Chinese Journal of Intelligent Science and Technology ›› 2019, Vol. 1 ›› Issue (2): 145-153.doi: 10.11959/j.issn.2096-6652.201922

• Regular Papers • Previous Articles     Next Articles

Method of functional brain network modeling with group similarity constraint for mild cognitive impairment classification

Weikai LI1,Xin GAO2(),Tongjian JI3,Zhengxia WANG1()   

  1. 1 College of Computer Science and Technology,Nanjing University of Aeronautics and Astronautics,Nanjing 210016,China
    2 Shanghai Universal Medical Imaging Diagnostic Center,Shanghai 200233,China
    3 Shandong Juxian Maternal and Child Health Hospital,Rizhao 276500,China
  • Revised:2019-05-20 Online:2019-06-20 Published:2019-09-09
  • Supported by:
    Scientific and Technological Research Program of Chongqing Municipal Education Commission(KJQN201800716);Scientific and Technological Research Program of Chongqing Municipal Education Commission(KJ175492);Natural Science Foundation Project of CQCSTC(2018jcyjAX0398);Scientific Research Project of Shanghai Municipal Planning Commission of Health and Family Planing(201740010)


Functional brain network provides an effective biomarker for understanding brain activation patterns,the development of neurodegenerative diseases,and the structure of brain signaling.How to use priori information to build accurate brain network is particularly important in subsequent applications.A functional brain network construction model based on group similarity constraints was proposed.By introducing a tensor low rank constraint and using the nuclear norm of the tensor,a brain network group with a low rank prior in the group was solved.The group similarity constraints to shrink the solution space of the brain network was used,thus constructing a better functional brain network effectively.For evaluating the performance of the proposed method,the constructed brain network was adopted for the discriminative task of mild cognitive impairment.The experimental results show that the proposed functional brain network modeling method based on group similarity constraints can construct more discriminative brain network.In addition,based on the brain network constructed by the proposed method,the significant connections consistent with previous studies are obtained,and the effectiveness of the proposed method is further verified.

Key words: low-rank of tensor, functional brain network, functional magnetic resonance imaging, group constraint

CLC Number: 

No Suggested Reading articles found!