The quality control of frozen semen samples from cattle is established by parameters such as percentage of progressive motility (% MP) because it is related to the fertilization capacity of bulls. Nowadays, sperm quality test is performed by direct visual inspection of sperm through a microscope in andrology laboratories. However, there is a high subjectivity in the observation and assessment depending on the observer; thus, causing unreliable diagnoses and non-repetitiveness in results. The development of a low cost computer tool was proposed to identify the individual sperm motility in cattle through the application of artificial vision algorithms. The methodology consisted of: the acquisition and pre-processing of videos obtained from thawed cryopreserved samples, the segmentation, filtering and detection of spermatozoa using the Fisher Discriminant Analysis and Adaptive Gaussian Models, followed by the assignment and construction of sperm trajectories through the Munkres Algorithm and Kalman Filter. Finally, the characterization and assessment of sperm motility parameters were performed based on the criteria present in computer-aided semen analysis (CASA) commercial systems. The results obtained showed high correlation for the individual sperm motility with a determination coefficient of 0.8143 for 10 different samples of bull sperm with respect to manual analysis. Likewise, a concordance coefficient of 0.966 was found in the 95% confidence interval using the Bland-Altman test, indicating that the measures were highly similar. In this way, the methodology is a reliable technological support that contributes to the improvement of quality control of semen samples from cattle.
Key words: Sperm motility, track detection, fisher discriminant analysis, Gaussian models, Kalman filter
Copyright © 2020 Author(s) retain the copyright of this article.
This article is published under the terms of the Creative Commons Attribution License 4.0