Related Experiment Video
Updated: Aug 30, 2025
08:19
Simultaneous Data Collection of fMRI and fNIRS Measurements Using a Whole-Head Optode Array and Short-Distance Channels
Published on: October 20, 2023
1.2K
Simultaneous cluster structure learning and estimation of heterogeneous graphs for matrix-variate fMRI data.
Dong Liu1, Changwei Zhao2, Yong He2
1Shanghai University of Finance and Economics, Shanghai, China.
Biometrics
|August 26, 2022
Summary
This study introduces a novel method for analyzing brain connectivity in heterogeneous groups using functional magnetic resonance imaging (fMRI). The Simultaneous Clustering and Estimation of Heterogeneous Graphs (SCEHG) method improves graph structure learning and cluster accuracy.
Area of Science:
- Neuroscience
- Statistical Learning
- Network Analysis
Background:
- Graphical models are crucial for brain connectivity analysis in neuroscience.
- Analyzing heterogeneous group data with unknown group membership presents significant challenges for graph structure learning.
- Functional magnetic resonance imaging (fMRI) data often involves such heterogeneous observations.
Purpose of the Study:
- To propose a method for Simultaneous Clustering and Estimation of Heterogeneous Graphs (SCEHG) for matrix-variate fMRI data.
- To address the challenge of learning graph structures from heterogeneous group data without prior group membership information.
- To develop a novel approach that leverages group differences in conditional dependencies for cluster structure learning.
Main Methods:
- The SCEHG method formulates clustering as penalized regression with grouping and sparsity pursuit, transforming unsupervised learning into supervised learning.
- Individual-level between-region network measures are constructed to capture group differences.
- A modified difference of convex programming with the alternating direction method of multipliers (DC-ADMM) algorithm is employed for optimization.
- A generalized criterion is proposed for specifying the number of clusters.
Main Results:
- Extensive simulations demonstrate the superiority of the SCEHG method over existing state-of-the-art approaches.
- The method achieves higher accuracy in both clustering and graph recovery.
- Application to fMRI data for attention-deficit hyperactivity disorder (ADHD) showcases its practical utility.
Conclusions:
- The SCEHG method offers a robust framework for analyzing heterogeneous fMRI data in neuroscience.
- It effectively addresses challenges in graph structure learning and clustering when group memberships are unknown.
- The approach demonstrates significant improvements in accuracy and empirical usefulness for complex neuroimaging studies.
