Research Of Benign And Malignant Breast Cancer Recognition Model Based On Raman Spectroscopy

Breast cancer is one of the most common cancers in the world,and the incidence of it is increasing year by year.Raman spectroscopy can be used to characterize and explain the changes of tissue composition on the basis of molecular level.It has the advantages of high sensitivity and no damage in the diagnosis of disease and in situ detection of living tissue.However,Raman spectrum data,acquired from the experiments,are high-dimensional,and noise.It is difficult to distinguish the benign and malignant breast tumors directly.Therefore,it is necessary to establish a model to distinguish benign and malignant breast tumors.The model can help doctors to carry out targeted treatment.The approved model,using Raman spectrum data and machine learning method,can improve the accuracy in the diagnosis of breast tumors,and achieve better therapeutic effect.Fresh tissues were obtained from 168 female patients in the First Hospital of Jilin University.The collected Raman spectra data are complex and could not directly used to construct a classification model.According to the researchers' work,we summed up the typical characteristics Raman data peaks of benign and malignant breast tissue.These peaks can characterize composition characterization of breast tissue lesions.After the dimension reduction,the support vector machine(SVM),extreme learning machine(ELM)and K nearest neighbor(KNN)are used to build the models.It is found that the prediction accuracy of classification are from 51.67% to 85.00%,and the models have obvious tendency to malignant tissue.In order to solve the problem above,we adopt the methods of feature selection and feature extraction to find out the best combination of feature subset.So that we could achieve higher classification accuracy and more stable model.Using sequence forward selection(SFS),Relief-F and joint sparse discriminant analysis(JSDA)to analyze the characteristic peaks of breast tissue,we find the best combination of feature subsets and build the models using the above mentioned classification method.The experimental results show that the prediction accuracy of the classification model constructed by the combination of feature selection and feature extraction is better than that of all the feature peaks.Among them,the classification model based on KNN and JSDA(KNN-JSDA)obtained the best classification accuracy.The accuracy rate of KNN-JSDA model for benign and malignant breast tumors was 93.33%.The Kappa coefficient is 0.84,which indicates the significative classification results.The results show that the KNN-JSDA model established in this paper has good recognition ability,and can identify benign and malignant breast tumors.