Surface-Enhanced Raman Spectroscopy-Based Detection of Micro-RNA Biomarkers for Biomedical Diagnosis Using a Comparative Study of Interpretable Machine Learning Algorithms

Abstract

Surface-enhanced Raman spectroscopy (SERS) has wide diagnostic applications due to narrow spectral features that allow multiplex analysis. We have previously developed a multiplexed, SERS-based nanosensor for micro-RNA (miRNA) detection called the inverse molecular sentinel (iMS). Machine learning (ML) algorithms have been increasingly adopted for spectral analysis due to their ability to discover underlying patterns and relationships within large and complex data sets. However, the high dimensionality of SERS data poses a challenge for traditional ML techniques, which can be prone to overfitting and poor generalization. Non-negative matrix factorization (NMF) reduces the dimensionality of SERS data while preserving information content. In this paper, we compared the performance of ML methods including convolutional neural network (CNN), support vector regression, and extreme gradient boosting combined with and without NMF for spectral unmixing of four-way multiplexed SERS spectra from iMS assays used for miRNA detection. CNN achieved high accuracy in spectral unmixing. Incorporating NMF before CNN drastically decreased memory and training demands without sacrificing model performance on SERS spectral unmixing. Additionally, models were interpreted using gradient class activation maps and partial dependency plots to understand predictions. These models were used to analyze clinical SERS data from single-plexed iMS in RNA extracted from 17 endoscopic tissue biopsies. CNN and CNN-NMF, trained on multiplexed data, performed most accurately with RMSElabel = 0.101 and 9.68 × 10–2, respectively. We demonstrated that CNN-based ML shows great promise in spectral unmixing of multiplexed SERS spectra, and the effect of dimensionality reduction on performance and training speed.

SERS microscopy as a tool for comprehensive biochemical characterization in complex samples

“…131,132 Apart from experiment-specific pre-processing steps, e.g., sorting of data or background correction, 127,133 smoothing, de-noising, truncating the spectra to a range of interest, or normalization can be applied, 130 with variations of the outcome depending on the order in which the different procedures are applied. [134][135][136][137] In order to avoid the propagation of errors and to verify whether the applied method has achieved the intended effect, inspection of spectra after individual processing steps is helpful. While there are studies on the effects of pre-processing on various kinds of spectral data for machine learning applications, 138,139 detailed evaluations of pre-processing for SERS data are relatively rare.…”

“…However, the challenge with the application of ANNs is that they require a very large amount of training data and are often used as a 'black box', i.e., it is not clear how the method arrives at its results. Approaches have been developed to address these challenges, e.g., to increase the amount of training data through data augmentation, 135,163 and visualization of the filters and outputs of neural networks to 'peek into the black box'. 164 For both reasons, it makes sense to also use machine learning methods other than neural networks, that may help to better reveal and analyze particular chemical differences between experiments or samples.…”

Gold–Silver Nanoboxes: A Systematic Investigation of Their SERS Capability and Dual Functions

“…The enhancement of SERS has been proportional to the fourth power of the localized enhanced electromagnetic field, with a SERS enhancement factor of 10 5 to 10 9 or higher in metallic nanoparticles (NPs). , The chemical mechanism created by the charge transfer between the molecules and the metallic surface is proved to contribute less to the SERS enhancement . SERS has emerged as a nondestructive technique renowned for its high photostability and sensitivity for wide-ranging applications across diverse domains, including bioassays, biomedical diagnostics, , environmental catalysis, and detection …”

“…1,2 The chemical mechanism created by the charge transfer between the molecules and the metallic surface is proved to contribute less to the SERS enhancement. 1 SERS has emerged as a nondestructive technique renowned for its high photostability and sensitivity for wide-ranging applications across diverse domains, including bioassays, 6 biomedical diagnostics, 7,8 environmental catalysis, 9 and detection. 10 Various gold and silver nanostructures with different shapes and sizes have expressed valuable properties for potential SERS applications.…”