A hybrid IMM-JPDAF algorithm for tracking multiple sperm targets and motility analysis

Semen analysis has received a lot of attention because of its important role in determining infertility in men. It involves several factors, the most important of which are sperm morphology, sperm concentration, and sperm motility. In addition, measurements of sperm cell mobility reflect important parameters in medical diagnosis. As computer-assisted semen analysis systems are very expensive and not prolific, especially in small medical laboratories, semen analysis is often done manually. This is a time-consuming and costly process. Therefore, we have developed an automated system that evaluates sperm motility parameters which can be of great help to clinicians in achieving more accurate results at a lower cost and time. The system tracks the movement of most spermatozoa cells in a semen sample taken from a microscope and then carefully measures all the parameters related to sperm movement to determine the fertility or infertility rate of the subject. Detecting large numbers of sperm cells is challenging because there are a large number of colliding targets that cause false alarms. In this work, the background subtraction method is used to determine the sperms within the video frames, and the joint probabilistic data association filter algorithm is used to estimate the sperm trajectory and to associate different tracks. Since the sperm represents maneuvering movements, the interacting multiple models technique was used along with the JPDAF algorithms to obtain more accurate results. Our evaluations on real and synthetic data reveal the superiority of our method over previous work in sperm cell tracking.