Automated analysis of intratidal dynamics of alveolar geometry from microscopic endoscopy

Alveolar parenchyma, the gas exchange area of the respiratory system, is prone to mechanical damage during mechanical ventilation. Development of lung protective ventilation strategies therefore requires a better understanding of alveolar dynamics during mechanical ventilation. In this paper, we propose a novel method for automated analysis of the intratidal geometry of subpleural alveoli based on the evaluation of video frames recorded from alveolar microscopy in an experimental setting. Our method includes the recording with a microscopic endoscope, feature extraction from image data, the analysis of a single frame, the tracking and analysis of single alveoli in a video sequence, and the evaluation of the obtained sequence of alveolar geometry data. Our method enables automated analysis of 2-D alveolar geometry with sufficient temporal resolution to follow intratidal dynamics. The developed method and the reproducibility of the results were successfully validated with manually segmented video frames.