Feasibility Study On The Application Of Raman Spectroscopy Technology In Screening And Distinguishing Benign And Malignant Breast Tumors

Objective:Breast cancer（BC）is a kind of malignant tumor with the highest incidence rate in women.Early diagnosis and early treatment can greatly improve the prognosis.At present,the Chinese Society of Clinic Oncology（CSCO）recommends ultrasound and molybdenum target as the first choice for breast cancer screening,but it is inevitable that overdiagnosis or missed diagnosis will occur.We propose to apply Raman Spectroscopy（RS）to the feasibility study of early breast cancer screening,and explore the application value of Raman spectroscopy in early breast cancer screening.Methods:Retrospective analysis was used to select the patients who were treated in breast surgery of Sichuan Cancer Hospital from October 2021 to October 2022.Among them,129 cases were diagnosed as malignant and 91cases were benign by pathological examination,and 113 cases were set up as health group.The serum of malignant,benign and healthy patients were detected by Raman spectroscopy.Principal Component Analysis（PCA）and Support Vector Machine（SVM）were used to classify healthy,malignant and benign sample groups,and the SVM model was evaluated by cross-validation.Further obtain the sensitivity and specificity of Raman spectrometer to distinguish benign,malignant and normal breast tissue.Finally,the receiver operating characteristic（ROC）was used to further measure the feasibility of this method in differentiating benign and malignant breast and normal breast.Results:The difference of the three groups of spectra mainly lies in the content changes of biological macromolecules such as proteins,lipids,carbohydrates and nucleotides.There are some similar changes in the serum Raman signals of the malignant tumor group and the benign group,but there are significant differences in the average Raman spectra between the healthy group and the malignant tumor group/benign group.Compared with healthy controls,the percentage of total Raman active components of some biological molecules in the serum of breast cancer patients increased or decreased.The main difference between the malignant tumor group and the control group is that the intensity at the peak spectrum of 784cm^-1,925cm^-1,986cm^-1,1020cm^-1,1056cm^-1,1114cm^-1,1127cm^-1,1139cm^-1,1346cm^-1,1367cm^-1in the cancer spectrum increases,and the intensity at the peak spectrum of 835cm^-1,989cm^-1,1002cm^-1,1285cm^-1,1295cm^-1,1437cm^-1,1531cm^-1,1650cm^-1decreases.Compared with the healthy group,the peak intensity of the benign group increased at 784cm^-1,925cm^-1,986cm^-1,1020cm^-1,1056cm^-1,1114cm^-1,1127cm^-1,1139cm^-1,1346cm^-1,1367cm^-1,and decreased at 835cm^-1,1002cm^-1,1285cm^-1,1295cm^-1,1437cm^-1,1531cm^-1,1650cm^-1.The malignant tumor group and the benign group have similar spectral signals.However,there are still obvious peaks of intensity differences at 925cm^-1,986cm^-1,989cm^-1,1002cm^-1,1020cm^-1,1056cm^-1,1114cm^-1,1285cm^-1,1295cm^-1,1346cm^-1,1367cm^-1,1437cm^-1,1650cm^-1.Through PCA-SVM analysis and cross-validation,the sensitivity and specificity of Raman spectrum in differentiating malignant tumor group,benign group and healthy group were 93.2%、90.1%、93.1%respectively,and 95.1%、97.1%、95.9%respectively.The accuracy of SVM model area classification of malignant,benign and healthy was 98.7%,98.6%and 99%respectively.The area under curve（AUC）of the three groups of samples in the malignant tumor group,the benign group and the healthy group were 0.990,0.987 and 0.987respectively.Conclusion:There are obvious differences in Raman spectra between the healthy group and the malignant tumor group and the benign group.The malignant tumor group and the benign group have similar Raman signals.The malignant tumor group shows earlier changes in biochemical components.The Raman spectrometer can distinguish the healthy,malignant tumor group and benign group at an early stage,and has high screening application value.