Background. Graph-theoretic approaches are increasingly popular for identifying the patterns of disrupted neural systems after traumatic brain injury (TBI). However, the patterns of neuroplasticity in brain organization after cognitive training in TBI are less well understood. Objective. We identified the patterns of training-induced neuroplasticity of the whole-brain network in TBI, using resting-state functional connectivity and graph theory. Methods. A total of 64 civilians and veterans with TBI were randomized into either a strategy-based cognitive training group (n = 33) or a knowledge-based training group (active control group; n = 31) for 8 weeks. The participants experienced mild to severe TBI without focal damage and persistent cognitive dysfunctions. A subset of participants complained of subclinical but residual psychiatric symptoms. We acquired their resting-state functional magnetic resonance imaging before training, immediately posttraining, and 3 months posttraining. From participants’ resting-state networks, we obtained the modularity, participation coefficient, within-module connectivity, global efficiency, and local efficiency over multiple network densities. We next performed longitudinal analyses on those measures corrected for multiple comparisons across network densities using false discovery rate (FDR). Results. Relative to the knowledge-based training group, the strategy-based cognitive training group had reduced modularity and increased participation coefficient, global efficiency, and local efficiency over time (Pnodal < .05; qFDR < 0.05). Brain behavior analysis revealed that the participation coefficient and global efficiency within the strategy-based cognitive training group correlated with trail-making scores in the context of training (Pnodal < .05; qFDR < 0.05). Conclusions. Cognitive training reorganized modular networks in TBI over the whole brain. Graph-theoretic approaches may be useful in identifying a potential brain-based marker of training efficacy in TBI.

中文翻译：

---

认知训练重组创伤性脑损伤中的网络模块

背景。图论方法越来越流行用于识别创伤性脑损伤 (TBI) 后神经系统中断的模式。然而，TBI 认知训练后大脑组织的神经可塑性模式还不太清楚。客观的。我们使用静息状态功能连接和图论确定了 TBI 中全脑网络的训练诱导的神经可塑性模式。方法。共有 64 名患有 TBI 的平民和退伍军人被随机分为基于策略的认知训练组（n = 33）或基于知识的训练组（主动对照组；n = 31），为期 8 周。参与者经历了轻度至重度的 TBI，没有局部损伤和持续的认知功能障碍。一部分参与者抱怨亚临床但残留的精神症状。我们在训练前、训练后立即和训练后 3 个月获得了他们的静息状态功能磁共振成像。从参与者的静止状态网络中，我们获得了多个网络密度下的模块性、参与系数、模块内连接性、全局效率和局部效率。接下来，我们对那些使用错误发现率 (FDR) 进行跨网络密度多重比较校正的度量进行了纵向分析。结果。相对于基于知识的训练组，基于策略的认知训练组随着时间的推移降低了模块性并增加了参与系数、全局效率和局部效率（Pnodal <.05；qFDR <0.05）。大脑行为分析显示，基于策略的认知训练组的参与系数和整体效率与训练背景下的开拓得分相关（Pnodal <.05；qFDR <0.05）。结论。认知训练在整个大脑的 TBI 中重组了模块化网络。图论方法可能有助于确定 TBI 中潜在的基于大脑的训练效果标记。