Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

HyperBrain: Anomaly Detection for Temporal Hypergraph Brain Networks

Sadaf Sadeghian, Xiaoxiao Li, Margo Seltzer

TL;DR

HyperBrain is presented, an unsupervised anomaly detection framework for temporal hypergraph brain networks that models fMRI time series data as temporal hypergraphs capturing dynamic higher-order interactions and outperforms all other baselines on detecting abnormal co-activations in brain networks.

Abstract

Identifying unusual brain activity is a crucial task in neuroscience research, as it aids in the early detection of brain disorders. It is common to represent brain networks as graphs, and researchers have developed various graph-based machine learning methods for analyzing them. However, the majority of existing graph learning tools for the brain face a combination of the following three key limitations. First, they focus only on pairwise correlations between regions of the brain, limiting their ability to capture synchronized activity among larger groups of regions. Second, they model the brain network as a static network, overlooking the temporal changes in the brain. Third, most are designed only for classifying brain networks as healthy or disordered, lacking the ability to identify abnormal brain activity patterns linked to biomarkers associated with disorders. To address these issues, we present HyperBrain, an unsupervised anomaly detection framework for temporal hypergraph brain networks. HyperBrain models fMRI time series data as temporal hypergraphs capturing dynamic higher-order interactions. It then uses a novel customized temporal walk (BrainWalk) and neural encodings to detect abnormal co-activations among brain regions. We evaluate the performance of HyperBrain in both synthetic and real-world settings for Autism Spectrum Disorder and Attention Deficit Hyperactivity Disorder(ADHD). HyperBrain outperforms all other baselines on detecting abnormal co-activations in brain networks. Furthermore, results obtained from HyperBrain are consistent with clinical research on these brain disorders. Our findings suggest that learning temporal and higher-order connections in the brain provides a promising approach to uncover intricate connectivity patterns in brain networks, offering improved diagnosis.

HyperBrain: Anomaly Detection for Temporal Hypergraph Brain Networks

TL;DR

HyperBrain is presented, an unsupervised anomaly detection framework for temporal hypergraph brain networks that models fMRI time series data as temporal hypergraphs capturing dynamic higher-order interactions and outperforms all other baselines on detecting abnormal co-activations in brain networks.

Abstract

Identifying unusual brain activity is a crucial task in neuroscience research, as it aids in the early detection of brain disorders. It is common to represent brain networks as graphs, and researchers have developed various graph-based machine learning methods for analyzing them. However, the majority of existing graph learning tools for the brain face a combination of the following three key limitations. First, they focus only on pairwise correlations between regions of the brain, limiting their ability to capture synchronized activity among larger groups of regions. Second, they model the brain network as a static network, overlooking the temporal changes in the brain. Third, most are designed only for classifying brain networks as healthy or disordered, lacking the ability to identify abnormal brain activity patterns linked to biomarkers associated with disorders. To address these issues, we present HyperBrain, an unsupervised anomaly detection framework for temporal hypergraph brain networks. HyperBrain models fMRI time series data as temporal hypergraphs capturing dynamic higher-order interactions. It then uses a novel customized temporal walk (BrainWalk) and neural encodings to detect abnormal co-activations among brain regions. We evaluate the performance of HyperBrain in both synthetic and real-world settings for Autism Spectrum Disorder and Attention Deficit Hyperactivity Disorder(ADHD). HyperBrain outperforms all other baselines on detecting abnormal co-activations in brain networks. Furthermore, results obtained from HyperBrain are consistent with clinical research on these brain disorders. Our findings suggest that learning temporal and higher-order connections in the brain provides a promising approach to uncover intricate connectivity patterns in brain networks, offering improved diagnosis.
Paper Structure (13 sections, 3 equations, 3 figures, 1 table)

This paper contains 13 sections, 3 equations, 3 figures, 1 table.

Table of Contents

  1. Introduction
  2. The Proposed Framework
  3. Modeling fMRI Data as Temporal Hypergraph Brain Network
  4. Brain Walk Sampling
  5. Neural Hyperedge Anomaly Detection
  6. Neural Walk Encoding
  7. Training
  8. Evaluation
  9. Datasets and Baselines
  10. Quantitative Evaluation on Synthetic Anomalous Hyperedges
  11. Clinical Relevance and Consistency on Real Datasets
  12. Conclusion
  13. Acknowledgments and Disclosure of Funding

Figures (3)

  • Figure 1: Schematic of HyperBrain. HyperBrain consists of four stages: (1) Modelling both temporal and higher-order interactions among brain regions [§\ref{['sec:Modeling']}], (2) Extracting temporal, higher-order patterns of brain activity [§\ref{['sec:Brainwalk']}] , (3) Neural encoding to merge information from the sequence of hyperedges and their timestamps in the extracted brain patterns [§\ref{['sec:neuralWalkEncode']}], and (4) Calculating anomaly scores for each brain co-activation [§\ref{['sec:anomalyScore']}]. In training HyperBrain, we only rely on healthy control data to detect anomalous co-activations in the brain [§\ref{['sec:hb_training']}] .
  • Figure 2: ADHD-Related Brain Regions Identified by HyperBrain
  • Figure 3: ASD-Related Brain Regions Identified by HyperBrain

Theorems & Definitions (1)

  • Definition 1: Temporal Hypergraph