Combining Dense Au Nanoparticle Layers and 2D Surface-Enhanced Raman Scattering Arrays for the Identification of Mutant Cyanobacteria Using Machine Learning

Abstract

We report the crowding of Au nanoparticles (Au NPs) on a surface-enhanced Raman scattering (SERS) 2D array substrate with high nanoparticle surface coverage in a combined approach for the identification of cyanobacteria with machine learning. By simply using the screening effect of NaCl, the crowding effect of PEG to overcome the repulsion between nanoparticles, and different dithiol chain lengths during the deposition process of Au NPs on a substrate, we provide a general approach to increase the deposition density of nanoparticles on the films over nanodisk-array SERS substrates. The optimized substrate was subsequently utilized for the discrimination of wild-type (WT) and mutant cyanobacteria using SERS and machine learning methods (principal component analysis, logistic model, Gaussian naïve Bayes model, K-nearest-neighbor model, and a support vector classifier model with radial basis function). The best performance to discriminate between WT and mutant cyanobacteria was achieved by using the support vector classifier (SVC) with a positive rate as high as 97% using five repeat tests for the congeneric cells. These results indicate that highly sensitive SERS substrates, in combination with efficient data analysis, can be employed in mutant identification by SERS, enabling high-throughput screening in the current biological research.

Machine learning-augmented surface-enhanced spectroscopy toward next-generation molecular diagnostics

“…The challenge can be addressed by ML-enhanced data processing in SERS. [221][222][223][224] Alstrom and co-workers 225 proposed a Bayesian Non-negative Matrix Factorization (NMF) approach to identify the locations of target molecules. This method can successfully analyze the spectra and extract the target spectrum, as shown in Fig.…”

Recent Trends in SERS-Based Plasmonic Sensors for Disease Diagnostics, Biomolecules Detection, and Machine Learning Techniques

“…The CNN model achieved an accuracy of 96%, which is better than that of PLS-DA, with 84% [ 387 ]. The SVM outperformed other techniques in the identification of cyanobacteria, using SERS spectra of mutant and wild-type strains [ 388 ]. Using a dimensionality reduction technique, followed by a probabilistic ML model, SARS-CoV-2 identification was performed with an accuracy of ~85% [ 389 ].…”

Nanophotonics for Chemical Imaging and Spectroscopy

“…Indeed, this work dissects the contributions of different phenomena to SERS (and ambient TERS) signal fluctuations, which continue to puzzle practitioners, even at this (late) stage of evolution of both techniques. , Equally interesting contributions analyzed ligand binding on silver nanoparticles, , a detailed understanding of which is a prerequisite to rationalizing the enhanced optical spectra and chemical properties of molecules on plasmonic nanostructures. Finally, both conventional and machine-learning-assisted analyses of nanoparticles and nanoparticle–cyanobacteria assemblies were described, with a goal of interfacing advanced theoretical treatments with SERS to significantly broaden the application space of this powerful technique.…”