Related Experiment Video
Updated: Jan 13, 2026
07:12
Using Informational Connectivity to Measure the Synchronous Emergence of fMRI Multi-voxel Information Across Time
Published on: July 1, 2014
12.7K
Topology Assisted Clustering of Temporal fMRI Brain Networks With Use-Case in Mitigating Non-Neural Multi-Site
Ahmedur Rahman Shovon1,2,3,4,5,6, Sidharth Kumar1, Gopikrishna Deshpande7,2,3,4,5,6
1Department of Computer Science, University of Illinois Chicago, Chicago, IL 60607, USA.
IEEE Access : Practical Innovations, Open Solutions
|January 12, 2026
Summary
Topological data analysis (TDA) offers a robust method for clustering dynamic functional connectivity in fMRI data, overcoming challenges from varying sampling rates and non-neural artifacts. This TDA-based pipeline enhances the reliability of brain connectivity analysis across diverse scanning conditions.
Area of Science:
- Neuroimaging and Computational Neuroscience
- Brain Connectivity Analysis
- Topological Data Analysis Applications
Background:
- Dynamic functional connectivity analysis using fMRI is crucial for understanding brain function over time.
- High dimensionality and non-neural artifacts, such as varying temporal sampling rates, pose significant challenges in fMRI data analysis.
- Existing graph-based methods for dynamic connectivity are sensitive to arbitrary threshold choices, limiting robustness.
Purpose of the Study:
- To develop a robust temporal clustering pipeline for fMRI-derived dynamic functional connectivity.
- To address challenges of high dimensionality, non-neural variability, and varying sampling rates in fMRI data.
- To leverage Topological Data Analysis (TDA) for enhanced feature extraction and clustering of brain network dynamics.
Main Methods:
- Developed a TDA-based temporal clustering pipeline to extract noise-invariant features from dynamic functional connectivity.
- Applied the pipeline to resting-state fMRI data from 316 subjects scanned with different temporal sampling periods.
- Compared the TDA pipeline's performance against direct time-series clustering, PCA-based clustering, and traditional network analysis with dimensionality reduction.
Main Results:
- The TDA pipeline demonstrated greater robustness, maintaining consistent cluster numbers across different sampling rates for the same subjects.
- Achieved higher overlap in optimal cluster numbers (59%) and pairwise similarity (74-77%) between cluster solutions across sampling cohorts.
- Validated on the ADHD-200 dataset, showing consistent clustering patterns across sites and protocols with high stability (>80% similarity) and better subject-level dynamics separation.
Conclusions:
- Incorporating network topology via TDA significantly enhances the reliability of temporal clustering in fMRI studies.
- The TDA-based pipeline offers a robust framework for studying brain dynamics across heterogeneous acquisition settings and multisite studies.
- This method effectively masks non-neural variability and preserves true neural signals, improving the characterization of dynamic brain connectivity.
