Discrimination of periodontal pathogens using Raman spectroscopy combined with machine learning algorithms

Abstract

Periodontitis is closely related to many systemic diseases linked by different periodontal pathogens. To unravel the relationship between periodontitis and systemic diseases, it is very important to correctly discriminate major periodontal pathogens. To realize convenient, efficient, and high-accuracy bacterial species classification, the authors use Raman spectroscopy combined with machine learning algorithms to distinguish three major periodontal pathogens Porphyromonas gingivalis (Pg), Fusobacterium nucleatum (Fn), and Aggregatibacter actinomycetemcomitans (Aa). The result shows that this novel method can successfully discriminate the three above-mentioned periodontal pathogens. Moreover, the classification accuracies for the three categories of the original data were 94.7% at the sample level and 93.9% at the spectrum level by the machine learning algorithm extra trees. This study provides a fast, simple, and accurate method which is very beneficial to differentiate periodontal pathogens.