Study On Raman Spectrum Disease Diagnosis Algorithms Based On Signal Processing

Raman spectroscopy is a non-invasive and highly sensitive optical detection technology based on inelastic scattering.Each molecule or sample has its own unique spectral "fingerprint" information,which has been widely used in the field of biomedicine.In this study,human serum Raman spectroscopy signals were taken as the research,and combined with the pattern recognition methods,the screening models of high renin hypertension and thyroid dysfunction diseases were established respectively.This research mainly focuses on the following two aspects:1.Raman spectroscopy combined with different classification algorithms has been used for the first time in the screening of high renin hypertension disease.The original Raman spectroscopy data were fitted by polynomial and the fluorescence background noise was subtracted.The high-dimensional spectra data were extracted by principal component analysis(PCA).Finally,support vector machine(SVM),linear discriminant analysis(LDA)and K nearest neighbor(KNN)classification models were established respectively.The classification accuracy of these three models was 93.5%,93.5% and 89.1%,respectively.The results showed that serum Raman spectroscopy combined with multivariate data analysis was feasible in the screening of high renin hypertension disease.2.An artificial fish coupled with uniform design(AFUD)algorithm was proposed and applied to the hyperparameter optimization of SVM.The hyperparameter selection of SVM has a great influence on its classification accuracy.In order to solve the hyperparameter optimization problem of SVM,the AFUD algorithm was proposed for the first time and used in the hyperparameter optimization of SVM.A diagnostic model of thyroid dysfunction was established based on serum Raman spectroscopy data.The results show that,compared with artificial fish swarm algorithm(AFSA)and grid search(GS)algorithm,the AFUD algorithm proposed in this paper not only improves the classification accuracy slightly,but also improves the optimization efficiency(in terms of search time)in the hyperparameter optimization of SVM.The research results not only provides a new method for hyperparameter optimization of SVM,but also demonstrates the feasibility of Raman spectroscopy combined with SVM algorithm for the screening of thyroid dysfunction diseases.The results of this study show that Raman spectroscopy combined with PCA-SVM,PCA-LDA and PCA-KNN models can effectively distinguish different subtypes of hypertension,which provides a new method for clinical classification of hypertension.It also shows that the method of hyperparameter optimization is helpful to improve the classification accuracy and efficiency of the model,and provides a new idea for developing a screening method for thyroid dysfunction diseases based on spectral data.