Clinical Trial Research On REIS Detection System Of Breast Cancer

Breast cancer has always been one of the most important questions of women’s health,especially in recent years,the incidence of breast cancer is increasing year by year.Early detection and early treatment are two important measures to control the development of breast cancer in clinic.The existing detection methods mainly are: ultrasound,magnetic resonance imaging and mammography.But deficiencies exist in these detection methods,such as breast ultrasound images lack of specificity for less than 10 mm tumor,breast molybdenum target has radiation risk,mammography is expensive.These methods are only applicable for a limited group of women,and not suitable for large-scale early detection of breast cancer.With the extensive use of Electrical Impedance Technique in the detection of breast cancer,Resonance-frequency-based Electrical Impedance Spectroscopy(REIS),which has the features of no damage,non-imaging,low cost,ease of operation and so on,has a unique advantage in early detection and early warning of the tumor.This paper target at the analysis and research of the clinical application of the breast cancer REIS detection system.By doing this,not only analysis and master the characteristics of female breast tissue REIS data curves,but also can evaluate the effectiveness of REIS system in breast cancer detection applications.For this purpose,collecting the clinical REIS data from voluntary female Patients.Patients are grouped into 215 benign cases including fibro adenomas and cyst,74 malignant cases of ductal carcinoma in situ and invasive carcinoma and 149 no tumor cases with lobular hyperplasia.All cases were confirmed by B-reports,and benign or malignant tumors are pathological report results verified.Breast cancer REIS detection system,a discriminative assessment of breast patients was performed by the symmetry of breast tissue identification.After that,the initial collected data pretreatment with data balance,electrode analysis,removing abnormal data and so on,according to the difference between the two patients’ breast impedance spectra obtained by the REIS detection system,22 REIS based features,including seven frequency and fifteen magnitude features were extracted.To reduce redundancy,we selected features using the Support Vector Machine Recursive Feature Elimination(SVM-RFE)and evolutionary algorithm.We also using logistic regression classifier to identifiy and evaluate no tumor and tumor,no tumor and benign tumor,non tumor and malignant tumor,benign and malignant disease cases in the different database.The results of classification of patients with benign and malignant tumors showed AUC(the area under a receiver operating characteristic curve),ACC(the accuracy for classification),sensitivity and specificity of 0.815,0.758,0.649 and 0.738 respectively for classifier.The AUC,ACC,sensitivity and specificity for the classifier after feature selection were 0.846,0.80,0.703 and 0.833,respectively.The results indicated that the REIS system is a promising tool for detection of breast cancer,it can provide valuable supplementary information for clinical diagnosis.