摘要
对每一个训练点都定义点模糊度,利用其隶属函数所包含的信息量来确定模糊度,在此基础上对传统的支持向量机算法进行了改进,提出了基于模糊支持向量机的医学图像分类技术。采用不同噪声图像进行的试验结果表明,模糊支持向量机方法能够较好地对MRI图像中脑组织进行分类,并且具有较高的精度。使用该方法还可以减少计算量,提高运算速度。
A new method for medical image classification was proposed which is based on fuzzy support vector machine. In this method a fuzzy degree is defined for each training point, and the level of the fuzzy degree is determined by the amount of information in the subjecting function. Thus the traditional support vector machine algorithm is improved. Experiments on images with different noise levels were conducted and results show that the proposed method is able to classify the brain tissues in the MRI images with high precision. In application of this method the cost and time of computation can also be reduced.
出处
《吉林大学学报(工学版)》
EI
CAS
CSCD
北大核心
2007年第3期630-633,共4页
Journal of Jilin University:Engineering and Technology Edition
基金
国家自然科学基金资助项目(60573182)
国家博士后基金资助项目(20060390300)
关键词
计算机应用
医学图像分类
支持向量机
模糊支持向量机
模糊度
computer application
medical image classification
support vector machine
fuzzy support vector machine
fuzzy degree