Relationship Between Basic Properties of BOLD Fluctuations and Calculated Metrics of Complexity in the Human Connectome Project.

Shella Keilholz,Xiaodi Zhang,Theodore J Lagrow,Eric Maltbie,Nan Xu,Behnaz Yousefi,Maysam Nezafati,Wen-Ju Pan,Ying Guo

doi:10.3389/fnins.2020.550923

Shella Keilholz, Xiaodi Zhang + Show 7 more

Open Access

https://doi.org/10.3389/fnins.2020.550923

Copy DOI

Abstract

Resting state functional MRI (rs-fMRI) creates a rich four-dimensional data set that can be analyzed in a variety of ways. As more researchers come to view the brain as a complex dynamical system, tools are increasingly being drawn from other fields to characterize the complexity of the brain’s activity. However, given that the signal measured with rs-fMRI arises from the hemodynamic response to neural activity, the extent to which complexity metrics reflect neural complexity as compared to signal properties related to image quality remains unknown. To provide some insight into this question, correlation dimension, approximate entropy and Lyapunov exponent were calculated for different rs-fMRI scans from the same subject to examine their reliability. The metrics of complexity were then compared to several properties of the rs-fMRI signal from each brain area to determine if basic signal features could explain differences in the complexity metrics. Differences in complexity across brain areas were highly reliable and were closely linked to differences in the frequency profiles of the rs-fMRI signal. The spatial distributions of the complexity and frequency metrics suggest that they are both influenced by location-dependent signal properties that can obscure changes related to neural activity.

Highlights

Resting state functional magnetic resonance imaging is a popular tool for characterizing the functional architecture of the brain
In the large dataset available from the Human Connectome Project (HCP), we look at the relationship between these fundamental features and the higher level metrics of correlation dimension, approximate entropy, and Lyapunov exponent
We assessed the reliability of the metrics between D1S1-D1S2 and D1S1-D2S1 using intraclass correlation (ICC)

Summary

Introduction

Resting state functional magnetic resonance imaging (rs-fMRI; Biswal et al, 1995) is a popular tool for characterizing the functional architecture of the brain. Areas with higher power and stronger low frequency fluctuations are thought to have higher spontaneous neural activity In line with this hypothesis, areas like the posterior cingulate cortex (PCC) that exhibit high resting state metabolism in PET scans have higher ALFF and fALFF than other cortical areas (Zou et al, 2008). These measures of low frequency power can discriminate between patient groups and healthy controls or between rest and task performance (Zang et al, 2007; Cui et al, 2019; Li et al, 2019; Zhou et al, 2019)

Methods

Results

Discussion

Conclusion