Detection of Coronavirus in Viral Transport Media using Ultraviolet and Near-Infrared Absorbance Spectra and Pattern Recognition Model

Abstract

Abstract SARS-CoV-2 causes individuals to become infected with respiratory disease known as COVID-19. Rapid and robust identification ensures that the infected patients can be quarantined. In this paper, the detection of SARS-CoV-2 utilizes ultraviolet (UV) and near-infrared (NIR) absorbance spectra, along with principal component analysis and linear discriminant analysis (PCA-LDA). A total of 75 negative and 75 positive swab samples are separately placed in vials of viral transport media and transferred into cuvettes. The absorbance spectra are acquired and processed before they undergo dimensionality reduction using PCA. The dataset is divided into training set and testing set to develop and evaluate the PCA-LDA model. The scree plot analysis reveals that the two principal components are optimal for both UV and IR absorbance spectra. By utilizing the first two principal components, the performance indicators demonstrate higher accuracy (97.00%), sensitivity (94.84%), and specificity (99.31%) on IR absorbance spectra. This is attributed to the overall difference in IR absorbance, as well as two peaks centred at 558.5 nm and 972 nm respectively. Utilizing IR absorbance spectra with PCA-LDA model is cost-effective while showing performance comparable to conventional methods such as polymerase chain reaction. This method provides an alternative for rapid and effective SARS-CoV-2 detection.