Sperm screening with neural networks model
Code: G-1001
Authors: Gilda Sharifi * ℗, Melika Yazdanpanah, Farangis Sharifi
Schedule: Not Scheduled!
Tag: Intelligent Virtual Assistant
Download: Download Poster
Abstract:
Abstract
Introduction: Infertility affects around 30 to 40 percent of men and is a significant concern for couples. Intracytoplasmic sperm injection (ICSI) is a leading treatment method for infertility. The accuracy of sperm selection in ICSI depends on embryologists' expertise, leading to potential errors. Current sperm analysis methods follow WHO guidelines and depend on lab technicians' skills, resulting in variability and errors. Some studies have utilized the ResNet-50 neural network model for human sperm screening, aiming to improve the success rate of ICSI by using healthy sperm. This study focuses on "Sperm Screening with Neural Network Models" to enhance the ICSI process. Materials and Methods: Research data was gathered by reviewing articles on Google Scholar, PubMed, Web of Science, and Scopus, focusing on "neural network, male infertility, sperm screening." Inclusion criteria involved articles on increasing fertility and sperm screening using neural networks. Data extraction involved screening article titles and abstracts by two researchers, extending to full-text articles when necessary. Data and Method: In reviewed articles, a dataset categorized sperm into healthy and unhealthy types after pre-processing. Image pre-processing consisted of three stages: noise removal using a median filter, image normalization to adjust pixel intensity, and data augmentation to enhance image quality and model accuracy. The proposed model achieved a 96.66% accuracy rate, effectively identifying healthy sperm. Conclusion: Incorporating neural networks and artificial intelligence to boost fertility rates shows promise. Utilizing a deep learning model like ResNet-50 for accurate detection of healthy human sperm with nearly 100% accuracy proved successful. This model is efficient in identifying healthy sperm, potentially improving the success of sperm injection into eggs and enhancing pregnancy rates. With simplified pre-processing steps, the model provides quick and accurate results, simplifying sperm injection procedures for andrologists.
Keywords
Screening, Male Infertility, Sperm, Neural Network
