Automatic Analysis of Human Sperm Motion

Abstract

A new computerized methodology is described in which sperm movement characteristics are analyzed automatically. Video fields containing images of spermatozoa are electronically digitized, and the centroid position of each sperm head is determined. Over time, strings of centroids identify the swimming paths of individual spermatozoa. Details of path acquisition are described for human spermatozoa in semen, and include a discussion of how individual cells are identified and distinguished from each other. A diversified set of movement characteristics is computed for each spermatozoon, including two new measures of path shape based on the instantaneous turning angle. The traditional and the new measures of vigor and swimming pattern are evaluated and compared for consistency and redundancy. Analysis of data from human semen indicates that the new angular measures may be particularly useful in discriminating between spermatozoa exhibiting widely different patterns of motion.