Interpretable Graph Neural Networks for Connectome-Based Brain Disorder Analysis

@inproceedings{Cui2022InterpretableGN,
  title={Interpretable Graph Neural Networks for Connectome-Based Brain Disorder Analysis},
  author={Hejie Cui and Wei Dai and Yanqiao Zhu and Xiaoxiao Li and Lifang He and Carl Yang},
  booktitle={MICCAI},
  year={2022}
}
. Human brains lie at the core of complex neurobiological systems, where the neurons, circuits, and subsystems interact in enigmatic ways. Understanding the structural and functional mechanisms of the brain has long been an intriguing pursuit for neuroscience research and clinical disorder therapy. Mapping the connections of the human brain as a network is one of the most pervasive paradigms in neuroscience. Graph Neural Networks (GNNs) have recently emerged as a potential method for modeling… 
1 Citations

Figures and Tables from this paper

On Positional and Structural Node Features for Graph Neural Networks on Non-attributed Graphs

TLDR
A practical guideline on the choices between different artificial node features for GNNs on non-attributed graphs, and insights on why each is more appropriate for certain tasks are provided.

References

SHOWING 1-10 OF 43 REFERENCES

Joint Embedding of Structural and Functional Brain Networks with Graph Neural Networks for Mental Illness Diagnosis

TLDR
This work develops a novel multiview GNN model which takes advantage of the message passing scheme by propagating messages based on degree statistics and brain region connectivities and employs contrastive learning for multimodal fusion.

BrainGB: A Benchmark for Brain Network Analysis with Graph Neural Networks

TLDR
BrainGB is presented, a benchmark for brain network analysis with GNNs and standardizes the process by summarizing brain network construction pipelines for both functional and structural neuroimaging modalities and modularizing the implementation of GNN designs.

Brain Network Transformer

TLDR
This work proposes an O RTHONORMAL C LUSTERING R EADOUT operation based on self-supervised soft clustering and orthonormal projection, and re-standardize the evaluation pipeline on the only one publicly available large-scale brain network dataset of ABIDE, to enable meaningful comparison of different models.

Data-Efficient Brain Connectome Analysis via Multi-Task Meta-Learning

TLDR
This paper proposes to meta-train the model on datasets of large sample sizes and transfer the knowledge to small datasets, and also explores brain-network-oriented designs, including atlas transformation and adaptive task reweighing.

FBNETGEN: Task-aware GNN-based fMRI Analysis via Functional Brain Network Generation

TLDR
FBNETGEN is developed, a task-aware and interpretable fMRI analysis framework via deep brain network generation which learns to transform raw time-series features into task-oriented brain networks and provides unique interpretations by highlighting prediction-related brain regions.

BrainNet Viewer: A Network Visualization Tool for Human Brain Connectomics

TLDR
This work has developed a graph-theoretical network visualization toolbox, called BrainNet Viewer, to illustrate human connectomes as ball-and-stick models, and helps researchers to visualize brain networks in an easy, flexible and quick manner.

Sub-network Based Kernels for Brain Network Classification

TLDR
A novel sub-network based kernel for brain networks is constructed and applied for mild cognitive impairment (MCI) classification and results demonstrate that the proposed method outperform several state-of-the-art graph kernel based methods.

Multi-View Multi-Graph Embedding for Brain Network Clustering Analysis

TLDR
This work proposes Multi-view Multi-graph Embedding M2E by stacking multi-graphs into multiple partially-symmetric tensors and using tensor techniques to simultaneously leverage the dependencies and correlations among multi-view and multi- graph brain networks.

Stability of graph theoretical measures in structural brain networks in Alzheimer’s disease

TLDR
It is suggested that cortical thickness measures should be preferred in future AD studies if using binary networks and studying cortical networks in small cohorts should be complemented by analyzing smaller, subsampled groups to reduce the risk that findings are spurious.