Application Of Improved Support Vector Machine To The Diagnosis Of Benign And Malignant Breast Tumors

Cancer is a kind of malignant tumor derived from the epithelial tissue,which is a common serious disease.In recent years,the prevalence rate of cancer constantly increases.The cancer risk in female is higher than that in male.Breast cancer is the highest rate of cancer among female,which has seriously affected the physical and psychological health of patients.Currently,the pathological diagnosis is a common method to diagnose the benign and malignant tumors.However,there is certain misjudgment rate in the diagnosis of benign and malignant tumors due to the subjective judgment of physicians.Auxiliary diagnosis of tumor pathological data of patients by establishing classifier can not only accelerate the process of the diagnosis and treat,but also can significantly improve the misdiagnosis situation due to the deficiency of the experience and professional skills of physicians.The data in this paper are 699 breast cancer data samples which are from the University of Wisconsin Hospital.Therefore an Boosting-SVM integration learning method with difference is proposed which is based on traditional support vector machine model.This method uses resampling to obtain the sub-training sets for local class balancing,which can train the support vector machine based classifiers with good generalization.Meanwhile,the the diversity measure to model tuning was introduced,which can increase the difference between the base classifiers and overcome the problem of insufficient sample size and improve the generalization of the model.The experimental results show that the improved method can raise the recognition rate and have more stable performance compared to the traditional support vector machine.The classifier with high precision can decrease the misdiagnosis due to the subjective judgement of physicians,which can guarantee the accurate and timely treatment for patients with tumors.