Multispectral fibre-optic reflectometer with extended UV-VIS-NIR LEDs for non-destructive classification of eggshell optical signatures associated with 𝑀𝑦𝑐𝑜𝑝𝑙𝑎𝑠𝑚𝑎 𝑠𝑦𝑛𝑜𝑣𝑖𝑎𝑒

Abstract

This paper presents the development and evaluation of the multispectral fibre-optic reflectometer expanded to the UV-VIS-NIR spectral range for the non-destructive classification of eggshell originating from healthy or Mycoplasma synoviae (MS) infected poultry. MS is a highly contagious pathogen causing significant economic losses, particularly by altering the structure and reflectance of eggshells. The multispectral fibre-optic reflectometer (MFOR) system evaluates back-reflected light using machine learning algorithms to accurately classify eggshells as originating from healthy or MS infected hens. In this study, four categories of eggshells (brown/white, infected/healthy) were tested using different LED combinations across an extended wavelength spectrum. The machine learning classification model achieved exceptional precision, identifying the optimal wavelength combinations as 340, 490, and 660 nm for white eggs (mean F-score = 0.998) and 340, 490, and 560 nm for brown eggs (mean F-score = 1.000). These findings suggest that incorporating UV wavelengths enhances the MFOR ability to rapidly and non-invasively diagnose MS in a flock based on eggshell optical properties.