Research On The Method Of Breast Cancer Recognition Based On Machine Learning

Breast cancer is a serious cancer disease threatening women’s health,which takes the first place in the female cancer mortality rate.At present,the incidence and incidence rate of breast cancer in China are the first in the world,and it is on the rise.In view of the serious harm of breast cancer to life and health,relevant researchers and institutions around the world are making unremitting efforts to find a perfect diagnosis and treatment scheme.With the improvement of computer performance and machine learning,intelligent algorithms have been able to replace human behavior and judgment in some fields.The traditional diagnosis process of breast cancer requires medical experts to repeatedly observe the cell morphology of the breast sample and the chest imaging picture of the patient.In this case,the algorithm technology can be used to quickly feedback a reference result with a high probability to the doctor,which is particularly important for increasing the diagnosis efficiency and reducing the burden of the doctor’s examination.In view of the lack of accuracy of existing breast cancer identification methods,this paper studies the characteristics of breast cancer diagnosis data of numerical type and ultrasonic image category respectively,and proposes the corresponding breast cancer classification and recognition algorithm.The main research contents and achievements of this paper are as follows:1)Proposed and implemented a scheme based on WOA(Whale Optimization Algorithm)to iteratively adjust the key parameters of Support Vector Machine(SVM)to improve the accuracy of breast cancer recognition.In order to verify the performance of woa-svm algorithm,the Wisconsin Breast Cancer database WBCD(Wisconsin Breast Cancer Databas)in UCI database was used for the performance verification experiment.The experiment shows that the accuracy of woa-svm model is higher than 98.53% under the reservation method validation,and 97.50% underthe 10-fold cross-validation,which is higher than the traditional breast cancer recognition model.2)Aiming at the low accuracy of WOA-SVM algorithm under 10 fold cross validation,the algorithm was improved.Firstly,SVM-RFE recursive feature elimination method is used to reduce the dimension of the original data set,and then MWOA Algorithm is improved by introducing the idea of adding high-quality operators dynamically and MWOA Algorithm is obtained.Using the WBCD data set experiment again,the experimental results prove that mwoa-svm algorithm is superior to woa-svm in performance,and the recognition accuracy is improved by0.44%.3)Based on the feature extraction method,an algorithm for breast cancer classification and recognition based on ultrasound image data is proposed and implemented.The algorithm first USES the visual machine learning tool Weka to extract PHOG features from ultrasonic images,and then USES the random forest algorithm to classify and recognize the feature data.BUSI(Breast Ultra Sound Images)was used in the performance verification experiment,and the results showed that the AUC of the algorithm was 95.60%,which had high reliability and application value.