2025

Paiva, Eduardo Maia; Hyttinen, Eevi; Dönsberg, Timo; Barth, Dorothee

<p>This study elucidates the utility and efficacy of UV–visible spectroscopy for the detection and characterization of biological contaminants within microalgae cultures, augmented by machine learning algorithms. Biological contamination, exemplified by flagellates and rotifers, poses a significant concern due to its potential to rapidly devastate entire cultures, thus jeopardizing commercial viability. Conventional analytical methods for monitoring contamination, such as microscopy and cytometry, are often labor-intensive, reliant on specialized expertise for microorganism identification, and may lack specificity in discerning the nature of contamination, impeding timely intervention. UV–visible spectroscopy offers a compelling solution by overcoming many of these challenges, affording specificity in analysis, real-time monitoring capabilities, and automation, owing to the intricate pigment chemistry inherent in the microalgae realm, which generates distinct UV–visible spectra. Through the measuring of contaminated and uncontaminated samples, coupled with machine learning analysis of their respective spectra, this study explores the underlying biochemical principles driving spectral data, thereby justifying the efficacy of the technique. The findings underscore the wealth of information encapsulated within UV–visible spectral data, which can be effectively harnessed through classification algorithms for early-stage identification of contamination in real-time applications.</p>