Video Dataset of Respiratory Cilia Motion Phenotypes

Respiratory cilia are important components in the lung defense mechanism. The coordinated beating of cilia cleans the airways of pathogens and foreign particles. We present a large-scale validation dataset of cilia motion for characterizing ciliary function. Ciliary beat frequency (CBF) is provided as benchmark metrics. The video dataset of cilia motion phenotypes contains four categories: temperatures, drugs and ACE2 manipulation. Under each category, mouse trachea samples were treated with different stimuli and imaged with a high-speed video microscope to acquire cilia motion. In addition, we generate ground truth masks labeling ciliary area for image segmentation. This validation dataset can serve as a benchmark for the computer vision community to develop models for analyzing ciliary beat pattern.

This video dataset contains 872 videos and their ground-truth masks with the ciliary area labeled. The videos were recorded at 250 frames per second for 1 second. The image size is 800x800. Each pixel is 0.07987 μm. The csv file contains the CBF values of each video.