Gottman Drew C, Smith Bradford J
University of Colorado School of Medicine, University of Colorado Denver, Aurora, CO, United States.
Department of Bioengineering, University of Colorado Denver, Anschutz Medical Campus, Aurora, CO, United States.
Front Netw Physiol. 2024 May 1;4:1392701. doi: 10.3389/fnetp.2024.1392701. eCollection 2024.
Acute respiratory distress syndrome (ARDS) presents a significant clinical challenge, with ventilator-induced lung injury (VILI) being a critical complication arising from life-saving mechanical ventilation. Understanding the spatial and temporal dynamics of VILI can inform therapeutic strategies to mitigate lung damage and improve outcomes.
Histological sections from initially healthy mice and pulmonary lavage-injured mice subjected to a second hit of VILI were segmented with Ilastik to define regions of lung injury. A scale-free network approach was applied to assess the correlation between injury regions, with regions of injury represented as 'nodes' in the network and 'edges' quantifying the degree of correlation between nodes. A simulated time series analysis was conducted to emulate the temporal sequence of injury events.
Automated segmentation identified different lung regions in good agreement with manual scoring, achieving a sensitivity of 78% and a specificity of 85% across 'injury' pixels. Overall accuracy across 'injury', 'air', and 'other' pixels was 81%. The size of injured regions followed a power-law distribution, suggesting a 'rich-get-richer' phenomenon in the distribution of lung injury. Network analysis revealed a scale-free distribution of injury correlations, highlighting hubs of injury that could serve as focal points for therapeutic intervention. Simulated time series analysis further supported the concept of secondary injury events following an initial insult, with patterns resembling those observed in seismological studies of aftershocks.
The size distribution of injured regions underscores the spatially heterogeneous nature of acute and ventilator-induced lung injury. The application of network theory demonstrates the emergence of injury 'hubs' that are consistent with a 'rich-get-richer' dynamic. Simulated time series analysis demonstrates that the progression of injury events in the lung could follow spatiotemporal patterns similar to the progression of aftershocks in seismology, providing new insights into the mechanisms of injury distribution and propagation. Both phenomena suggest a potential for interventions targeting these injury 'hubs' to reduce the impact of VILI in ARDS management.
急性呼吸窘迫综合征（ARDS）是一项重大的临床挑战，呼吸机诱导的肺损伤（VILI）是挽救生命的机械通气引发的关键并发症。了解VILI的时空动态可为减轻肺损伤和改善预后的治疗策略提供依据。
使用Ilastik对最初健康的小鼠以及遭受VILI二次打击的肺灌洗损伤小鼠的组织学切片进行分割，以定义肺损伤区域。应用无标度网络方法评估损伤区域之间的相关性，损伤区域在网络中表示为“节点”，“边”量化节点之间的相关程度。进行模拟时间序列分析以模拟损伤事件的时间顺序。
自动分割识别出的不同肺区域与人工评分高度一致，在“损伤”像素上的灵敏度达到78%，特异性为85%。“损伤”、“空气”和“其他”像素的总体准确率为81%。损伤区域的大小遵循幂律分布，表明肺损伤分布中存在“富者愈富”现象。网络分析揭示了损伤相关性的无标度分布，突出了损伤中心，这些中心可作为治疗干预的重点。模拟时间序列分析进一步支持了初始损伤后发生二次损伤事件的概念，其模式类似于地震学余震研究中观察到的模式。
损伤区域的大小分布强调了急性和呼吸机诱导的肺损伤在空间上的异质性。网络理论的应用表明出现了与“富者愈富”动态一致的损伤“中心”。模拟时间序列分析表明，肺损伤事件的进展可能遵循与地震学中余震进展相似的时空模式，为损伤分布和传播机制提供了新见解。这两种现象都表明，针对这些损伤“中心”进行干预有可能在ARDS管理中减少VILI的影响。
