文献类型：期刊文献

英文题名：Deep Learning-Based Spectral Extraction for Improving thePerformance of Surface-Enhanced Raman Spectroscopy Analysis onMultiplexed Identification and Quantitation

作者：Zhang, Jie[1];Xin, Pei-Lin[1];Wang, Xiao-Yuan[1];Chen, Hua-Ying[1];Li, Da-Wei[1]

机构：[1]East China Univ Sci & Technol, Frontiers Sci Ctr Materiobiol & Dynam Chem, Sch Chem & Mol Engn, Key Lab Adv Mat,Shanghai Key Lab Funct Mat Chem, Shanghai 200237, Peoples R China

年份：2022

卷号：126

期号：14

起止页码：2278

外文期刊名：JOURNAL OF PHYSICAL CHEMISTRY A

收录：;EI(收录号：20221712033260);WOS:【SCI-EXPANDED(收录号:WOS:000791468200013)】；

基金：The authors appreciate financial support from the National Natural Science Foundation of China (21788102, 21777041, 21974046, 22176058, and 21977031), the Science and Technology Commission of Shanghai Municipality (19391901700, 19520744000, and 19ZR1472300), and the Fundamental Research Funds for the Central Universities.

语种：英文

外文关键词：Neural network models - Sulfur compounds - Mixtures - Light transmission - Deep learning - Raman spectroscopy

摘要：Surface-enhanced Raman spectroscopy (SERS) hasbeen recognized as a promising analytical technique for itscapability of providing molecularfingerprint information andavoiding interference of water. Nevertheless, direct SERS detectionof complicated samples without pretreatment to achieve the high-efficiency identification and quantitation in a multiplexed way isstill a challenge. In this study, a novel spectral extraction neuralnetwork (SENN) model was proposed for synchronous SERSdetection of each component in mixed solutions using ademonstration sample containing diquat dibromide (DDM),methyl viologen dichloride (MVD), and tetramethylthiuramdisulfide (TMTD). A SERS spectra dataset including 3600 spectraof DDM, MVD, TMTD, and their mixtures wasfirst constructedto train the SENN model. After the training step, the cosine similarity of the SENN model can achieve 0.999, 0.997, and 0.994 forDDM, MVD, and TMTD, respectively, which means that the spectra extracted from the mixture are highly consistent with thosecollected by the SERS experiment of the corresponding pure samples. Furthermore, a convolutional neural network model forquantitative analysis is combined with the SENN, which can simultaneously and rapidly realize the qualitative and quantitative SERSanalysis of mixture solutions with lower than 8.8% relative standard deviation. The result demonstrates that the proposed strategyhas great potential in improving SERS analysis in environmental monitoring, food safety, and so on