Predicting Blastocyst Formation in IVF: Integrating DINOv2 and Attention-Based LSTM on Time-Lapse Embryo Images

TL;DR

This study introduces a hybrid deep learning model combining DINOv2 and attention-based LSTM to predict blastocyst formation from limited embryo images, achieving high accuracy and robustness for IVF embryo selection.

Contribution

The paper presents a novel hybrid model that effectively predicts embryo development using limited images, improving upon existing methods and accommodating incomplete data.

Findings

Achieved 96.4% accuracy in predicting blastocyst formation.

Model performs well even with missing frames in embryo videos.

Outperforms existing prediction methods on real dataset.

Abstract

The selection of the optimal embryo for transfer is a critical yet challenging step in in vitro fertilization (IVF), primarily due to its reliance on the manual inspection of extensive time-lapse imaging data. A key obstacle in this process is predicting blastocyst formation from the limited number of daily images available. Many clinics also lack complete time-lapse systems, so full videos are often unavailable. In this study, we aimed to predict which embryos will develop into blastocysts using limited daily images from time-lapse recordings. We propose a novel hybrid model that combines DINOv2, a transformer-based vision model, with an enhanced long short-term memory (LSTM) network featuring a multi-head attention layer. DINOv2 extracts meaningful features from embryo images, and the LSTM model then uses these features to analyze embryo development over time and generate final predictions. We tested our model on a real dataset of 704 embryo videos. The model achieved 96.4% accuracy, surpassing existing methods. It also performs well with missing frames, making it valuable for many IVF laboratories with limited imaging systems. Our approach can assist embryologists in selecting better embryos more efficiently and with greater confidence.