Machine Learning-Based Heavy Metal Ion Detection Using Surface-Enhanced Raman Spectroscopy

Abstract

Surface-Enhanced Raman Spectroscopy (SERS) is often used for heavy metal ion detection. However, large variations in signal strength, spectral profile, and nonlinearity of measurements often cause problems that produce varying results. It raises concerns about the reproducibility of the results. Consequently, the manual classification of the SERS spectrum requires carefully controlled experimentation that further hinders the large-scale adaptation. Recent advances in machine learning offer decent opportunities to address these issues. However, well-documented procedures for model development and evaluation, as well as benchmark datasets, are missing. Towards this end, we provide the SERS spectral benchmark dataset of lead(II) nitride (Pb(NO3)2) for a heavy metal ion detection task and evaluate the classification performance of several machine learning models. We also perform a comparative study to find the best combination between the preprocessing methods and the machine learning models. The proposed model can successfully identify the Pb(NO3)2 molecule from SERS measurements of independent test experiments. In particular, the proposed model shows an 84.6% balanced accuracy for the cross-batch testing task.

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

“…Based on the current reported work, the benefits of machine learning for SERS and SEIRA can be summarized as follows. 45,112–125…”

“…Based on the current reported work, the benefits of machine learning for SERS and SEIRA can be summarized as follows. 45,[112][113][114][115][116][117][118][119][120][121][122][123][124][125] Machine learning algorithms enable the automated design of SERS/SEIRA substrates to avoid time-consuming and onerous design processes. Taking SEIRA's antenna design as an example, 126 its first step is to analyze the infrared spectrum of the analyte molecule and obtain the position of the molecular vibration.…”

Self-supervised learning for inter-laboratory variation minimization in surface-enhanced Raman scattering spectroscopy

“…Similar to other popular preprocessing techniques, e.g., baseline correction and power-spectral normalization, it is possible to combine MVNet with neural network-based regression models. 18,19 In such an end-to-end deep learning framework, one can predict the concentration of an unseen target analyte without any preprocessing. This is a particularly attractive feature for solving the reproducibility issues in quantitative SERS measurements.…”

“…[10][11][12][13][14][15][16][17] Although these studies have mainly focused on building application-specific predictive models, some studies have suggested methods to improve the consistency of predictive performance under independent test conditions. 18,19 In addition, Liu et al 20 proposed a convolutional neural network (CNN)-based unified solution for Raman signal analysis. Park et al 21 also proposed a pseudosiamese network to identify the best matching signal from reference spectra.…”

Optical imaging spectroscopy coupled with machine learning for detecting heavy metal of plants: A review

“…Although these methods have high detection accuracy, they are time-consuming, laborious, complicated sample pretreatment and have a limited detection range, which cannot meet the requirements of green, rapid, and large-scale detection ( Bian et al., 2020 ; Shojaei et al., 2021 ; Zhao et al., 2021 ). Recently, with the development of optical instruments and machine learning, the sensitivity of optical imaging spectroscopy has been improved and gradually applied to the detection of heavy metals for plants because of the simple, rapid, and in situ advantages provided by spectroscopy ( Zhou et al., 2020a ; Park et al., 2022 ).…”