Related Experiment Video
Updated: Jun 17, 2026
05:55
Modeling the Functional Network for Spatial Navigation in the Human Brain
Published on: October 13, 2023
Accurately modeling resting-brain functional connectivity using hypergraph neural field-Fourier deep neural network.
Jichao Ma1, Jiebin Luo2, Dandan Liu3
1School of Intelligent Manufacturing and Materials, Qingdao Binhai University, Qingdao, 266555, China. majichao@qdbhu.edu.cn.
Scientific Reports
|June 15, 2026
Summary
This study introduces a new hypergraph neural field model to better predict brain functional connectivity from structural data. The advanced HNF-FDNN model significantly improves prediction accuracy, aiding in understanding cognition and disorders.
Area of Science:
- Neuroscience
- Computational Biology
- Machine Learning
Background:
- Understanding the link between brain structure and function is crucial for cognitive science and neuropsychiatric disorder research.
- Existing graph diffusion models predict functional connectivity (FC) from structural connectivity (SC) but ignore inter-regional information interactions, failing to capture negative correlations.
Purpose of the Study:
- To develop a novel model that accurately predicts human brain functional connectivity (FC) from structural connectivity (SC).
- To incorporate information interactions between excitatory and inhibitory neurons for improved FC prediction.
- To enhance prediction accuracy and robustness using a Fourier deep neural network.
Main Methods:
- Established a hypergraph neural field (HNF) model to represent information interactions between neurons in different brain regions.
- Calculated interactive connectivity (IC) using Pearson correlation coefficients from excitatory membrane potentials.
- Developed a hypergraph neural field-Fourier deep neural network (HNF-FDNN) integrating spectral information for enhanced FC representation.
Main Results:
- The HNF model captured negative correlations in FC, unlike previous methods.
- The HNF-FDNN model achieved a mean Pearson correlation coefficient of 0.8168 on the Human Connectome Project dataset, significantly outperforming the graph diffusion model (0.5499).
- The HNF-FDNN demonstrated superior robustness and stability in predicting functional connectivity.
Conclusions:
- The HNF-FDNN model offers a significant advancement in predicting brain functional connectivity from structural data.
- Combining brain activity signals with machine learning holds great potential for modeling complex brain functions.
- This approach can advance our understanding of cognitive processes and the mechanisms underlying neuropsychiatric disorders.
