Automated Morphology Detection from Human Sperm Images

Background: Sperm morphology is an important factor impacting male fertility potential. Thus, the assessment of this parameter is a key step during the infertility diagnosis and in real-time selection of sperm for intracytoplasmic sperm injection (ICSI). In recent years, one interesting subject in sperm morphology analysis (SMA) involves use of an automated algorithm. These algorithms can accelerate detection of typical and atypical forms. This chapter presents one of the most successful SMA algorithms for use in real-time analysis of human sperm selection. Methods: A human sperm morphology analysis dataset (HSMA-DS) was created using sperm images (n = 1457) from 235 patients. Different human sperm components were detected and analyzed using this SMA method. Initially, noise removal was performed to enhance image contrast followed by recognition and analysis of size and shape of different sperm regions (i.e., head, large vacuole, midpiece, and tail). Classification of each sperm was done by this algorithm as normal or abnormal forms. Therefore, the SMA algorithm recognizes the different atypical forms in the head, midpiece, and tail of a sperm. This analysis by the SMA algorithm is performed on non-stained images with low resolution.