Fig 1.
Summary of image processing and analysis steps for quantifying alveolar duct and alveolar area.
(A) Representative RGB color image acquired at 20x magnification. (B) Isolated green channel of A. (C) Vessels removed from B. (D) Binarized version of C. (E) Disconnected exudate and debris removed from D. (F) Smoothed version of E. (G) Small holes removed from F. Selective quantification of (H) alveolar ducts or (I) alveoli.
Table 1.
Base area and circularity values tested for the quantification of alveolar ducts.
Table 2.
Base area and circularity values tested for the quantification of alveoli.
Fig 2.
Optimizing method parameters on images from mice exposed to an acute hyperoxia-induced model of BPD.
(A) Representative H&E-stained lung images from P14 mice exposed to 21% O2 (n = 3) or 95% O2 (n = 4) for 72 hours. (B) Quantification of mean linear intercept (LM) in the mice from A. (C) Representative colored images depicting alveolar ducts. (D) Evaluation of Pearson’s r across minimum area and circularity combinations for alveolar ducts. Red dots depict the r coefficient for all 7 tested samples. Black dots depict the median r coefficient for all 7 LOO iterations. Bars represent the range of values across all 7 LOO iterations. (E) Scatter plot of mean duct area plotted against measured LM. Solid black line depicts the fitted linear regression line. Dashed lines represent the 95% confidence intervals. (F) Bland-Altman plot of z-scores for alveolar duct area and LM. Dashed lines represent the 95% limit of agreement (LOA). Red line depicts bias. (G) Representative colored images depicting alveoli. (H) Evaluation of Pearson’s r across minimum area and circularity combinations for alveoli. Red dots depict the r coefficient for all 7 tested samples. Black dots depict the median r coefficient for all 7 LOO iterations. Bars represent the range of values across all 7 LOO iterations. (I) Scatter plot of mean alveolar area plotted against measured LM. Solid black line depicts the fitted linear regression line. Dashed lines represent the 95% confidence intervals. (J) Bland-Altman plot of z-scores for alveolar area and LM. Dashed lines represent the 95% limit of agreement (LOA). Red line depicts bias. (K) Bar graph depicting processing time per image across a test set of 6 images (1 animal). (○) female mice, (●) male mice. *P < 0.05. Student’s t-test used in B. Simple linear regression used in E, I. Pearson’s correlation used in D,H. Bland-Altman analysis used in F,J. Error bars are mean ± SD.
Fig 3.
Exposure to postnatal hyperoxia increases alveolar duct size and number.
(A) Representative H&E-stained images from mice exposed to 21% O2 (n = 12) or 95% O2 (n = 10) for 72 hours. (B) Quantification of LM in the mice from A. (C) Representative colored images depicting alveolar ducts. (D) Quantification of mean alveolar duct area from the same mice as A. (E) Quantification of the number of ducts per 20x FOV in the same mice as A. (F) Quantification of total alveolar duct area in the same mice as A. (G) Scatter plot of mean alveolar duct area plotted against measured LM. Solid black line depicts the fitted linear regression line. Dashed lines represent the 95% confidence intervals. (H) Bland-Altman plot of z-scores for alveolar area and LM. Dashed lines represent the 95% limit of agreement (LOA). Red line depicts bias. (○) female mice, (●) male mice. ****P < 0.0001, ***P < 0.001, P** < 0.01. Student’s t-test used in B,D,E,F. Simple linear regression used in G. Bland-Altman analysis used in H. Error bars are mean ± SD.
Fig 4.
Neonatal hyperoxia impairs alveolar septation.
(A) Representative colored images depicting alveoli from mice exposed to 21% O2 (n = 12) or 95% O2 (n = 10) for 72 hours. (B) Quantification of mean alveolar area from the same mice as A. (C) Quantification of the number of alveoli per 20x FOV in the same mice as A. (D) Quantification of total alveolar area in the same mice as A. (E) Scatter plot of mean alveolar duct area plotted against measured LM. Solid black line depicts the fitted linear regression line. Dashed lines represent the 95% confidence intervals. (F) Bland-Altman plot of z-scores for alveolar area and LM. Dashed lines represent the 95% limit of agreement (LOA). Red line depicts bias. (○) female mice, (●) male mice. ****P < 0.0001, ***P < 0.001, P** < 0.01. Student’s t-test used in B,C,D. Simple linear regression used in E. Bland-Altman analysis used in F. Error bars are mean ± SD.
Fig 5.
Quantification of intermediate airspace characteristics on the validation dataset.
(A) Representative colored images depicting alveoli from mice exposed to 21% O2 (n = 12) or 95% O2 (n = 10) for 72 hours. (B) Quantification of mean airspace area from the same mice as A. (C) Quantification of the number of airspaces per 20x FOV in the same mice as A. (D) Quantification of total airspace area in the same mice as A. (E) Quantification of percent airway area for each subdivision of airspace in the same mice as A. Student’s t-test used in B,C,D,E. Error bars are mean ± SD.