一种基于深度学习的精子尾部识别方法

A Method of Sperm Tail Recognition Based on Deep Learning

精子数量的多少是衡量男性生殖能力的重要指标,传统方法是通过显微镜和摄像系统放大并采集显微镜下动态图像,对图像中的精子进行计数或识别,但由于精液中含有很多的非精子成分(圆细胞、杂质等),会影响到对精子数量的判断。通过实验证明,本方法取得了比较好的效果,优点是通过精子尾部的识别,过滤掉精液中的非精子细胞或杂质,清晰的地呈现精液中的精子数量,为临床提供了很好的数据支撑。

The number of sperm is an important indicator to measure male reproductive ability.The traditional method is to magnify and collect dynamic images under the microscope through a microscope and camera system,and to count or identify the sperm in the image,however the semen contains a lot of non-sperm components(round cells,impurities,etc.),which will affect the judgment of sperm count.Experiments have shown that this method has achieved relatively good results.The advantage is that through the identification of the sperm tail,non-sperm cells or impurities in the semen are filtered,and the number of sperm in the semen is clearly displayed,which provides good data support for clinical practice.