Benchmarking Brain Connectivity Graph Inference: A Novel Validation Approach

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Chevaux, Alice, Fahkar, Ali, Polisano, Kévin, Gannaz, Irène, Achard, Sophie
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913744830857216
author Chevaux, Alice
Fahkar, Ali
Polisano, Kévin
Gannaz, Irène
Achard, Sophie
author_facet Chevaux, Alice
Fahkar, Ali
Polisano, Kévin
Gannaz, Irène
Achard, Sophie
contents Inferring a binary connectivity graph from resting-state fMRI data for a single subject requires making several methodological choices and assumptions that can significantly affect the results. In this study, we investigate the robustness of existing edge detection methods when relaxing a common assumption: the sparsity of the graph. We propose a new pipeline to generate synthetic data and to benchmark the state of the art in graph inference. Simulated correlation matrices are designed to have a set of given zeros and a constraint on the signal-to-noise ratio. We compare approaches based on covariance or precision matrices, emphasizing their implications for connectivity inference. This framework allows us to assess the sensitivity of connectivity estimations and edge detection methods to different parameters.
format Preprint
id arxiv_https___arxiv_org_abs_2503_15012
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Benchmarking Brain Connectivity Graph Inference: A Novel Validation Approach
Chevaux, Alice
Fahkar, Ali
Polisano, Kévin
Gannaz, Irène
Achard, Sophie
Methodology
Inferring a binary connectivity graph from resting-state fMRI data for a single subject requires making several methodological choices and assumptions that can significantly affect the results. In this study, we investigate the robustness of existing edge detection methods when relaxing a common assumption: the sparsity of the graph. We propose a new pipeline to generate synthetic data and to benchmark the state of the art in graph inference. Simulated correlation matrices are designed to have a set of given zeros and a constraint on the signal-to-noise ratio. We compare approaches based on covariance or precision matrices, emphasizing their implications for connectivity inference. This framework allows us to assess the sensitivity of connectivity estimations and edge detection methods to different parameters.
title Benchmarking Brain Connectivity Graph Inference: A Novel Validation Approach
topic Methodology
url https://arxiv.org/abs/2503.15012