摘要
为了有效提高轻度认知障碍(mild cognitive impairment,MCI)的早期诊断效果,提出了基于SVM的fMRI数据分类方法,并构建分类准确率更高的集成分类器。传统的MCI诊断过程,检验周期长、主观误差较大,为此,利用数据挖掘技术,采用SVM数据分类方法,通过提取单个体素的分类特征,对fMRI图像作分类,并分析分类准确率较高的体素分布区域。通过加权平均的方法,构建集成分类器,更好地辅助临床诊断。
To improve the quality of early diagnosis of MCI （mild cognitive impairment） effectively, a data classification method based on SVM （support vector machine） is presented and a ensemble classifier with higher classification accuracy is constructed. The traditional MCI diagnosis processes have long inspection cycles and high subjective errors. To achieve the data mining, the SVM data classification method is proposed. The feature is extracted from each voxel for the classification of fMRI images. Then, the distribution of the voxels with higher classification accuracy is analyzed. Finally, the ensemble classifier is constructed by using weighted means method so as to assist in the clinical diagnosis better.
出处
《计算机工程与设计》
CSCD
北大核心
2013年第9期3313-3317,共5页
Computer Engineering and Design
基金
国家自然科学基金项目(60975032)
山西省自然科学基金项目(2011011015-4)
北京市博士后科研活动经费基金项目(Q6002020201201)
关键词
轻度认知障碍
数据分类
支持向量机
分类算法
分类器集成
mild cognitive impairment （MCI）
data classification
support vector machine （SVM）
classification algorithm
classifiers ensemble
