Rigorous and reproducible evaluation of lung tissue under different conditions is necessary to interpret development, injury, and pharmacologic interventions. Common histological measurements in the distal lung include mean linear intercept (MLI) as a metric of alveolarization and airspace volume density (ASVD) as a metric of airspaces relative to tissue. Historically, these have been performed manually in a time-intensive process, with reproducible trends but a high degree of variability between individuals. To improve the reproducibility and throughput of lung morphometry, we developed AlveolEye, an open-source, semiautomated, computer vision-assisted tool that rapidly and reproducibly calculates MLI and ASVD from images of standard hematoxylin and eosin-stained tissue sections. AlveolEye-assisted MLI calculation closely aligns with manually derived measurements for corresponding images, with preservation of trends in measurements between noninjured controls and neonatal mice subjected to two different injury models. Analyzing human tissue of varying ages suggests that the approach developed in AlveolEye is generalizable across species. Notably, AlveolEye markedly reduced the average variation across individual analyzers, with the greatest improvement in precision among individuals with the least experience in performing lung morphometry. The design of AlveolEye is intentionally semiautomated, preserving the investigator's ability to assess and adjust parameters based on sample characteristics. AlveolEye facilitates efficient lung morphological measurements on larger sample sizes, allowing for greater statistical power for preclinical studies, and improves precision across individual observers, allowing for improved rigor in experimental design and execution.NEW & NOTEWORTHY Calculating lung morphometric measurements is time-intensive, with a high degree of interrater variability. We developed a semiautomated tool, AlveolEye, which uses computer vision to perform rapid and precise lung morphometry, reducing variability and improving efficiency. AlveolEye presents an opportunity for improved rigor, efficiency, and reproducibility in studies of lung development, injury, and regeneration.
Keywords: lung morphometry; machine learning.