Uncertainty in fMRI Functional Networks of Autism Brain Imaging Data

TL;DR

This paper reviews the preprocessing of fMRI data into brain networks, emphasizing the importance of modeling uncertainties in functional correlations with probabilistic edges to better understand brain connectivity.

Contribution

It introduces a probabilistic modeling approach for functional correlations in fMRI brain networks, highlighting the impact of preprocessing parameters on network interpretation.

Findings

Uncertainty in fMRI network construction can be modeled probabilistically.

Preprocessing parameters significantly influence network analysis.

Probabilistic edges better represent inherent uncertainties.

Abstract

In this paper we review the preprocessing pipeline through which fMRI data is transformed into a network. We discuss three parameters that mostly affect our understanding of the existence of functional correlations between the brain regions. In the end, we conclude that the existence of functional correlations between pairs of the brain's regions can be modeled with probabilistic edges, not to lose the uncertainty that is inherent in the network generation process.