Analysis of sperm beating characteristics using microfluidic trapping and machine-learning-based flagellum tracking

Male infertility affects a significant portion of couples worldwide, with standard semen analysis often failing to identify functional deficiencies in sperm performance. This study presents a microfluidic platform for characterizing sperm flagellar beating patterns with unprecedented detail, providing insights into sperm functional parameters potentially linked to unexplained infertility. We combined microcontact printing of fibronectin adhesion spots with machine-learning-based flagellum tracking to immobilize sperm heads while allowing free flagellar movement, enabling precise analysis of beating characteristics. Our tracking algorithm utilizes a YOLOv8 (You Only Look Once) machine learning-based computer vision model, which we trained using 950 manually annotated images of sperm cells. We used keypoint detection along the sperm flagellum to calculate critical beating parameters including the beating amplitude, frequency, and asymmetry patterns. To validate the platform, we investigated the effects of established capacitation and hyperactivation agents on sperm motility on the flagellum beating pattern on bovine sperm cells. Caffeine treatment (10-40 mM) increased flagellar beating amplitude by up to 65% while decreasing frequency by approximately 50%, with pronounced asymmetrical beating consistent with hyperactivation. Heparin exposure (10-100 μg ml^-1) similarly enhanced beating amplitude by approximately 25% without significantly altering frequency. We also analyzed the beating pattern of sperm cells immobilized inside a microchannel under different flow velocities. Results revealed a decrease in the beating frequency when sperm cells were subjected to flow. The platform eliminates the need for sophisticated sperm tracking techniques and facilitates high-throughput analysis under controlled physicochemical conditions. By enabling detailed characterization of sperm flagellar behaviour under various stimuli, our platform offers a valuable tool for investigating molecular mechanisms underlying idiopathic male infertility and evaluating potential therapeutic interventions.