Journal of Computer Applications ›› 2020, Vol. 40 ›› Issue (1): 62-70.DOI: 10.11772/j.issn.1001-9081.2019061026

• Artificial intelligence • Previous Articles     Next Articles

Construction of brain functional hypernetwork and feature fusion analysis based on sparse group Lasso method

LI Yao1, ZHAO Yunpeng2, LI Xinyun1, LIU Zhifen3, CHEN Junjie1, GUO Hao1   

  1. 1. College of Information and Computer Science, Taiyuan University of Technology, Jinzhong Shanxi 030600, China;
    2. College of Art, Taiyuan University of Technology, Jinzhong Shanxi 030600, China;
    3. Department of Mental Health, First Hospital of Shanxi Medical University, Taiyuan Shanxi 030000, China
  • Received:2019-06-18 Revised:2019-09-19 Online:2020-01-10 Published:2019-10-11
  • Supported by:
    This work is partially supported by the National Natural Science Foundation of China (61672374, 61741212, 61876124, 61873178), the China Scholarship Fund supported Study Abroad Program (201708140216), the Shanxi Provincial Science and Technology Department Applied Basic Research Youth Project (201601D021073, 201801D121135), the Shanxi Provincial Education Department Science and Technology Innovation Research Project for the Universities (2016139), the CERENT Next Generation Internet Technology Innovation Project of Ministry of Education (NGII20170712), the Key Research and Development Project of Shanxi Province (201803D31043).

基于sparse group Lasso方法的脑功能超网络构建与特征融合分析

李瑶1, 赵云芃2, 李欣芸1, 刘志芬3, 陈俊杰1, 郭浩1   

  1. 1. 太原理工大学 信息与计算机学院, 山西 晋中 030600;
    2. 太原理工大学 艺术学院, 山西 晋中 030600;
    3. 山西医科大学第一医院 精神卫生科, 太原 030000
  • 通讯作者: 郭浩
  • 作者简介:李瑶(1996-),女,山西运城人,博士研究生,主要研究方向:人工智能、智能信息处理、脑影像学;赵云芃(1993-),女,山西太原人,硕士研究生,主要研究方向:人工智能、智能信息处理、脑影像学;李欣芸(1999-),女,山西太原人,主要研究方向:人工智能;刘志芬(1981-),女,山西太原人,副主任医师,博士,主要研究方向:情感障碍基础与临床研究;陈俊杰(1956-),男,河北定州人,教授,博士,主要研究方向:人工智能、智能信息处理、脑影像学;郭浩(1981-),男,山西祁县人,副教授,博士,CCF会员,主要研究方向:人工智能、智能信息处理、脑影像学。
  • 基金资助:
    国家自然科学基金资助项目(61672374,61741212,61876124,61873178);国家留学基金资助出国留学项目(201708140216);山西省科技厅应用基础研究项目青年面上项目(201601D021073,201801D121135);山西省教育厅高等学校科技创新研究项目(2016139);教育部赛尔网络下一代互联网技术创新项目(NGII20170712);山西省重点研发计划项目(201803D31043)。

Abstract: Functional hyper-networks are widely used in brain disease diagnosis and classification studies. However, the existing research on hyper-network construction lacks the ability to interpret the grouping effect or only considers the information of group level information of brain regions, the hyper-network constructed in this way may lose some useful connections or contain some false information. Therefore, considering the group structure problem of brain regions, the sparse group Lasso (Least absolute shrinkage and selection operator) (sgLasso) method was introduced to further improve the construction of hyper-network. Firstly, the hyper-network was constructed by using the sgLasso method. Then, two groups of attribute indicators specific to the hyper-network were introduced for feature extraction and feature selection. The indictors are the clustering coefficient based on single node and the clustering coefficient based on a pair of nodes. Finally, the two groups of features with significant difference obtained after feature selection were subjected to multi-kernel learning for feature fusion and classification. The experimental results show that the proposed method achieves 87.88% classification accuracy by using the multi-feature fusion, which indicates that in order to improve the construction of hyper-network of brain function, the group information should be considered, but the whole group information cannot be forced to be used, and the group structure can be appropriately expanded.

Key words: hyper-network, sparse group Lasso (Least absolute shrinkage and selection operator), clustering coefficient based on a pair of nodes, multi-kernel learning, depression, machine learning

摘要: 功能超网络广泛地应用于脑疾病诊断和分类研究中,而现有的关于超网络创建的研究缺乏解释分组效应的能力或者仅考虑到脑区间组级的信息,这样构建的脑功能超网络会丢失一些有用的连接或包含一些虚假的信息,因此,考虑到脑区间的组结构问题,引入sparse group Lasso(sgLasso)方法进一步改善超网络的创建。首先,利用sgLasso方法进行超网络创建;然后,引入两组超网络特有的属性指标进行特征提取以及特征选择,这些指标分别是基于单一节点的聚类系数和基于一对节点的聚类系数;最后,将特征选择后得到的两组有显著差异的特征通过多核学习进行特征融合和分类。实验结果表明,所提方法经过多特征融合取得了87.88%的分类准确率。该结果表明为了改善脑功能超网络的创建,需要考虑到组信息,但不能逼迫使用整组信息,可以适当地对组结构进行扩展。

关键词: 超网络, sparse group Lasso, 基于一对节点的聚类系数, 多核学习, 抑郁症, 机器学习

CLC Number: