李广涵,刘健,武敬平,田艳,刘将,马立勇,刘跃军,张波,郑敏.基于支持向量机超声多模态诊断肾疾病的研究[J].中国医学影像技术,2020,36(6): |
基于支持向量机超声多模态诊断肾疾病的研究 |
Study on multi-modal ultrasound diagnosis of renal diseases based on support vector machine |
投稿时间:2020-02-15 修订日期:2020-06-13 |
DOI: |
中文关键词: 支持向量机 Logistic回归 超声弹性成像 肾病 |
英文关键词:support vector machine logistic regression ultrasound elastography nephropath |
基金项目:1,国家政府间国际科技创新合作重点专项 项目编号:2017YFE0110500 2,山东省自然科学基金 编号:ZR2018MF026 |
|
摘要点击次数: 1922 |
全文下载次数: 628 |
中文摘要: |
目的:比较支持向量机(support vector machine,SVM)和传统的Logistic回归构建基于常规超声、彩色多普勒超声和弹性成像多模态联合诊断肾脏疾病的诊断能力。
方法:收集有病理诊断的肾脏疾病患者94例及正常对照组109例,分别进行常规超声、彩色超声和剪切波弹性检查。建模的方法采用支持向量机和Logistic回归。其中SVM法根据随机数字法分为3:1两组,153例作为训练样本,进行单因素变量判断和模型建立,50例患者作为验证样本,用于评价根据训练样本建立的SVM模型预测效果。结果:共入组203例患者,Logistic回归中左肾皮质弹性硬度和右肾宽度进入回归方程,Logistic回归模型准确性83.74%,支持向量机模型的诊断准确性85.10%,二者具有相似的诊断能力。
结论:多模态超声具有较高的肾脏疾病诊断效能,支持向量机和Logistic模型具有相似的诊断能力。 |
英文摘要: |
Methods: 94 cases of renal diseases with pathological diagnosis and 109 cases of normal control group were collected and examined by conventional ultrasound, color Doppler ultrasound and shear wave elastic imaging respectively. The method of modeling is support vector machine and logistic regression. Support vector machine and logistic regression are used in the modeling. The SVM method is divided into 3:1 two groups according to the random number method. 153 cases are used as training samples for single factor variable judgment and model establishment, and 50 cases are used as validation samples to evaluate the prediction effect of SVM model established based on training samples. Results: a total of 203 patients were enrolled in this study. The elastic hardness of left renal cortex and the width of right kidney entered the regression equation in logistic regression. The accuracy of logistic regression model was 83.74%, and that of SVM model was 85.10%. They had similar diagnostic ability.
The SVM method is divided into 3:1 groups according to the random number method. 153 cases are used as training samples to select variables and establish models, and 50 cases are used as validation samples to evaluate the prediction effect of SVM model based on training samples. Results: through the prediction and verification of 203 patients, the elastic hardness of left renal cortex and the width of right kidney entered the regression equation, and the accuracy of logistic regression model was 83.74%. The diagnostic accuracy of SVM model is 85.10%, which is slightly higher than that of logistic regression model.
Conclusion: Multimodal ultrasound has a high diagnostic efficiency of kidney disease, and support vector machine and logistic model have similar diagnostic ability.support vector function is better than logistic regression model in the diagnosis of renal injury. Renal width and shear wave elastic value of renal cortex can better reflect renal injury.
Conclusion: support vector function is better than logistic regression model in the diagnosis of renal injury. |
查看全文 查看/发表评论 下载PDF阅读器 |