Abstract / Summary
Abstract Primary ciliary dyskinesia (PCD) is a genetic condition causing dysmotile cilia and abnormal mucociliary clearance. Despite advances in understanding disease pathogenesis, diagnosis remains challenging. Here, we show that machine learning analysis of cilia-driven particle transport can objectively detect primary ciliary dyskinesia-related dysfunction. Using high-speed video microscopy, we captured fluorescent microsphere movement on cultured multiciliated cells from 14 individuals with primary ciliary dyskinesia and 10 healthy donors (602 videos). We developed CiliaTracks, an open-source computational framework that extracts measures of micropsphere movement, visualizes transport patterns, and classifies cilia-propelled particle transport. Quantitative and visual analyses reveal distinct motility profiles between PCD and non-primary ciliary dyskinesia and normal recordings. Classical machine learning models and a convolutional neural network classify recordings with 95-97% accuracy and differentiate affected from normal samples. Machine learning provides an objective, accurate framework for characterizing ciliary dysfunction, offering potential as a diagnostic tool for primary ciliary dyskinesia.