Tissue Surface-enhanced Raman Spectroscopy Combined with Machine Learning Algorithms for Rapid Detection of Alveolar Echinococcosis

• Published in: 2024 IEEE 25th China Conference on System Simulation Technology and its Application (CCSSTA) pp. 194 - 197
• Main Authors: Huang, Jiahui, Zheng, Xiangxiang, Lv, Guodong, Li, Xiaojing, Wu, Guohua, Xu, Liang
• Format: Conference Proceeding
• Language: English
• Published: IEEE 21.07.2024
• Subjects: Accuracy; alveolar echinococcosis; Analytical models; Biological system modeling; Interference; Kernel; Machine learning; Machine learning algorithms; Raman scattering; rapid diagnosis; Support vector machines; Surface resistance; surface-enhanced Raman scattering; tissue samples
• DOI: 10.1109/CCSSTA62096.2024.10691877

Alveolar echinococcosis (AE) remains a serious worldwide public health problem as a very harmful human tapeworm disease. Rapid and accurate diagnosis of AE is crucial for treatment and postoperative recovery. This study explored the applicability of tissue surface Raman spectroscopy (SERS) combined with machine learning for rapid AE detection. Specifically, liver tissue section samples were obtained by creating an AE mouse model and SERS spectra were acquired. Five kinds of machine learning methods were used to construct AE diagnosis model, among which the linear kernel support vector machine (SVM) model had the best classification result, and the accuracy rate reached 95.7 \%. Then, the linear SVM model was subjected to feature significance analysis, which demonstrated that the biochemicals corresponding to the spectra of 641,716 and 1372 \mathrm{~cm}^{-1} might be potential biomarkers. Gaussian noise is further added to test the noise interference resistance of different machine learning algorithms, and it is found that SVM with Gaussian radial basis function (RBF) kernel function has better noise interference resistance. The results show that tissue SERS spectroscopy combined with SVM algorithm for rapid detection of AE has a promising application.