AUTHOR=Guo Tingting , Zhang Yining , Xue Yanfang , Qiao Lishan , Shen Dinggang 

TITLE=Brain Function Network: Higher Order vs. More Discrimination

JOURNAL=Frontiers in Neuroscience

VOLUME=15

YEAR=2021

URL=https://www.frontiersin.org/journals/neuroscience/articles/10.3389/fnins.2021.696639

DOI=10.3389/fnins.2021.696639

ISSN=1662-453X

ABSTRACT=<p>Brain functional network (BFN) has become an increasingly important tool to explore individual differences and identify neurological/mental diseases. For estimating a “good” BFN (with more discriminative information for example), researchers have developed various methods, in which the most popular and simplest is Pearson's correlation (PC). Despite its empirical effectiveness, PC only encodes the low-order (second-order) statistics between brain regions. To model high-order statistics, researchers recently proposed to estimate BFN by conducting two sequential PCs (denoted as <italic>PC</italic><sup>2</sup> in this paper), and found that <italic>PC</italic><sup>2</sup>-based BFN can provide additional information for group difference analysis. This inspires us to think about (1) what will happen if continuing the correlation operation to construct much higher-order BFN by <italic>PC</italic><sup><italic>n</italic></sup> (<italic>n</italic>&gt;2), and (2) whether the higher-order correlation will result in stronger discriminative ability. To answer these questions, we use <italic>PC</italic><sup><italic>n</italic></sup>-based BFNs to predict individual differences (Female vs. Male) as well as identify subjects with mild cognitive impairment (MCI) from healthy controls (HCs). Through experiments, we have the following findings: (1) with the increase of n, the discriminative ability of <italic>PC</italic><sup><italic>n</italic></sup>-based BFNs tends to decrease; (2) fusing the <italic>PC</italic><sup><italic>n</italic></sup>-based BFNs (<italic>n</italic>&gt;1) with the <italic>PC</italic><sup>1</sup>-based BFN can generally improve the sensitivity for MCI identification, but fail to help the classification accuracy. In addition, we empirically find that the sequence of BFN adjacency matrices estimated by <italic>PC</italic><sup><italic>n</italic></sup> (<italic>n</italic> = 1,2,3,⋯ ) will converge to a binary matrix with elements of ± 1.</p>