Improved accuracy for estrous cycle staging using supervised object detection.

Area of Science:

• Reproductive biology
• Computational biology
• Veterinary science

Background:

• The estrous cycle is critical for female mammal reproduction and health, impacting research outcomes.
• Traditional estrous cycle monitoring via vaginal cytology is time-consuming and prone to accuracy issues.
• Accurate estrous cycle staging is essential for interpreting results in studies involving female subjects.

Purpose of the Study:

• To evaluate the feasibility and reliability of machine learning for estrous cycle staging.
• To develop and assess an object detection model for automated estrous cycle classification.
• To improve the accuracy and efficiency of estrous cycle monitoring in research settings.

Main Methods:

• An object detection model, Object Detection Estrous Staging (ODES), was developed.
• A dataset of 555 mouse vaginal cytology images with various stains was annotated.
• The ODES model was trained, validated, and tested on the image dataset.

Main Results:

• ODES achieved an average accuracy of 87% in classifying estrous cycle stages.
• The model analyzed 175 test images in just 3.9 minutes.
• Machine learning significantly outperformed previous models (33-45% accuracy) and human accuracy (66%).

Conclusions:

• Machine learning, specifically ODES, offers a reliable and efficient method for estrous cycle monitoring.
• This technology enhances research practices in studies involving female mammals.
• Accurate estrous cycle identification improves the quality and interpretability of scientific findings.