摘要
针对功能磁共振成像(functional Magnetic Resonance Imaging,fMRI)数据分类方法中多余的预测变量和实验噪声等导致无法准确提取数据的有效特征,影响分类准确率的问题,提出将独立成分分析(Independent Component Analysis,ICA)算法与Adaboost数据分类方法相结合用于分析fMRI数据.利用ICA算法通过线性变换将体素信息分解为统计独立的源信号的线性组合;不断更新分离矩阵提取脑组织边缘变化的特征信息;利用ICA算法得到的特征信息训练Adaboost分类器.实验结果显示该方法得到的平均分类准确率达到84.72%,表明其有助于对大脑中形成的视觉图像信息进行分类,为解码fMRI数据提供了一种方法.
Redundant predictive variables and experimental noise in functional magnetic resonance imaging(fMRI)data classification methods lead to inaccurate extraction of effective features of data,which affects the classification accuracy.This paper proposed an independent component analysis(ICA)algorithm combined with Adaboost data classification method to analyze fMRI data.Using ICA algorithm,voxel information was decomposed into a linear combination of statistically independent source signals through linear transformation.Then the separation matrix was updated continuously to extract the feature information of brain tissue edge changes.Adaboost classifier was trained with the feature information obtained by ICA algorithm.The experimental results show that the average classification accuracy of the method is 84.72%,and it is helpful to classify the visual image information formed in the brain and provides a method for decoding fMRI data.
作者
张芳芳
李楠
Zhang Fangfang;Li Nan(College of Information and Computer Science,Taiyuan University of Technology,Taiyuan 030024,Shanxi,China)
出处
《计算机应用与软件》
北大核心
2019年第11期107-111,共5页
Computer Applications and Software
作者简介
张芳芳,硕士生,主研领域:磁共振图像处理,模式识别;李楠,硕士生。
