基于免疫克隆选择算法搜索GMM的脑岛功能划分

发布时间：2018-03-29 08:20

  本文选题：脑岛功能划分　切入点：高斯混合模型(GMM)　出处：《浙江大学学报(工学版)》2017年12期

【摘要】：为了得到更好的脑岛功能划分结构,加深人们对其功能组织性的理解,提出一种基于免疫克隆选择(ICS)算法搜索高斯混合模型(GMM)的脑岛功能划分方法(NICS-GMM).该方法基于功能磁共振成像(fMRI)数据,将GMM映射到抗体上;利用ICS算法搜索能够反映脑岛功能分布的GMM,并在搜索过程中融入具有抗噪能力的动态邻域信息,以提高其搜索质量;利用最优的GMM实现对脑岛的功能划分.在划分数为2~12的脑岛功能划分上,新方法搜得的GMM具有最高的似然分数,而且相应划分结果的轮廓系数也达到了最大值.真实脑岛fMRI数据上的实验结果表明,该方法不仅具有更强的全局搜索能力,还可以得到具有较高功能一致性与更强区域连续性的脑岛功能划分结构.
[Abstract]:In order to get a better functional division structure of the brain island and deepen people's understanding of its function organization, This paper presents a method of functional division of cerebral islands based on immune Clone selection (ICS) algorithm to search Gao Si mixed model (GMM). The method is based on functional magnetic resonance imaging (fMRI) data and maps GMM to antibodies. The ICS algorithm is used to search the GMMs which can reflect the functional distribution of the cerebral islands, and the dynamic neighborhood information with anti-noise ability is incorporated into the search process to improve the search quality. The optimal GMM is used to realize the functional division of the cerebral islands. The GMM obtained by the new method has the highest likelihood fraction in the functional division of the islands with the number of 2o 12. The contour coefficients of the corresponding partition results also reach the maximum. The experimental results on the real fMRI data show that the proposed method not only has a stronger global search ability, Furthermore, the functional division structure with higher functional consistency and stronger regional continuity can be obtained.
【作者单位】： 北京工业大学信息学部多媒体与智能软件技术北京市重点实验室;南阳师范学院软件学院;
【基金】：国家“973”重点基础研究发展规划资助项目(2014CB744601) 国家自然科学基金资助项目(61375059,61672065) 河南省科技厅科技攻关资助项目(142102210588) 南阳师范学院校级青年科研资助项目(QN2017040)
【分类号】：R338;TP301.6
，

本文编号：1680312