Automatic detection and tracking of animal sperm cells in microscopy images

Sperm tracking-and-analysis is one of the interesting topics in biological research and reproductive medicine, as it helps to assess the quality of the sperm for the male infertility. Computer-Assisted Sperm Analysis (CASA) systems provide a rapid and automated assessment of the parameters of sperm motion, together with improved standardization and quality control. In this paper, we propose a method to detect and track animal sperms automatically. First, we detect the sperms in the first frame of all the sequences using a bag-of-words approach and SVM classifier. Then, the detected sperm cells are tracked in the rest of all sequences using mean-shift. The proposed algorithm is evaluated on three videos in our datasets which have sperms as groundtruth. The experimental results show that our method achieves a precision of 0.94, 0.93 and 0.96, and a recall of 0.96, 0.92, and 0.97 for the three videos respectively in terms of sperm detection. RMSE (Root mean square error) is calculated to evaluate our results in terms of sperms tracking. The results show that we achieve high performance with RMSE of 8.06, 9.01, and 7.09 pixels for three different videos.