A novel approach to smart-assisted schizophrenia screening based on Raman spectroscopy and deep learning

• Published in: Scientific reports Vol. 15; no. 1; pp. 28487 - 11
• Main Authors: Xiao, Meng, Xiaokaiti, Sulidan, Shang, Meng, Xu, Pan, Zhu, Xiaofen
• Format: Journal Article
• Language: English
• Published: London Nature Publishing Group UK 05.08.2025; Nature Publishing Group; Nature Portfolio
• Subjects: 631/114; 631/443; 631/45; 639/624; 692/308; 692/617; 692/699; Adult; Biomarkers; Body fluids; Carotenoids; Deep Learning; Disease; Female; Health care; Humanities and Social Sciences; Humans; Lipids; Male; Markov Chains; Medical diagnose; Medical screening; Mental disorders; Methods; multidisciplinary; Neural networks; Neural Networks, Computer; Proteins; Raman spectroscopy; Schizophrenia; Schizophrenia - blood; Schizophrenia - diagnosis; Science; Science (multidisciplinary); Signal processing; Spectral analysis; Spectrum analysis; Spectrum Analysis, Raman - methods; Time series; Triglycerides; Deep learning; Schizophrenia; Medical diagnose; Raman spectroscopy
• ISSN: 2045-2322; 2045-2322
• DOI: 10.1038/s41598-025-14015-x

In this study, serum Raman spectra are introduced into the screening of schizophrenia. We collect serum Raman spectra from schizophrenic and healthy individuals, classified them based on four convolutional neural networks, and developed an assisted screening method for schizophrenia based on serum Raman spectra. We also introduce Markov transition field (MTF), which is commonly used in time-series signal processing, into Raman spectral analysis, and convert 1D Raman spectral sequences into 2D spectrograms to enrich the method of Raman spectral analysis. The experimental results show that the performance of the model trained based on MTF is overall better than that of the model trained based on 1D spectral sequences.