Automated Sperm Morphology Analysis with Deep Learning and Feature Extraction

Accurate assessment of sperm morphology is critical for improving artificial insemination procedures. This project introduces an AI-driven approach using deep learning to enhance sperm morphology evaluation. Integrating advanced image processing and deep learning models, the goal is to automate and refine sperm analysis, enhancing procedural efficiency and reliability. Our study evaluates six preprocessing techniques, two feature extraction techniques and employs models like VGG16, VGG19, ResNet50, Xception, and DenseNet121 to classify sperm into Acrosome, Head, and Vacuole categories. Results show significant performance boosts with techniques like adaptive denoising and CLAHE, achieving up to 88% accuracy for Vacuole classification with VGG16. Comparative analysis with existing methods validates the effectiveness of our approach, highlighting preprocessing's crucial role in medical image classification and suggesting avenues for further research and clinical application.