Abstract:Finding an appropriate representation of dynamic activities in the brain is crucial for many downstream applications. Due to its highly dynamic nature, temporally averaged fMRI (functional magnetic resonance imaging) can only provide a narrow view of underlying brain activities. Previous works lack the ability to learn and interpret the latent dynamics in brain architectures. This paper builds an efficient graph neural network model that incorporates both region-mapped fMRI sequences and structural connectivities obtained from DWI (diffusion-weighted imaging) as inputs. We find good representations of the latent brain dynamics through learning sample-level adaptive adjacency matrices and performing a novel multi-resolution inner cluster smoothing. We also attribute inputs with integrated gradients, which enables us to infer (1) highly involved brain connections and subnetworks for each task, (2) temporal keyframes of imaging sequences that characterize tasks, and (3) subnetworks that discriminate between individual subjects. This ability to identify critical subnetworks that characterize signal states across heterogeneous tasks and individuals is of great importance to neuroscience and other scientific domains. Extensive experiments and ablation studies demonstrate our proposed method's superiority and efficiency in spatial-temporal graph signal modeling with insightful interpretations of brain dynamics.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to more accurately capture and interpret the temporal variation characteristics of brain signals in the dynamic representation of brain activity. Specifically: 1. **Representation of Temporal - Variation Signals**: Traditional fMRI (functional magnetic resonance imaging) methods usually use time - averaging methods to process brain signals. This method can only provide a narrow view of brain activity and cannot fully capture the dynamic changes of brain activity. The paper proposes a new graph neural network model (ReBraiD), which can simultaneously consider the fMRI sequences of regional mapping and the structural connectivity obtained through DWI (diffusion - weighted imaging), thereby better representing the dynamic activities of the brain. 2. **Learning and Interpretation of Dynamic Structures**: Existing methods have deficiencies in learning the underlying dynamic structures of brain activity, especially in interpreting these dynamic structures. Through learning sample - level adaptive adjacency matrices and performing multi - resolution internal clustering smoothing, ReBraiD can not only find better representations of underlying brain dynamics, but also, through integrated gradient attribution techniques, identify highly involved brain connections and sub - networks in each task, key time frames of image sequences characterizing tasks, and sub - networks that distinguish individual subjects. 3. **Generality across Tasks and Individuals**: The paper emphasizes the ability of its method to identify key sub - networks across different tasks and individuals, which is of great significance for research in neuroscience and other scientific fields. Through extensive experiments and ablation studies, the paper demonstrates the superiority and efficiency of the proposed method in spatio - temporal graph signal modeling and provides profound insights into brain dynamics. In conclusion, this paper aims to solve the limitations of existing methods in capturing and interpreting the dynamic changes of brain activity by proposing a new graph neural network model, providing a more powerful tool for neuroscience research.

Deep Representations for Time-varying Brain Datasets

DSAM: A Deep Learning Framework for Analyzing Temporal and Spatial Dynamics in Brain Networks

Spatio-temporal Modeling of Connectome-Scale Brain Network Interactions Via Time-Evolving Graphs

Dynamic Adaptive Spatio-temporal Graph Convolution for fMRI Modelling

BrainTGL: A dynamic graph representation learning model for brain network analysis

A Deep Graph Neural Network Architecture for Modelling Spatio-temporal Dynamics in resting-state functional MRI Data

A Deep Probabilistic Spatiotemporal Framework for Dynamic Graph Representation Learning with Application to Brain Disorder Identification

Functional2Structural: Cross-Modality Brain Networks Representation Learning

Inferring Brain Dynamics via Multimodal Joint Graph Representation EEG-fMRI

Towards a predictive spatio-temporal representation of brain data

DeepGraphDMD: Interpretable Spatio-Temporal Decomposition of Non-linear Functional Brain Network Dynamics

Deep Representation Learning For Multimodal Brain Networks

Encoding Multi-Resolution Brain Networks Using Unsupervised Deep Learning

DynDepNet: Learning Time-Varying Dependency Structures from fMRI Data via Dynamic Graph Structure Learning

A Hybrid Spatio-Temporal Deep Belief Network and Sparse Representation-Based Framework Reveals Multi-Level Core Functional Components in Decoding Multi-Task fMRI Signals

Learning low-dimensional dynamics from whole-brain data improves task capture

Identifying Brain Networks at Multiple Time Scales via Deep Recurrent Neural Network

Interpretable Spatio-Temporal Embedding for Brain Structural-Effective Network with Ordinary Differential Equation

Recognizing Brain States Using Deep Sparse Recurrent Neural Network

Generalized Recurrent Neural Network accommodating Dynamic Causal Modeling for functional MRI analysis

Sparse Deep Neural Network for Encoding and Decoding the Structural Connectome