Multi-hops functional connectivity improves individual prediction of fusiform face activation via a graph neural network

Wu, Dongya; Li, Xin; Feng, Jun

doi:10.1101/2020.09.21.305839

Cited by 3 publications

(4 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Later on, (72) leveraged siamese graph convolutional neural network (s-GCN) to learn the similarity between a pair of graphs and incorporated the learned similarity into the classification step. From another perspective, (73) assume that detecting the brain disease state is relative to study the brain region's function that can be represented by its multi-hops connectivity profile. Therefore, a GNN model taking as input the functional brain graph learns how many levels of nearest neighbors of a specific ROI that need to be considered in a brain graph classification task.…”

Section: B-graph-based Classificationmentioning

confidence: 99%

Graph Neural Networks in Network Neuroscience

Bessadok¹,

Mahjoub²,

Rekik³

2021

Preprint

View full text Add to dashboard Cite

Noninvasive medical neuroimaging has yielded many discoveries about the brain connectivity. Several substantial techniques mapping morphological, structural and functional brain connectivities were developed to create a comprehensive road map of neuronal activities in the human brain -namely brain graph. Relying on its non-Euclidean data type, graph neural network (GNN) provides a clever way of learning the deep graph structure and it is rapidly becoming the state-of-the-art leading to enhanced performance in various network neuroscience tasks. Here we review current GNN-based methods, highlighting the ways that they have been used in several applications related to brain graphs such as missing brain graph synthesis and disease classification. We conclude by charting a path toward a better application of GNN models in network neuroscience field for neurological disorder diagnosis and population graph integration.

show abstract

Section: B-graph-based Classificationmentioning

confidence: 99%

Graph Neural Networks in Network Neuroscience

Bessadok¹,

Mahjoub²,

Rekik³

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…In order to deal with the curse of dimensionality, different statistical models have been proposed with certain low-dimensional structures, such as sparse linear regression, low-rank matrix regression and so on. Specifically, the multi-response regression model, which is an important instance of matrix regression, has been deeply investigated in theoretical aspects [3,4] and widely used in real applications such as neuroimage analysis [5,6]. Consider the following multi-response regression model…”

Section: Introductionmentioning

confidence: 99%

“…Unfortunately, the low-rankness of a matrix is rather different from the sparsity of a vector due to the more sophisticated manifold structure [23]. Moreover, the multivariate nature of the responses enables one to build more complex models for modern large-scale association analysis, such as fMRI image analyses [6] and physiological network analyses [24], and thus has a substantial wider application than that of the univariate model.…”

Section: Introductionmentioning

confidence: 99%

Minimax Lower Bounds for High-Dimensional Multi-Response Errors-in-Variables Regression

Li¹,

Wu²

2022

Preprint

Self Cite

View full text Add to dashboard Cite

Noisy data is always encountered in real applications, such as bioinformatics, neuroimage and remote sensing. Existing methods mainly consider linear or generalized linear errors-in-variables regression, while relatively little attention is paid for the multivariate response case, and how to evaluate the estimation performance under perturbed covariates is still an open question. In this paper, we consider the information-theoretic limitations of estimating a low-rank matrix in the multi-response errors-in-variables regression model. By application of the information theory and statistical techniques on concentration inequalities, the minimax lower bound is provided in terms of the squared Frobenius loss, which recaptures the rate provided under the clean covariate assumption in previous literatures. Hence our result further indicates that though under the more realistic errors-in-variables situation, no more samples are required so as to achieve a rate-optimal estimation.

show abstract

“…In addition, these applications are based on the node-driven GNNs with parameters concentrating on the transformation of node features, a strategy that is not specialized for utilizing topological properties of the brain connectome. Edge-driven GNN as used by Wu et al (2021) is needed to explore topological structures of the brain connectome.…”

Section: Introductionmentioning

confidence: 99%

Connectome-based individual prediction of cognitive behaviors via the graph propagation network reveals directed brain network topology

Feng

2021

Preprint

Self Cite

View full text Add to dashboard Cite

The brain connectome supports the information flow underlying human cognitions and should reflect the individual variability in human cognitive behaviors. Various studies have utilized the brain connectome to predict individual differences in human behaviors. However, traditional studies viewed the brain connectome feature as a vector of one dimension, a method which neglects topological structures of the brain connectome. To utilize topological properties of the brain connectome, we proposed that graph neural network which combines graph theory and neural network can be adopted. Different from previous node-driven graph neural networks that parameterize on the node feature transformation, we designed an edge-driven graph neural network named graph propagation network that parameterizes on the information propagation within the brain connectome. We compared various models in predicting the individual total cognition based on the resting-state functional connectome. The edge-driven graph propagation network showed the highest prediction accuracy and outperformed the node-driven graph neural network and traditional partial least square regression. The graph propagation network also revealed a directed network topology encoding the information flow, indicating that the high-level association cortices are responsible for the information integration underlying the total cognition. These results suggest that the edge-driven graph propagation network can better explore the topological structure of the brain connectome and can serve as a new method to associate the brain connectome and human behaviors.

show abstract

Multi-hops functional connectivity improves individual prediction of fusiform face activation via a graph neural network

Cited by 3 publications

References 42 publications

Graph Neural Networks in Network Neuroscience

Graph Neural Networks in Network Neuroscience

Minimax Lower Bounds for High-Dimensional Multi-Response Errors-in-Variables Regression

Connectome-based individual prediction of cognitive behaviors via the graph propagation network reveals directed brain network topology

Contact Info

Product

Resources

About