摘要
Abstract
Objective To propose a multimodal ultrasound diagnosis method,which combines deep learning with multimodal ultrasound imaging technology,in order to further improve the application value of multimodal ultrasound imaging technology in the diagnosis of liver diseases,and achieve higher precision disease diagnosis.Methods Firstly,the concept of edge perception was introduced,and a network framework for image segmentation was proposed.The image segmentation model was obtained by training,and the multi-modal ultrasound images were preprocessed.Then,support vector machine(SVM)was used as the base classifier,and multiple base classifiers were established by combining the characteristics of multimodal ultrasound images.Finally,the coefficients of the base classifiers were determined according to the information entropy,dispersion and mean-variance ratio,and the fused multimodal ultrasound diagnosis results were obtained.Results The research results showed that the diagnostic accuracy of the multi-modal ultrasonic imaging diagnostic model combined with deep learning method reached 95.78%,the false positive rate and false negative rate were 4.23%and 4.50%respectively,F1 score reached 0.93,and Kappa statistic reached 0.92,with high diagnostic accuracy.Conclusion The multimodal ultrasound diagnosis method proposed in this study has high application value in the diagnosis of liver diseases,which can help doctors to determine the disease status of patients.关键词
多模态/超声成像/肝脏疾病/诊断/图像分割/分类/支持向量机Key words
multimodal/ultrasonic imaging/liver disease/diagnosis/image segmentation/categorize/support vector machine(SVM)分类
预防医学
