Automatic Healthy Sperm Head Detection using Deep Learning

Infertility is one of the diseases in which researchers are interested. disease a global health concern, and andrologists constantly looking for more advanced solutions this disease. The intracytoplasmic sperm injection (ICSI) process considered as most common procedures achieving fertilization. Sperm selection performed using visual assessment dependent upon skills laboratory technicians such prone to human errors. Therefore, an automatic detection system needed quick accurate results. This study utilizes deep learning technique classification heads sperms indicate healthy sperms. Convolutional Neural Network (CNN) model Geometry Group 16 layers (VGG16) was used classification, it best architectures image classification. dataset consists 1200 images divided into unhealthy. Here, VGG16 fine-tuned achieved accuracy 97.92% sensitivity 98.82%. Moreover, F1 score 98.53%. effective real-time detecting that can be injected eggs successful quickly recognizes makes easier andrologists.