摘要
针对乳腺肿块具有大小不固定和个体差异等特性,提出了一种多支持向量机对乳腺肿块的识别方法.选择八个方向上的支持向量生成分类器,选取高斯核函数作为核函数,其中取σ=30时分割正确率达到97.3%.表明多支持向量机应用于乳腺肿块识别可以获得较好的识别效果,为进一步的医学诊断提供可靠的依据.
In view of the varying size and individnal differences of breast tamors,this paper provides a method for breast tumour recognition——Multi-Support Vector Machine（MSVM）.Support Vector Machine（SVM） in eight direction of bump area are taken to generate a vector classifier.Gauss kernel function was as the selected kernel function.The breast tumour recognition accuracy can reach 97.3% when σ=30.The experiment shows that the application of MSVM in breast tumor recognition can achieve good result,and provide a reliable basis for further medical diagnosis.
出处
《西安工业大学学报》
CAS
2011年第2期160-163,共4页
Journal of Xi’an Technological University
关键词
多支持向量机
核函数
肿瘤识别
图像分割
边缘提取
multi-support vector machine
kernel function
tumour recognition
image segmentation
edge detection
