Characterization of Bacteria Inducing Chronic Sinusitis Using Surface-Enhanced Raman Spectroscopy (SERS) with Multivariate Data Analysis

• Published in: Analytical letters 2023-05, Vol.56 (8), p.1351-1365
• Main Authors: Bari, Rana Zaki Abdul, Nawaz, Haq, Majeed, Muhammad Irfan, Rashid, Nosheen, Tahir, Muhammad, ul Hasan, Hafiz Mahmood, Ishtiaq, Shazra, Sadaf, Nimra, Raza, Ali, Zulfiqar, Anam, Rehman, Aziz ur, Shahid, Muhammad
• Format: Article
• Language: English
• Subjects: Bacteria; Bacterial strains; Classification; Data analysis; Differentiation; Discriminant analysis; Klebsiella; Multivariate analysis; Paranasal sinuses; partial least squares-discriminate analysis (PLS-DA); principal component analysis (PCA); Principal components analysis; Raman spectroscopy; Sinuses; Sinusitis; Spectrum analysis; surface-enhanced Raman spectroscopy (SERS)
• ISSN: 0003-2719; 1532-236X
• DOI: 10.1080/00032719.2022.2130349

Sinusitis is the inflammation of the mucous membrane lining the paranasal sinuses, and if symptoms and signs of sinusitis last for more than 12 weeks, it is categorized to be chronic. In this work, the characterization of cell mass/pellets of three bacterial strains, Klebsiella pneumoniae, Enterococcus faecalis, and Staphylococcus aureus, which cause chronic sinusitis, was performed by surface-enhanced Raman Spectroscopy (SERS). These bacteria that induce chronic sinusitis were cultured and isolated from the nasal swab of a patient and identified by the 16S rRNA sequences performed on isolated strains. The bacteria were characterized by their SERS characteristics, showing the potential of this method. SERS features at 594, 822, 831, 944, 1030, 1170, and 1268 cm −1 were the differentiating features of these bacteria. Moreover, multivariate data analysis was performed by principal component analysis (PCA) and partial least squares-discriminate analysis (PLS-DA) and shown to be suitable for the differentiation and classification of these bacteria. The spectral features were characterized by PCA for classification. PLS-DA was applied for further validation of differentiation which provides accuracy and sensitivity above 90% in all of the models. The area under curve (AUC) was near 1 for all PLS-DA models.