Akademik Veri Yönetim Sistemi
15th Nanoscience and Nanotechnology Conference, Antalya, Türkiye, 3 - 06 Kasım 2019, ss.97
Rapid and reliable identification of bacteria will decrease infectious diseases and antimicrobial resistance which has become a growing
worldwide problem in recent years. Conventional methods which are used for diagnosis of bacterial infections are time-consuming, required
trained analysts, and include complex sample preparation processes. Surface Enhanced Raman Scattering (SERS) is a powerful technique for
rapid, sensitive and label free detection of biological samples. Metallic nanoparticles which are used in SERS technique enhance localized
electric fields and thus the intensity of Raman scattering light increases significantly. In this study, we introduced the application of SERS
combined with machine learning technique to classify 10 different subtypes of methicillin resistant Staphylococcus aureus (MRSA) species
which are resistant to antibiotic at different levels. SERS spectra were collected using 785 nm laser excitation source, after each strain of
MRSA was treated with Ag nanoparticle colloids to ensure signal enhancement. Raw spectra were preprocessed to remove noises from different
sources by using some smoothing algorithms. Spectral dataset which includes 10 classes of MRSA consists of 92 observations and 1015
features. Principal component analysis (PCA) was applied to dataset in order to reduce feature space dimension. Support vector machine
(SVM) is an elegant and powerful classification algorithm and gives better results than other conventional machine learning techniques. SVM
together with PCA has clearly separated the classes each other with %95.86, %96.6, %99.4 accuracy, sensitivity, and specificity, respectively.
The obtained results are highly promising for the application of Raman spectroscopy combined with machine learning in the determination of
MRSA strains