Clustering Dynamics of Human Spermatozoa Motility During ICSI And Its Correlation With Fertilization And Blastocyst Quality

A change in kinetic activity of spermatozoa called hyperactivation and the acrosome reaction are two interrelated processes of sperm activation observed in vitro. Although studies have investigated hyperactivation, no correlation has been found between individual sperm motility patterns with fertilization outcomes. In this study, we observed individual human sperm trajectories to determine analytical categories based on their dynamics that correlates to fertilization and blastocyst formation outcomes during ICSI. A set of 1,637 sperm trajectories in PVP during ICSI, acquired using SiD 2.0 software (IVF 2.0 Ltd., UK), was analyzed retrospectively. The first three Fourier coefficients, the peak velocity and the area of the convex hull from each trajectory were obtained to summarize sperm motion dynamics. Clustering analyses such as k-means, Gaussian mixture model (GMM) and spectral clustering were carried out to explore whether spermatozoa paths fall into similar analytical categories i.e clusters. Principal component analysis (PCA) was used to obtain a two-dimensional visualization of the cluster distribution. Findings were assessed according to fertilization and subsequent embryo development. All three clustering methods consistently identified three mathematical clusters as the most structurally coherent groupings, with k-means achieving the highest silhouette score of 0.56. One mathematical cluster was predictive of blastocyst formation (65% of embryos with ICM and TE), indicating a significant correlation between extracted trajectory features and blastocyst quality. This association was statistically validated by a chi-squared test, resulting in a p-value of 0.0029, suggesting a strong association between cluster assignment and blastocyst quality. The 2-dimensional PCA further confirmed the clear separation of the clusters. The congruence of sperm clustering methods identified three patterns of sperm grouping. The significant presence of high-quality blastocysts in a specific cluster, may illustrate the potential of using sperm trajectory characteristics as a non-invasive predictive tool of embryonic developmental capacity. These findings pave the way for advanced analytical techniques in reproductive medicine, particularly for improving in vitro fertilization (IVF) outcomes by allowing sperm selection based on quantitative motility patterns. Future studies may seek to identify new features or apply advanced machine learning techniques to improve predic