Spectroscopic Characterization of Bacterial Colonies through UV Hyperspectral Imaging Techniques

Josune J Ezenarro ¹ Mohammad Al Ktash ^2* Nuria Vigues ^1* Jordi Mas Gordi ^1* Xavier Munoz-Berbel ^3,4 Marc Brecht ^5*

• ¹ Departament Genètica i Microbiologia, Universitat Autònoma de Barcelona, 08193 Bellaterra, Barcelona, Spain
• ² Process Analysis and Technology PA & T, Reutlingen University, Alteburgstraße 150, Reutlingen, Germany
• ³ Departament Genètica i Microbiologia, Universitat Autònoma de Barcelona, 08193 Bellaterra (Barcelona), Spain, Barcelona, Balearic Islands, Spain
• ⁴ CIBER de Bioingeniería, Biomateriales y Nanomedicina, Instituto de Salud Carlos III, E-28029, Madrid, Spain
• ⁵ Process Analysis and Technology PA & T, Reutlingen University, Alteburgstraße 150, 72762 Reutlingen, Germany, Reutlingen, Baden-Württemberg, Germany

Although molecular biology techniques are now gaining much attention, plate culturing and visual inspection is still the gold standard method for bacterial identification. Colony identification with agar plates is manual, interpretative, strongly dependent on human experience and subjected to human errors. Advanced imaging techniques, such as hyperspectral imaging, have emerged as promising alternatives. The VIS-NIR region has been previously utilized for colony identification; however, its routine application can be challenging due to its sensitivity to multiple components and functional groups, including changes in the culture medium itself, which could interfere with measurements and lead to inaccurate results. However, its application in the ultraviolet (UV) region has been unexplored, despite the presence of specific absorption and emission peaks in many bacterial components. To address this gap, we present a predictive model for bacterial colony detection and identification using UV hyperspectral imaging. The model leverages hyperspectral images acquired in the UV wavelength range of 225-400 nm, processed with principal component analysis (PCA) and discriminant analysis (DA). This measurement setup integrates a hyperspectral imager, a PC for automated data analysis, and a conveyor belt system for transporting agar plates to the detection region for their automated analysis. Four bacterial species-Escherichia coli, Staphylococcus, Pseudomonas, and Shewanellawere analyzed on two generic culture media, Luria Bertani and Tryptic Soy, to train and validate the model. The PCA-DA-based model achieved a high accuracy of 90% in differentiating bacterial species based on the first three principal components. These results highlight the untapped potential of UV hyperspectral imaging combined with advanced data analysis tools, offering a robust and automated alternative to traditional, human-dependent methods for bacterial identification.