Pathogenic bacteria infections are a major public health problem in current society. Rapid and reliable identification of these pathogens can help avoid the misuse of antibiotics and enable precision therapy. In this study, we present a large-spot confocal Raman system based on fiber array (LSCR-FA) for the in situ detection of microbial colonies on agar plates. This method can alleviate the problem of spatial heterogeneity of colonies to a certain extent and is fast and high-throughput. Additionally, we also applied machine learning algorithms with 5-fold cross-validation to analyze colony Raman spectral data and classify seven different pathogenic bacteria. Among them, the Support Vector Machine (SVM) achieved a high accuracy of 98.74 %. The results of the study demonstrate that the mentioned LSCR-FA system combined with machine learning algorithms provides a new, fast, and effective strategy for the identification of pathogenic bacteria and precise clinical treatment.
Keywords: Bacterial colony; Confocal Raman; In-situ detection; Large spot; Round-to-linear fiber array.
Copyright © 2024 Elsevier B.V. All rights reserved.