumonia in patients, a key clinical distinction.

A team at the University of Zurich's Institute of Experimental Immunology investigated whether the numerous alterations in a patient's immune response could be classified into those specific to Covid-19 and those driven by general inflammatory response common to pneumonia patients. The results were published in Immunity.

The team used a Cytek Aurora fluorescence cytometry platform and algorithmic analysis to analyze blood samples from mild and severe Covid-19 patients alongside a cohort of pneumonia patients, looking for immune signatures specific to Covid-19 and those shared between the two groups.

The Cytek Aurora device can employ up to five lasers in its spectral flow cytometry operation, in which a laser's interaction with controlled numbers of cells can be used as a spectral fingerprint for different cell types or components. Ideally, the operation aims to record data from a single cell at a time, for maximum accuracy.

The potential versatility of the principle is reflected in the growth of the cytometry sector, long predicted to be a significant market for biophotonics technology.

The Zurich group found that both patient cohorts showed increased emergency generation of immune cells from their progenitor cells in the body, along with adaptive immune paralysis, a term used when the immune system loses potency under sustained attack. However, only the Covid-19 patients also showed immune signatures suggestive of T cell exhaustion, a scenario in which that important class of white cell shows a potential range of reduced functionality and lack of response.

The project also used the same experimental platform to determine that the number of a particular form of T cells termed NKT cells could be a predictive biomarker for patient outcomes, with NKT frequency offering a way to distinguish between severe Covid-19 patients and those with a milder case of the disease.

As well as indicating a way to delineate treatment strategies in the clinic, these results also prove the value of spectral flow cytometry in efforts to tackle the pandemic or similar future events.

"It displays the power of high dimensional flow cytometry for making discoveries within huge complexity, from which single pieces of information like NKT percentage can be extracted to inform treatment strategies or patient outcomes," commented Zurich-based immunologist Andrew Croxford on twitter.

Autonomous vehicle technology helps people keep their distance

A project at Swiss research institute EPFL has developed an algorithm able to analyze images of groups of people and determine whether individuals are maintaining appropriate social distancing, without collecting any personal data.

The technique is an evolution of algorithms intended for use with autonomous self-driving vehicles, that the team was working on in 2020 when Switzerland entered lockdown.

"We quickly saw that by adding just a few features, we could make our program a useful tool for managing the pandemic," said Lorenzo Bertoni from EPFL's Visual Intelligence for Transportation Laboratory (VITA).

The project builds on MonoLoco, a computational framework developed at EPFL to "tackle the fundamentally ill-posed problem of 3D human localization from monocular RGB images," using the silhouettes of individual people to determine their distance apart.

Using monocular rather than stereo images or multiple cameras is one potential route to simplifying perception systems for vehicles and other applications, and MonoLoco is ultimately intended to be suitable for fixing to any camera or video recorder.

According to the project's findings, published by IEEE, the platform uses an off-the-shelf pose analysis system to identify the silhouetted postures of human bodies in an image, producing a map of 2D "keypoints" and their planar spatial relationship.

These are fed into a bespoke neural network to predict where these keypoints lie in 3D orientations and locations, with appropriate confidence intervals, and do so without capturing facial images or other personal data.

The project is now planning an initial deployment of its new system on Swiss postal buses through a joint project with Swiss Post, to see if it assists effective social distancing while maximizing movement of individuals.

"To monitor social distancing effectively, we should go beyond a measure of distance," commented the project. "Orientations and relative positions of people strongly influence the risk of contagion."