功能性消化不良患者脑网络的模块结构发现方法

Group Structure Analysis of Intrinsic Functional Brain Connectivity in Functional Dyspepsia

现有研究表明功能性消化不良(FD)疾病与大脑功能网络的异常拓扑模块结构有关。脑网络并不是静态固化的,而人工识别网络结构的成本很高。因此,数据驱动的自动识别方法具有重要的研究意义。然而,传统的静态识别模型很难有效挖掘动态网络中的功能模块结构。为了解决这个问题,本文从网络演化的角度提出新的识别方法。我们首先引入了时间平滑性来估算网络的动态演化过程,并结合节点的拓扑结构与属性相似度来缓解数据稀疏性问题。这样可以有效地减少噪音的影响,提高识别的区分度,从而提供更准确和可靠的结果。在多个数据集上的实验显示了该模型在网络结构分析方面的有效性。发现FD患者脑网络中的功能模块有助于下一步探索FD与异常模块之间关联关联,进而揭示FD的病理机制,为早期诊断和治疗评估提供判别依据。

Existing studies have found that functional dyspepsia(FD)is related to the abnormal topological module structure of the brain's functional network.Such a brain network is not static and fixed.It is high-cost to manually identify the network structure.Thus,the data-driven automatic recognition method has great research value.However,it is hard for traditional static models to effectively mine the functional module structure from the dynamic network.To solve this problem,we propose a new identification method from the view of network evolution.We introduce time smoothness to estimate the dynamic evolution process of the network.In addition,we incorporate the topology of nodes and attribute similarity to alleviate the data sparsity problem.That can effectively reduce the impact of noise to improve the discrimination ability of the model,which can boost performance.Experiments on multiple data sets show the model's effectiveness in analyzing network structure.The discovery of functional modules in the brain network of patients with FD is helpful for the next step to explore the relations between FD and abnormal modules.That can facilitate revealing the FD pathological mechanism which provides a discriminant basis for early diagnosis and treatment evaluation.