Nerrise Favour, Zhao Qingyu, Poston Kathleen L, Pohl Kilian M, Adeli Ehsan
Department of Electrical Engineering, Stanford University, Stanford, CA, USA.
Department of Psychiatry and Behavioral Sciences, Stanford University, Stanford, CA, USA.
Med Image Comput Comput Assist Interv. 2023 Oct;14221:723-733. doi: 10.1007/978-3-031-43895-0_68. Epub 2023 Oct 1.
One of the hallmark symptoms of Parkinson's Disease (PD) is the progressive loss of postural reflexes, which eventually leads to gait difficulties and balance problems. Identifying disruptions in brain function associated with gait impairment could be crucial in better understanding PD motor progression, thus advancing the development of more effective and personalized therapeutics. In this work, we present an explainable, geometric, weighted-graph attention neural network () to identify functional networks predictive of the progression of gait difficulties in individuals with PD. predicts the multi-class gait impairment on the MDS-Unified PD Rating Scale (MDS-UPDRS). Our computational- and data-efficient model represents functional connectomes as symmetric positive definite (SPD) matrices on a Riemannian manifold to explicitly encode pairwise interactions of entire connectomes, based on which we learn an attention mask yielding individual- and group-level explainability. Applied to our resting-state functional MRI (rs-fMRI) dataset of individuals with PD, identifies functional connectivity patterns associated with gait impairment in PD and offers interpretable explanations of functional subnetworks associated with motor impairment. Our model successfully outperforms several existing methods while simultaneously revealing clinically-relevant connectivity patterns. The source code is available at https://github.com/favour-nerrise/xGW-GAT.
帕金森病（PD）的标志性症状之一是姿势反射的逐渐丧失，这最终会导致步态困难和平衡问题。识别与步态障碍相关的脑功能破坏对于更好地理解PD运动进展至关重要，从而推动更有效和个性化治疗方法的发展。在这项工作中，我们提出了一种可解释的、几何的、加权图注意力神经网络（），以识别预测PD患者步态困难进展的功能网络。该网络预测多类步态障碍的MDS统一PD评定量表（MDS-UPDRS）。我们的计算高效且数据高效的模型将功能连接组表示为黎曼流形上的对称正定（SPD）矩阵，以明确编码整个连接组的成对相互作用，在此基础上，我们学习一个注意力掩码，从而实现个体和群体层面的可解释性。应用于我们的PD患者静息态功能磁共振成像（rs-fMRI）数据集时，该网络识别出与PD步态障碍相关的功能连接模式，并提供与运动障碍相关的功能子网的可解释性解释。我们的模型成功超越了几种现有方法，同时揭示了临床相关的连接模式。源代码可在https://github.com/favour-nerrise/xGW-GAT获取。
