A mid-IR spectroscopic setup for identifying viruses in human saliva under high throughput conditions

Kurzfassung

Mid-IR-Spectroscopy promises high selective capabilities for the detection of viruses such as SARS-CoV2 in human saliva. The motivation for this work was driven by the demand for enhanced COVID testing capabilities, especially in scenarios with large number of people passing, e.g. at airports. To achieve this, high and fast throughput of taken saliva samples is required. Additionally, high true positive rates are essential, when identifying the viral signatures e.g. by machine learning techniques. In this progress report, we present our approach for a high and fast throughput detection setup: The mid-IR spectroscopy part is based on quantum cascade lasers (QCL) and attenuated total reflection flow cells. To enrich the concentration of the viral load in the saliva sample, we use ctionalized magnetic beads. The sample handling part is designed to be automatized for the later application. This can allow for a much faster spectral scan than compared to for example FTIR methods and a possible higher throughput of samples, as for example for Raman microscopy approaches. In our setting, detecting low concentrations of viral loads in aqueous solutions is challenging due to the dynamic range required for the detectors. We present a balanced detection approach to achieve a reasonable dynamic range and discuss the implications for the automatized identification. The diversion of sample material during the cleaning procedure of the flow cells, the purification process with the protein-coated magnetic beads and other impurities can cause variations in the amide I and amide II vibration bands, disturbing the virus signatures. Along our measured spectra we discuss these upcoming challenges for the AI-based reliable identification of the viruses with respect to the creation of training datasets.