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.

翻译：近期，图神经网络（GNNs）在从功能磁共振成像（fMRI）数据中学习脑图表示方面取得了成功。然而，现有的大多数GNN方法假设脑图随时间静态不变，且图邻接矩阵在模型训练前已知。这些假设与神经科学的证据相矛盾——脑图是时变的，其连接结构取决于功能连接度量方式的选择。不能真实反映底层fMRI数据的噪声脑图会对GNN的性能产生不利影响。为此，我们提出动态脑图结构学习（DBGSL），一种新颖的方法，通过学习由下游预测任务驱动的fMRI数据的最优时变依赖结构。实验表明，DBGSL在使用真实静息态和任务态fMRI数据进行性别分类时达到了最先进的性能。此外，对学习到的动态图的分析突显了与预测相关的脑区，这与现有神经科学文献一致。