Embryo Selection for IVF using Machine Learning Techniques Based on Light Microscopic Images of Embryo and Additional...

Area of Science:

• Reproductive Medicine
• Artificial Intelligence in Healthcare
• Embryology

Background:

• Current In Vitro Fertilization (IVF) embryo selection relies on subjective morphological grading (e.g., Istanbul scoring system).
• Lack of objective criteria for prioritizing embryos with similar morphology hinders effective selection.
• Existing methods offer limited guidance when multiple embryos share the same grade.

Purpose of the Study:

• To develop a deep learning model for classifying viable and non-viable embryos using light microscopic images.
• To integrate additional clinical features, including the Istanbul grading system and patient age, into the model.
• To enhance the prediction of embryo implantation potential.

Main Methods:

• Utilized light microscopic images of embryos for analysis.
• Developed and evaluated various deep learning models.
• Fused embryo image data with supplementary features (Istanbul grading, patient age, ICM, TE quality, developmental stage).

Main Results:

• The best-performing model achieved 65% accuracy, 74.29% sensitivity, and an AUC of 0.72.
• Analysis revealed that patient age, embryo development stage, Inner Cell Mass (ICM), and Trophectoderm (TE) quality positively impact prediction.
• The model demonstrated improved prediction of implantation potential when incorporating these additional factors.

Conclusions:

• Deep learning models integrating embryo images and clinical data offer a promising approach for objective embryo selection in IVF.
• Incorporating features like patient age and specific morphological indicators (ICM, TE) significantly enhances predictive accuracy.
• This AI-driven approach can potentially improve IVF success rates by optimizing embryo transfer decisions.