摘要
目的:使用基于机器学习的Fisher线性分类判别方法,对分割的乳腺微钙化数据进行线性变换,预测乳腺微钙化的性质。方法:基于Fisher线性分类判别分析原理,建立预测判别模型对乳腺微钙化的良、恶性进行分类。选取在医院行乳腺癌筛查的432例患者的原始数据,将原始数据中的30项569条乳腺癌特征数据为输入变量,以乳腺微钙化良、恶性的预测准确率为输出变量,建立乳腺微钙化分类判别模型。结果:将测试样本代入训练后的Fisher线性判别模型中,其预测乳腺微钙化的良、恶性分类准确率达到93.86%,受试者工作特征(ROC)曲线下面积(AUC)值为0.99,模型的分类性能良好。结论:建立的Fisher线性判别模型对乳腺微钙化良、恶性的预测分类效果较好,能够方便快捷地为乳腺疾病的临床诊断起辅助作用。
Objective:To predict the nature of breast microcalcification by linear transformation for the segmented microcalcification data of breast on the basis of using Fish linear classification discriminant method of machine learning.Methods:To establish predictive discriminant model for classing benign and malignant microcalcification of breast on the basis of the principle of Fisher linear discriminant analysis.The original data of 432 patients who underwent the screening of breast cancer in hospital were selected,and the 569 characteristic data of breast cancer of 30 items in original data were used as input variable,and the predictive accuracies of benign and malignant microcalcifications of breast were used as output variable in establishing the classification discriminant model of breast microcalcifications.Results:After the test samples were put into the trained Fisher linear discriminant model,the accuracy of predicting benign and malignant microcalcifications of breast reached 93.86%,and the area under curve(AUC)value of the receiver operating characteristic(ROC)curve was 0.99.The classification performance of this model was favorable.Conclusion:The established Fisher linear discriminant model has a favorable effect on the prediction and classification of benign and malignant microcalcifications of breast,and it can conveniently and quickly play an auxiliary role for the clinical diagnosis of breast diseases.
作者
汪家清
张鑫
曹彤
王能才
张海英
WANG Jia-qing;ZHANG Xin;CAO Tong(Specialist Department of Breast,Jiugang Hospital of Jiayuguan City,Jiayuguan 735100,China;不详)
出处
《中国医学装备》
2022年第2期5-9,共5页
China Medical Equipment
基金
第三批甘肃省民生课题(1303FCMA018)“远程会诊医疗信息服务平台建设与示范应用”。
关键词
乳腺微钙化
FISHER线性判别
线性变换
预测分类
机器学习
Breast microcalcification
Fisher liner discriminant
Linear transformation
Prediction and classification
Machine learning
