Department of Radiation Oncology, William Beaumont Hospital, Royal Oak, Michigan 48073, USA.
Radiat Res. 2010 Apr;173(4):536-44. doi: 10.1667/RR1712.1.
Pulmonary damage after radiotherapy is typically characterized by an initial alveolar inflammation (pneumonitis) followed by chronic fibrosis. In the present study, changes in lung architecture were measured in the pneumonitis phase after whole-body low-dose X irradiation of C57BL/6 mice. Radiation damage was evaluated at 24 h and 1-8 weeks postirradiation. Three distinct scoring systems were used: ( 1 ) manually evaluating alveolar distortion and infiltration of inflammatory cells into the alveolar space using a continuous numerical scale across an entire lung section, ( 2 ) physically measuring the average thickness of the alveolar septa from multiple representative microscope fields, and ( 3 ) a new rapid automated mathematical algorithm based on image segmentation of alveolar space across an entire section. Each scoring method detected significant changes in alveolar architecture at the earliest times compared with sham-treated controls and gave comparable evaluations of injury. The results from the automated mathematical algorithm correlated significantly with both the manual evaluation method (Spearman's correlation coefficient rho = 0.044) and the direct physical measurement of septa thickness (rho = 0.002). These data demonstrate that evaluating alveolar space by segmentation analysis provides a reliable method for scoring early pulmonary radiation damage that is consistent with more established methodologies but is more rapid and is independent of potential operator and selection bias.
放射性肺损伤的典型特征是初始的肺泡炎症(放射性肺炎),随后是慢性纤维化。本研究在 C57BL/6 小鼠全身低剂量 X 照射后的放射性肺炎期测量了肺结构的变化。在照射后 24 小时和 1-8 周评估了放射性损伤。使用了三种不同的评分系统:(1)使用连续数值标度手动评估整个肺切片中肺泡的变形和炎症细胞浸润到肺泡腔,(2)从多个代表性显微镜视野物理测量肺泡间隔的平均厚度,(3)基于整个切片的肺泡空间图像分割的新的快速自动数学算法。与假处理对照相比,每种评分方法在最早的时间都检测到肺泡结构的显著变化,并对损伤进行了可比的评估。来自自动数学算法的结果与手动评估方法(Spearman 相关系数 rho = 0.044)和对隔厚度的直接物理测量(rho = 0.002)显著相关。这些数据表明,通过分割分析评估肺泡空间提供了一种可靠的方法来评分早期放射性肺损伤,该方法与更成熟的方法一致,但更快速且独立于潜在的操作者和选择偏见。