| 基于医学图像的数学规划支持向量机 |
| |
| 引用本文： | 孙蕾,周明全,耿国华.基于医学图像的数学规划支持向量机[J].计算机工程,2005,31(19):4-5,31. |
| |
| 作者姓名： | 孙蕾 周明全 耿国华 |
| |
| 作者单位： | 西北大学计算机科学系,西安,710069 |
| |
| 基金项目： | 国家自然科学基金资助项目（60372072） |
| |
| 摘 要： | 构造一个基于医学图像的机器学习机制。应用数学规划方法对传统的SVM进行修改，通过优化分类面的方向和位置来确定最优分类面，并将产生的非线性支持向量机应用在乳腺肿瘤的计算机辅助诊断上。在提取统计特征、纹理特征和形状特征的基础上，该文的算法取得了较低的伪阳性和伪阴性。
|
| 关 键 词： | 支持向量机 数学规划 特征提取 区域分割 |
| 文章编号： | 1000-3428（2005）19-0004-02 |
| 收稿时间： | 2004-08-04 |
| 修稿时间： | 2004-08-04 |
Mathematic Programming Support Vector Machine Based on Medical Images |
| |
| SUN Lei,ZHOU Mingquan,GENG Guohua.Mathematic Programming Support Vector Machine Based on Medical Images[J].Computer Engineering,2005,31(19):4-5,31. |
| |
| Authors: | SUN Lei ZHOU Mingquan GENG Guohua |
| |
| Abstract: | A machine learning mechanism is constructed based on medical images. The standard support vector machine is improved by mathematic programming in this work. The optimal hyperplane is achieved by optimizing both its direction and location. And the produced nonlinear SVM is applied to computer-aided diagnosis of breast tumor. Low false positive and false negative are achieved based on statistic features, texture features and shape features utilizing the presented algorithm. |
| |
| Keywords: | Support vector machine Mathematic programming Feature extraction Region segmentation |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
