The thresholding problem and variability in the EEG graph network parameters

Abstract

Graph thresholding is a frequently used practice of eliminating the weak connections in brain functional connectivity graphs. The main aim of the procedure is to delete the spurious connections in the data. However, the choice of the threshold is arbitrary, and the effect of the threshold choice is not fully understood. Here we present the description of the changes in the global measures of a functional connectivity graph depending on the different proportional thresholds based on the 146 resting-state EEG recordings. The dynamics is presented in five different synchronization measures (wPLI, ImCoh, Coherence, ciPLV, PPC) in sensors and source spaces. The analysis shows significant changes in the graph’s global connectivity measures as a function of the chosen threshold which may influence the outcome of the study. The choice of the threshold could lead to different study conclusions; thus it is necessary to improve the reasoning behind the choice of the different analytic options and consider the adoption of different analytic approaches. We also proposed some ways of improving the procedure of thresholding in functional connectivity research.

Unified topological inference for brain networks in temporal lobe epilepsy using the Wasserstein distance

“…Although the method is applicable to arbitrary edge weights, in this study, the edge weights will be derived from correlations across different brain regions in rs-fMRI. Determining optimal methods to threshold the weight adjacency matrix in order to quantify brain connectivity remains a persistent challenge in graph theory ( Adamovich et al, 2022 ). Unfortunately, choices in the threshold parameter can significantly impact the results, hindering both reproducibility and cross-study comparisons, as well as the biological interpretation of results.…”

DISCOVER-EEG: an open, fully automated EEG pipeline for biomarker discovery in clinical neuroscience

“…Matrices were binarized by keeping the 20% strongest connections. This is an arbitrary threshold and it is good practice to test the reliability of results with different binarizing thresholds 47 . This threshold can be easily changed in the function define_params .…”

The time-evolving epileptic brain network: concepts, definitions, accomplishments, perspectives

“…Instead of estimating properties of interactions in sensor-space , it has been proposed to do so in source-space to overcome the problems of volume conduction (EEG) and field spread (MEG) [see, e.g., van Mierlo et al (2019) ; Koutlis et al (2021) ; Sadaghiani et al (2022) ; Chiarion et al (2023) and references therein]. Nevertheless, localizing the sources of EEG/MEG activities in the brain constitutes an inverse problem that lacks a unique solution ( von Helmholtz, 1853 ) and source-space-based network approaches continue to be critically discussed ( Colclough et al, 2016 ; Palva et al, 2018 ; Koutlis et al, 2021 ; Adamovich et al, 2022 ; Pourmotabbed et al, 2022 ; Schaworonkow and Nikulin, 2022 ; Hatlestad-Hall et al, 2023 ).…”