TY - JOUR
T1 - Blood detection of autoimmune encephalitis based on laser-induced breakdown spectroscopy and Raman spectroscopy
AU - Zhao, Zhifang
AU - Xu, Wangshu
AU - Teng, Geer
AU - Xu, Xiangjun
AU - Lu, Bingheng
AU - Zhou, Hao
AU - Wang, Leifu
AU - Liu, Yuge
AU - Xu, Shuai
AU - Wang, Qianqian
AU - Ma, Wenping
N1 - Publisher Copyright:
© Elsevier B.V.
PY - 2025/6/1
Y1 - 2025/6/1
N2 - Background: Recently, the incidence range of autoimmune encephalitis (AE) in people has rapidly expanded, and the diagnosis procedure of clinical criteria for AE remains complicated. Herein, with advantages of rapid speed, simple pre-treatment, and slightly destructive or non-destructive analysis, the feasibility of integrating laser-induced breakdown spectroscopy (LIBS) and Raman techniques to identify blood of AE patients was explored, and the mechanisms of medical diagnosis from atomic and molecular perspectives were further analyzed. Results: In the experiment, etched mesh silicon wafers were used as serum substrates to reduce the spectral variability during measurements. Totally, 1785 LIBS spectra and 1785 Raman spectra were collected from 119 people (79 healthy people and 40 AE patients), respectively. Fusion spectra were formed by connecting LIBS spectra in series behind with Raman spectra. With mutual information (MI) method, 537 features were selected from fusion spectra, and the accuracy and test time of long short-term memory model using these features were 95.04 % and 0.95 s, an improvement by 14.36 %, 8.03 %, 2.22 % and 0.48 s, 0.08 s, 0.55 s compared to using LIBS spectra, Raman spectra and fusion spectra, respectively. Besides, the correlations between spectra and cytokines were analyzed by the Pearson's correlation coefficient. Both metal atoms such as Na and K and molecules such as tryptophan, deoxyribose and phenylalanine were related to cytokines, corresponding to their MI importance in the AE diagnosis. Significance: We made the first attempt to identify AE blood using fusion of spectral techniques and analyze correlation mechanism between spectra and cytokines. All results demonstrated that it is feasible to accurately identify AE serum by fusing LIBS and Raman techniques, which is expected to effectively assist the clinical diagnosis of AE in the future.
AB - Background: Recently, the incidence range of autoimmune encephalitis (AE) in people has rapidly expanded, and the diagnosis procedure of clinical criteria for AE remains complicated. Herein, with advantages of rapid speed, simple pre-treatment, and slightly destructive or non-destructive analysis, the feasibility of integrating laser-induced breakdown spectroscopy (LIBS) and Raman techniques to identify blood of AE patients was explored, and the mechanisms of medical diagnosis from atomic and molecular perspectives were further analyzed. Results: In the experiment, etched mesh silicon wafers were used as serum substrates to reduce the spectral variability during measurements. Totally, 1785 LIBS spectra and 1785 Raman spectra were collected from 119 people (79 healthy people and 40 AE patients), respectively. Fusion spectra were formed by connecting LIBS spectra in series behind with Raman spectra. With mutual information (MI) method, 537 features were selected from fusion spectra, and the accuracy and test time of long short-term memory model using these features were 95.04 % and 0.95 s, an improvement by 14.36 %, 8.03 %, 2.22 % and 0.48 s, 0.08 s, 0.55 s compared to using LIBS spectra, Raman spectra and fusion spectra, respectively. Besides, the correlations between spectra and cytokines were analyzed by the Pearson's correlation coefficient. Both metal atoms such as Na and K and molecules such as tryptophan, deoxyribose and phenylalanine were related to cytokines, corresponding to their MI importance in the AE diagnosis. Significance: We made the first attempt to identify AE blood using fusion of spectral techniques and analyze correlation mechanism between spectra and cytokines. All results demonstrated that it is feasible to accurately identify AE serum by fusing LIBS and Raman techniques, which is expected to effectively assist the clinical diagnosis of AE in the future.
KW - AE serum
KW - Cytokines
KW - LIBS
KW - Raman
UR - http://www.scopus.com/inward/record.url?scp=105000303867&partnerID=8YFLogxK
U2 - 10.1016/j.aca.2025.343948
DO - 10.1016/j.aca.2025.343948
M3 - Article
C2 - 40221195
AN - SCOPUS:105000303867
SN - 0003-2670
VL - 1353
JO - Analytica Chimica Acta
JF - Analytica Chimica Acta
M1 - 343948
ER -