改进谱聚类算法在MCI患者检测中的应用研究

doi:10.11959/j.issn.1000-436x.2015181

摘要/Abstract

摘要：

摘要：为了利用功能核磁影像（fMRI，functional magnetic resonance imaging）数据进行轻度认知障碍（MCI，mild cognitive impairment）自动检测，对患者的 fMRI 数据进行聚类分析，得到患者大脑血氧依赖水平(BOLD，blood oxygen level dependence)的变化模式，并将异常模式用于疾病检测中。由于传统谱聚类算法需要计算相似矩阵所有的特征值和特征向量、时间与空间复杂度较高。提出一种改进的谱聚类方法，在相似矩阵的构造以及σ与k值的确定等方面进行了改进，将其用于MCI fMRI数据的聚类与诊断研究中。与传统谱聚类及Nystr?m算法进行的对比实验结果表明，改进的谱聚类方法可以更准确得到患者异常BOLD模式，分类正确率较高，且时间和空间复杂度均小于传统算法。

关键词: 谱聚类, Nystr?m, fMRI-BOLD, 轻度认知障碍, MCI诊断

Abstract:

In order to detect mild cognitive impairment (MCI) using functional magnetic resonance imaging (fMRI)，a method based on fMRI clustering was proposed fMRI data were clustered to obtain the blood oxygen level dependence( BOLD) change model of MCI patients，then abnormal patterns were used to detect disease.The traditional spectral clustering algorithm needs to calculate all of the eigenvalue and eigenvector，so time and space complexity is higher.An improved spectral clustering method was proposed which modified the similar matrix construction method and the setting method of σ and k，and then this method was applied to clustering and detection of MCI patients.To verify the performance of the proposed method，the comparison of the clustering result，classification accuracy using traditional algorithm and Nystr?m is also done.The comparative experimental results show that the proposed method can get BOLD pattern more accurately，the accuracy of MCI detection is higher than the other two algorithms，and the time and space complexity are less than the traditional algorithm.

Key words: spectral clustering, Nystr?m, fMRI-BOLD, MCI, MCI detection

相洁,赵冬琴. 改进谱聚类算法在MCI患者检测中的应用研究[J]. 通信学报, 2015, 36(4): 27-34.

Jie XIANG,Dong-qin ZHAO. Improved spectral clustering algorithm and its application in MCI detection[J]. Journal on Communications, 2015, 36(4): 27-34.

图/表 8

图1

图2

图3

图4

图5

图6

表1

表2

3种聚类算法的复杂度对比"

算法	时间	空间
经典谱聚类算法	计算N×N维的相似矩阵W，计算n×n的对角矩阵D，所有特征值和特征向量	需要存放 n×n 维的数组和所有的特征向量
Nystr?m算法	只需要计算矩阵维数为 n×n 的A和n×(N?n)的B，高斯核函数具有正定性，直接估计 $L = D^{-}^{_{2}^{1}} A D^{{^{-}}^{_{2}^{1}}}$ 2的主特征向量	不需要将相似矩阵 W放到内存中，估计是一个理论过程，只需存放 $L = D^{-}^{_{2}^{1}} A D^{{^{-}}^{_{2}^{1}}}$ 的主特征向量
改进算法	计算相似矩阵，对所有特征值进行排序。计算特征距最大的特征值对应的特征向量	存放稀疏矩阵

表2

参考文献 15

[1]	BROOKMEYER R ， JOHNSON E ， ZIEGLER G K ，et al. Forecasting the global burden of Alzheimer's disease[J]. Alzheimers Dement， 2007，3(3): 186-191.
[2]	MISRA C ， FAN Y ， DAVATZIKOS C . Baseline and longitudinal patterns of brain atrophy in MCI patients，and their use in prediction of short-term conversion to AD:results from ADNI[J]. NeuroImage， 2009，44: 1414-1422.
[3]	蔡晓妍，戴冠中，杨黎斌 . 谱聚类算法综述[J]. 计算机科学， 2008，35(7): 14-18. CAI X Y ， DAI G Z ， YANG L B . Survey on spectral clustering algorithms[J]. Computer Science， 2008，35(7): 14-18.
[4]	JORDAN M I ， NG A Y ， WEISS Y . On spectral clustering:analysis and an algorithm[A]. Proceedings of the 14th Advances in Neural Information Processing Systems (NIPS 2002)[C]. Cambridge，MA， 2002. 849-856.
[5]	YEO B T ， OU W . Clustering fMRI time series[EB/OL]. . 2004
[6]	WU C . Feature selection for fMRI classification[J].Program of Computational Biology Carnegie Mellon University Pittsburgh，PA 15213.
[7]	WANG C ， TIAN J ， CHEN S ，et al. Image segmentation using spectral clustering[A]. 2012 IEEE 24th International Conference on Tools with Artificial Intelligence.IEEE Computer Society[C]. 2005. 677-678.
[8]	ALKAN S ， YARMAN V F T . Localization of semantic category classification in fMRI images[A]. Signal Processing and Communications Applications Conference (SIU)[C]. 2014. 2178-2181.
[9]	FOWLKES C ， BELONGIE S , CHUNG F , et al.Spectral groupingusing the Nystrom method[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence， 2004，26(2): 214-225.
[10]	CHEN W Y ， SONG Y ， BAI H ，et al. Parallel spectral clustering in distributed systems[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence， 2011，33(3): 568-586.
[11]	孔万增，孙志海，杨灿等 . 基于本征间隙与正交特征向量的自动谱聚类[J]. 电子学报， 2010，38(8): 1880-1891. KONG W Z ， SUN Z M ， YANG C.et al . Automatic spectral clustering based on eigengap and orthogonal eigenvector[J]. Acta Electronica Sinica， 2010，38(8): 1880-1891.
[12]	相洁 . 启发式问题解决认知神经机制及fMRI数据分析方法研究[D]. 太原理工大学， 2010. XIANG J . Study of Cognitive Neuroscience Mechanism of Heuristic Problems Solving and Methods of fMRI Data Analysis[D]. Taiyuan University of Technology， 2010.
[13]	YAN C G ， ZANG Y F . DPARSF:a MATLAB toolbox for “pipeline”data analysis of resting-state fMRI[J]. Frontiers in Systems Neuroscience， 2010，5(4): 1-3.
[14]	CHANG C G ， LIN C J . LIBSVM:a library for support vectormachines[EB/OL]. .
[15]	吕艳阳，相洁 . 基于SVM的 fMRI 数据分类及MCI诊断应用[J]. 计算机工程与设计， 2013，34(9): 3313-3317. LV Y Y ， XIANG J . fMRI data classification based on SVM and its application in diagnosis of MCI[J]. Computer Engineering and Design， 2013，34(9): 3313-3317.