Modeling Dynamic Functional Information Flows on Large-Scale Brain Networks

Peili Lv, Lei Guo, Xintao Hu, Xiang Li, Changfeng Jin, Junwei Han, Lingjiang Li, Tianming LiuMICCAI 2013 · OpenReview

Abstract

Growing evidence from the functional neuroimaging field suggests that human brain functions are realized via dynamic functional interactions on large-scale structural networks. Even in resting state, functional brain networks exhibit remarkable temporal dynamics. However, it has been rarely explored to computationally model such dynamic functional information flows on large-scale brain networks. In this paper, we present a novel computational framework to explore this problem using multimodal resting state fMRI (R-fMRI) and diffusion tensor imaging (DTI) data. Based on the large-scale structural brain networks constructed from DTI data, we employ a dynamic programming strategy to infer functional information transition routines on structural networks, based on which hub routers that most frequently participate in these routines are identified. Our findings reveal that the Default Mode Network (DMN) acts as a critical functional hub in resting state, while patients with post-traumatic stress disorder (PTSD) exhibit significant differences in hub router distributions.

Introduction

While the brain's structural connectivity (the "hardware") is well-mapped, understanding how functional signals (the "software") traverse these pathways remains a significant challenge. Most neuroimaging studies focus on "static" connectivity, averaging activity over several minutes. However, brain activity is highly dynamic, shifting through various "quasi-stable states" even at rest. This paper addresses the challenge of modeling these transitions: How does a signal move between brain regions, and which areas act as the "routers" facilitating this traffic? By analogizing the brain to a telecommunications network, we can decipher how information is processed and how disorders like PTSD disrupt these flows.

Computational Framework

The proposed framework integrates structural (DTI) and functional (R-fMRI) data through a multi-step pipeline designed to trace information "packets" across the cortex.

  1. Landmark Localization: Using the DICCCOL system, 358 consistent cortical landmarks are identified across subjects to provide a common reference frame.
  2. Dynamic State Extraction: A sliding-window approach combined with sparse coding parcellates the fMRI timeline into distinct quasi-stable functional states.
  3. Sub-network Clustering and Peak Detection: Within each state, landmarks are grouped into sub-networks. Cubic polynomial curve fitting identifies the temporal peak of activity for each sub-network, establishing a temporal order for information flow.
  4. Routine Tracing: A dynamic programming strategy, similar to the Viterbi algorithm used in telecommunications, infers the optimal transition routines between sub-networks by maximizing functional connectivity strength along the path.
Overview of the three-step computational framework for modeling information flow.

Key Insights and Results

The most significant conceptual contribution is the identification of "hub routers" — nodes that most frequently participate in the inferred transition routines.

The DMN as a Central Hub

The study found that a majority of hub routers are located within the Default Mode Network (DMN), particularly the posterior cingulate cortices. This suggests that the DMN’s role at rest extends beyond internal reflection; it acts as a primary traffic controller for information flow across the entire brain. Comparison with group-wise Independent Component Analysis (ICA) confirmed this spatial overlap.

PTSD and "Hyper-routing"

Application of this framework to a PTSD dataset revealed distinct routing abnormalities. In PTSD patients, hub routers were significantly more widespread and "hyper-active," particularly in anterior frontal regions. This suggests that trauma may force the brain to utilize non-standard, potentially less efficient routing pathways, aligning with existing literature on frontal hyper-activity in PTSD neural circuits.

Router frequency maps comparing healthy controls and PTSD patients, highlighting differences in anterior regions.

Conclusion

This novel framework successfully models the brain as a dynamic routing network. By identifying hub routers, we gain a clearer picture of the DMN's functional importance and discover potential biomarkers for psychiatric conditions. Future work will focus on refining window size sensitivity and integrating more direct measures of effective connectivity to further validate these "router signatures" for clinical use.

Summary updated

Abstract

Growing evidence from the functional neuroimaging field suggests that human brain functions are realized via dynamic functional interactions on large-scale structural networks. Even in resting state, functional brain networks exhibit remarkable temporal dynamics. However, it has been rarely explored to computationally model such dynamic functional information flows on large-scale brain networks. In this paper, we present a novel computational framework to explore this problem using multimodal resting state fMRI (R-fMRI) and diffusion tensor imaging (DTI) data. Basically, recent literature reports including our own studies have demonstrated that the resting state brain networks dynamically undergo a set of distinct brain states. Within each quasi-stable state, functional information flows from one set of structural brain nodes to other sets of nodes, which is analogous to the message package routing on the Internet from the source node to the destination. Therefore, based on the large-scale structural brain networks constructed from DTI data, we employ a dynamic programming strategy to infer functional information transition routines on structural networks, based on which hub routers that most frequently participate in these routines are identified. It is interesting that a majority of those hub routers are located within the default mode network (DMN), revealing a possible mechanism of the critical functional hub roles played by the DMN in resting state. Also, application of this framework on a post trauma stress disorder (PTSD) dataset demonstrated interesting difference in hub router distributions between PTSD patients and healthy controls.
BibTeX
@inproceedings{lv2013modeling,
  author = {Peili Lv and Lei Guo and Xintao Hu and Xiang Li and Changfeng Jin and Junwei Han and Lingjiang Li and Tianming Liu},
  title = {{Modeling Dynamic Functional Information Flows on Large-Scale Brain Networks}},
  booktitle = {International Conference on Medical Image Computing and Computer-Assisted Intervention},
  year = {2013},
  url = {https://openreview.net/forum?id=nEvG7lIxof7},
}
Figure