DBGSL: Dynamic Brain Graph Structure Learning

Alexander Campbell, Antonio Giuliano Zippo, Luca Passamonti, Nicola Toschi, Pietro Lio
Medical Imaging with Deep Learning, PMLR 227:1318-1345, 2024.

Abstract

Recently, graph neural networks (GNNs) have shown success at learning representations of brain graphs derived from functional magnetic resonance imaging (fMRI) data. The majority of existing GNN methods, however, assume brain graphs are static over time and the graph adjacency matrix is known prior to model training. These assumptions are at odds with neuroscientific evidence that brain graphs are time-varying with a connectivity structure that depends on the choice of functional connectivity measure. Noisy brain graphs that do not truly represent the underling fMRI data can have a detrimental impact on the performance of GNNs. As a solution, we propose Dynamic Brain Graph Structure Learning (DBGSL), a novel method for learning the optimal time-varying dependency structure of fMRI data induced by a downstream prediction task. Experiments demonstrate DBGSL achieves state-of-the-art performance for sex classification using real-world resting-state and task fMRI data. Moreover, analysis of the learnt dynamic graphs highlights prediction-related brain regions which align with existing neuroscience literature.

Cite this Paper

BibTeX
@InProceedings{pmlr-v227-campbell24a, title = {DBGSL: Dynamic Brain Graph Structure Learning}, author = {Campbell, Alexander and Zippo, Antonio Giuliano and Passamonti, Luca and Toschi, Nicola and Lio, Pietro}, booktitle = {Medical Imaging with Deep Learning}, pages = {1318--1345}, year = {2024}, editor = {Oguz, Ipek and Noble, Jack and Li, Xiaoxiao and Styner, Martin and Baumgartner, Christian and Rusu, Mirabela and Heinmann, Tobias and Kontos, Despina and Landman, Bennett and Dawant, Benoit}, volume = {227}, series = {Proceedings of Machine Learning Research}, month = {10--12 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v227/campbell24a/campbell24a.pdf}, url = {https://proceedings.mlr.press/v227/campbell24a.html}, abstract = {Recently, graph neural networks (GNNs) have shown success at learning representations of brain graphs derived from functional magnetic resonance imaging (fMRI) data. The majority of existing GNN methods, however, assume brain graphs are static over time and the graph adjacency matrix is known prior to model training. These assumptions are at odds with neuroscientific evidence that brain graphs are time-varying with a connectivity structure that depends on the choice of functional connectivity measure. Noisy brain graphs that do not truly represent the underling fMRI data can have a detrimental impact on the performance of GNNs. As a solution, we propose Dynamic Brain Graph Structure Learning (DBGSL), a novel method for learning the optimal time-varying dependency structure of fMRI data induced by a downstream prediction task. Experiments demonstrate DBGSL achieves state-of-the-art performance for sex classification using real-world resting-state and task fMRI data. Moreover, analysis of the learnt dynamic graphs highlights prediction-related brain regions which align with existing neuroscience literature.} }
Endnote
%0 Conference Paper %T DBGSL: Dynamic Brain Graph Structure Learning %A Alexander Campbell %A Antonio Giuliano Zippo %A Luca Passamonti %A Nicola Toschi %A Pietro Lio %B Medical Imaging with Deep Learning %C Proceedings of Machine Learning Research %D 2024 %E Ipek Oguz %E Jack Noble %E Xiaoxiao Li %E Martin Styner %E Christian Baumgartner %E Mirabela Rusu %E Tobias Heinmann %E Despina Kontos %E Bennett Landman %E Benoit Dawant %F pmlr-v227-campbell24a %I PMLR %P 1318--1345 %U https://proceedings.mlr.press/v227/campbell24a.html %V 227 %X Recently, graph neural networks (GNNs) have shown success at learning representations of brain graphs derived from functional magnetic resonance imaging (fMRI) data. The majority of existing GNN methods, however, assume brain graphs are static over time and the graph adjacency matrix is known prior to model training. These assumptions are at odds with neuroscientific evidence that brain graphs are time-varying with a connectivity structure that depends on the choice of functional connectivity measure. Noisy brain graphs that do not truly represent the underling fMRI data can have a detrimental impact on the performance of GNNs. As a solution, we propose Dynamic Brain Graph Structure Learning (DBGSL), a novel method for learning the optimal time-varying dependency structure of fMRI data induced by a downstream prediction task. Experiments demonstrate DBGSL achieves state-of-the-art performance for sex classification using real-world resting-state and task fMRI data. Moreover, analysis of the learnt dynamic graphs highlights prediction-related brain regions which align with existing neuroscience literature.
APA
Campbell, A., Zippo, A.G., Passamonti, L., Toschi, N. & Lio, P.. (2024). DBGSL: Dynamic Brain Graph Structure Learning. Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 227:1318-1345 Available from https://proceedings.mlr.press/v227/campbell24a.html.

Related Material