Ensemble deep learning with advanced feature engineering for embryo evaluation on in-vitro fertilisation procedures...

Area of Science:

• Reproductive Medicine
• Artificial Intelligence
• Biomedical Imaging

Background:

• In vitro fertilisation (IVF) is a key assisted reproductive technology for infertility treatment.
• Low IVF success rates are often linked to poor embryo quality, necessitating precise embryo assessment.
• Automated embryo grading is complex due to intricate morphology, posing challenges for traditional methods.

Purpose of the Study:

• To propose an automated embryo grading method using ensemble deep learning for improved IVF outcomes.
• To enhance the selection of viable embryos for successful implantation and pregnancy.
• To develop an advanced feature engineering system for biomedical images in IVF procedures.

Main Methods:

• An ensemble deep learning-enabled embryo evaluation system (EDLEVS-AFEBI) was developed.
• Image pre-processing utilized an adaptive Gaussian bilateral filter (AGBF) for noise reduction.
• Feature extraction employed an improved DenseNet model, followed by ensemble classification using TCN, ENN, and CVAE.

Main Results:

• The EDLEVS-AFEBI model achieved a high accuracy of 99.39% in embryo classification.
• The system demonstrated superior performance compared to other models on a microscopic image dataset.
• Advanced feature engineering and ensemble learning contributed to precise embryo quality assessment.

Conclusions:

• The proposed automated embryo grading system significantly enhances the accuracy of embryo selection in IVF.
• This AI-driven approach has the potential to improve implantation success rates and reduce healthcare costs.
• The EDLEVS-AFEBI model represents a significant advancement in applying deep learning to reproductive medicine.