ASABE Technical Library - Abstract

Click on “Download PDF” for the PDF version or on the title for the HTML version.
If you are not an ASABE member or if your employer has not arranged for access to the full-text, view access and subscription options

Hyperspectral Imaging–Based Breath Analysis for Detection of Bovine Coronavirus Infection in Beef Cattle

Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan www.asabe.org

Citation:  2026 ASABE Annual International Meeting  2600991.(doi:10.13031/aim.202600991)
Authors:   Safal Kshetri, Steve Wakham, Hao Gan, Yanqiu Yang
Keywords:   Bovine coronavirus (BCoV); Disease detection; Exhaled breath analysis; Hyperspectral imaging; Machine learning; Precision livestock farming; Volatile organic compounds (VOCs).

Abstract. Bovine coronavirus (BCoV) is a globally prevalent pathogen in cattle that causes enteric and respiratory disease, negatively affecting animal health, productivity, and economic sustainability. Current diagnostic methods rely on laboratory-based assays requiring biological sample collection, which, while accurate, are invasive, labor-intensive, and impractical for continuous and large-scale herd monitoring. In this study, the potential of hyperspectral imaging (HSI) was investigated as a non-invasive sensing approach for differentiating BCoV-infected and healthy beef cattle based on spectral information. Hyperspectral data were collected from 14 beef cattle with a sensor suite designed to capture animal breath samples. Clean (ambient) air was also measured concurrently as a control group (CA) for the analysis. Individual cattle was restrained in chutes with headgates during the data collection process. Nasal samples were collected and analyzed using PCR (Polymerase Chain Reaction) tests to classify animals as BCoV-positive and healthy reference individuals, providing ground-truth labels for analysis. Multivariate statistical and machine learning–based hyperspectral analyses were performed to evaluate the ability of HSI data to classify the infected and healthy animals. The KNN classifier achieved a validation accuracy of 60.7%, while a quadratic support vector machine (SVM) classifier achieved an improved validation accuracy of 67.3% after temporal drift correction. The study and results demonstrate the potential of hyperspectral imaging as a non-invasive tool for early and automated disease screening in beef production systems, supporting the development of precision livestock technologies to improve animal welfare and reduce dependence on labor-intensive diagnostic practices.

(Download PDF)    (Export to EndNotes)