Artificial intelligence-powered oocyte evaluation: Correlating cytoplasmic features with blastocyst development.

Area of Science:

• Reproductive Biology
• Artificial Intelligence in Medicine
• Biomedical Imaging

Background:

• Oocyte quality is crucial for successful fertilization and embryo development.
• Traditional methods for assessing oocyte quality are often subjective.
• Developing objective, quantitative measures of oocyte quality is essential for improving assisted reproductive technologies.

Purpose of the Study:

• To develop a quantitative and interpretable artificial intelligence (AI)-driven method for assessing oocyte quality.
• To analyze cytoplasmic morphology and intensity features for predicting oocyte developmental competence.
• To establish a standardized, non-invasive approach for oocyte evaluation.

Main Methods:

• Collected 695 oocyte images from young and aged mice.
• Performed radiomics analysis on manually annotated cytoplasmic regions to extract morphological and intensity features.
• Utilized Gaussian mixture models for clustering oocytes based on cytoplasmic characteristics.

Main Results:

• Identified three distinct oocyte subtypes based on cytoplasmic features.
• The most spherical and compact oocytes (Cluster 2) showed the highest blastocyst formation rate (42.9%).
• Oocyte cytoplasmic quality, assessed by AI, proved to be a more informative predictor of developmental potential than chronological age alone.

Conclusions:

• Cytoplasmic features serve as objective indicators of oocyte developmental competence.
• AI-driven analysis of oocyte morphology and intensity can enhance embryo selection in assisted reproductive technologies.
• This approach offers a standardized, non-invasive method for oocyte quality assessment, potentially improving clinical outcomes.