Expert-Annotated Optical Microscopy Images of Human Sperm for Detection and DNA Fragmentation Assessment

Area of Science:

• Reproductive Biology
• Medical Imaging
• Computational Biology

Background:

• Automated analysis of human sperm is crucial for male fertility diagnostics.
• High-quality, labeled datasets are essential for developing accurate machine learning models.
• Assessing sperm DNA fragmentation is a key indicator of male reproductive health.

Purpose of the Study:

• To present a collection of expert-labeled image datasets for automated human sperm analysis.
• To provide resources for training and benchmarking machine learning models in male fertility diagnostics.
• To facilitate research in reproductive health through openly available data.

Main Methods:

• Creation of three distinct image datasets: raw images, binary classification (sperm/non-sperm), and multiclass (DNA fragmentation levels).
• Annotation of over 400 high-resolution bright-field images by five experienced embryologists.
• Determination of final labels through independent expert agreement to ensure annotation quality.

Main Results:

• A comprehensive collection of labeled images supporting sperm detection and classification.
• Datasets enabling the assessment of sperm chromatin integrity based on DNA fragmentation levels (halo size).
• Open availability of data to promote reproducibility and further scientific investigation.

Conclusions:

• The presented datasets form a robust foundation for advancing automated sperm analysis.
• These resources will accelerate the development of AI-driven tools for male fertility diagnostics.
• The open data initiative supports collaborative research and innovation in reproductive medicine.