Application Of Raman Spectroscopy In The Rapid Detection Of Echinococcosis

Echinococcosis is a zoonotic chronic parasitic disease caused by tapeworms of the genus Echinococcus.This serious worldwide disease remains a major public health problem,and western China is a highly endemic area of the disease.Echinococcosis has a long incubation period,and patients may have no obvious clinical symptoms for many years.There is currently no standardized and widely accepted treatment and early and accurate diagnosis are crucial for the prevention and treatment of echinococcosis.The current clinical diagnostic methods have many drawbacks,and the fatal one is that they cannot be used to detect the early stage of infection.In recent years,Raman spectroscopy has shown great potential for clinical application in the rapid,non-invasive,and label-free detection of diseases.In this paper,Raman spectroscopy was used to systematically detect biochemical samples of echinococcosis patients and echinococcosis mouse models,and combined with machine learning and deep learning algorithms,the early and rapid diagnosis of echinococcosis was realized.The main research contents of this paper are as follows:1.Using serum Raman spectroscopy technology combined with a deep learning algorithm to realize the identification and differentiation of hepatic echinococcosis and other chronic liver diseases.Hepatic echinococcosis,liver cirrhosis,and hepatocellular carcinoma are all serious chronic liver diseases.At present,there is still a lack of clinical methods to quickly and accurately distinguish patients with hepatic echinococcosis from patients with other chronic liver diseases.To this end,a large number of serum Raman spectral data of patients with chronic liver disease and control group were collected.Based on the one-dimensional convolutional neural network(1D-CNN)model,the prediction accuracy for various chronic liver diseases reached 95.7%.Compared with traditional machine learning classification models,the 1D-CNN model has less dependence on spectral preprocessing steps and can achieve a prediction accuracy of 93.9%even using raw spectral data,thus avoiding tedious preprocessing steps for spectra and saving time and computational costs.Further comparing the results of the 1D-CNN model with the results of the echinococcosis rapid detection kit,the 1D-CNN method based on serum Raman spectroscopy has the advantages of rapidity,strong repeatability,and high accuracy.2.The rapid screening of patients with hepatic echinococcosis was realized by using serum vibrational spectroscopy technology combined with a support vector machine(SVM)algorithm.Previous research has fully demonstrated the enormous potential of Raman spectroscopy technology in the rapid screening of echinococcosis.As a vibrational spectroscopy technology complementary to Raman spectroscopy,the feasibility of infrared spectroscopy in the rapid detection of echinococcosis is still unknown.To this end,the diagnostic performance,advantages,and disadvantages of the two spectral techniques were further compared,so as to provide a useful reference for the development of clinical auxiliary diagnostic techniques for echinococcosis.The vibrational spectroscopy data of serum samples from patients with hepatic echinococcosis and the control group were collected,and combined with the SVM algorithm,the diagnostic models were established respectively.The results showed that both Raman spectroscopy and infrared spectroscopy had achieved satisfactory diagnostic results,with a classification accuracy of 97.7%,and the diagnostic performance of the two was comparable.When collecting spectra,Raman spectroscopy can directly detect liquid serum,while infrared spectroscopy needs to measure air-dried serum samples to avoid water interference.However,compared with infrared spectroscopy,the scattering signal of Raman spectroscopy is relatively low,and it is also easily interfered by strong fluorescent background.3.Using label-free surface-enhanced Raman spectroscopy(SERS)technology to detect the serum samples of patients with hepatic echinococcosis and normal controls,and analyze the differences in biochemical components between serum samples of patients with hepatic echinococcosis and the control group.The SVM algorithm was used to establish a screening model,and excellent classification results were achieved.The diagnostic accuracy,specificity,and sensitivity were 97.7%,100%,and 94,4%,respectively.Further analysis of the SERS characteristic peaks corresponding to uric acid and hypoxanthine in serum proved that the content of purine metabolites in serum is of great value in the clinical screening of hepatic echinococcosis.The research results show that the label-free serum SERS spectroscopy combined with the SVM algorithm has great application potential in the non-invasive detection of hepatic echinococcosis.4.In view of the lack of reliable and effective early screening technology for echinococcosis,and the lack of clinical samples from patients with early stages of echinococcosis infection.To this end,the mouse animal model was innovatively constructed,and a strategy based on the combination of label-free serum SERS spectroscopy and machine learning was proposed for the rapid and non-invasive diagnosis of early cystic echinococcosis(CE).The combination of the SVM algorithm successfully distinguished the serum SERS spectra of patient groups and control groups at different infection stages.The accuracy of early and midterm diagnostic models reached 91.7%and 95.7%,respectively.The results showed that label-free serum SERS analysis is a promising early CE detection method.In order to interpret the serum SERS spectroscopy more accurately,the possible assignment of spectral characteristic peaks was further correlated with untargeted metabolomics analysis,and some serum metabolic biomarkers that may be associated with CE disease progression were identified,such as uric acid and L-Glutamate,these potential biomarkers may contribute to the diagnosis and treatment of early CE.5.The rapid detection of alveolar echinococcosis(AE),CE,and the control group of mouse liver tissue sections was achieved by using the label-free SERS technology combined with the deep learning algorithm.Using silver nanoparticles as the SERS-enhanced substrate,a large number of tissue SERS spectral data were collected.A classification model was established using a 1D-CNN network to classify and differentiate three groups of tissue slice SERS spectral data.The prediction accuracy,precision,recall,and F1-score of the model reached 95.4%,94.7%,94.3%,and 94.5%,respectively.Five commonly used machine learning spectral classification models were further selected for comparison,which confirmed the superiority of the 1D-CNN classification model.The research results show that the combination of tissue slice SERS spectroscopy technology with deep a learning algorithm has great potential in the clinical detection of AE and CE.This study focuses on the research hotspots of biomedicine,combined with the urgent need for clinical echinococcosis detection,and uses Raman spectroscopy to conduct pioneering and systematic research on the rapid detection and early screening of echinococcosis.It is expected to develop a new method for early detection of echinococcosis with label-free,noninvasive,fast,and high accuracy.