摘要
提出了一套结合特征筛选及参数设定的方法,使用支持向量机来辨别肿瘤良恶性,并利用人工免疫算法进行特征筛选及决定支持向量机的参数。针对由PHILIPS ATL HDI 3000超声波扫描仪获得的220幅图片的处理结果显示,在此所提出的方法能使乳房肿瘤的分类正确率达到95.71%,并大幅缩短支持向量机的训练时间。
An approach combining the feature selection and parameter setting is proposed,in which SVM is adopted to rec-ognize whether a tumour is malignant or not,and the AIS algorithm is utilized to select the tumor features and determin the pa-rameters of SVM. The experimental results indicate that the approach can make the classification accuracy of the breast tumour reach to 94.63%,improve the correctness of adjusting the quale of a breast tumour,and shorten training time of the computer-aided diagnosis system based on ultrasound breast image.
出处
《现代电子技术》
2014年第4期108-111,共4页
Modern Electronics Technique
