AI Embryo Image Screening for Non-Invasive Aneuploidy Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current pre-implantation genetic screening (PGT-A) methods for embryos in IVF are invasive, time-consuming, and have reduced accuracy due to embryo mosaicism, leading to uncertain embryo development effects and extended time to pregnancy.
Innovation Solution
A non-invasive AI model is developed using computer vision and deep learning to analyze embryo images for aneuploidy, employing chromosomal group labels, training datasets, and hierarchical layered models to identify morphological features and provide rapid aneuploidy screening.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If PGT-A biopsy is performed to obtain genetic information from embryos, then genetic screening accuracy is improved, but embryo development reliability deteriorates due to invasive cell removal
Solution Approach 1:
The patent replaces the mechanical biopsy procedure with an optical imaging system. The AI model analyzes morphological features from time-lapse images of embryos, substituting physical cell removal with non-invasive visual assessment. This eliminates the harmful mechanical intervention while maintaining screening capability through pattern recognition in image data.
Solution Approach 2:
The patent creates a virtual model of the embryo based on its visual appearance in images. The AI model generates predictions about chromosomal status by analyzing morphological patterns in the image copy, rather than requiring physical sampling. This virtual representation allows genetic assessment without touching the actual embryo.
2Loss of information
If PGT-A biopsy is performed to screen embryos, then genetic information is obtained, but time to pregnancy is extended due to laboratory transport and freezing delays
Solution Approach 1:
The patent performs genetic screening assessment during the embryo culture period itself, before implantation. The AI model analyzes images captured throughout the standard 3-5 day culture process, providing genetic information simultaneously with morphological assessment. This eliminates the need for post-biopsy waiting periods and freezing delays, as screening is integrated into the existing culture timeline.
Solution Approach 2:
The patent maintains continuous monitoring of embryos through time-lapse imaging during culture. The AI model processes images captured throughout the entire culture period, providing ongoing assessment without interrupting the embryo development process. This continuous action eliminates the breaks and delays inherent in biopsy-based workflows.
3Loss of information
If PGT-A biopsy is performed on mosaic embryos, then genetic screening is attempted, but measurement precision deteriorates due to unrepresentative cell sampling
Solution Approach 1:
The patent replaces cellular-level sampling with whole-embryo visual assessment. The AI model analyzes morphological patterns that reflect the overall chromosomal status of the entire embryo, rather than sampling individual cells that may not be representative. This shifts the measurement from discrete cell samples to integrated embryo-level characteristics.
Solution Approach 2:
The patent creates a screening method that works universally for all embryos regardless of mosaic status. The AI model's morphological analysis approach does not depend on the uniformity of chromosomal distribution across cells, making it equally effective for euploid, aneuploid, and mosaic embryos. This universal applicability overcomes the sampling problem inherent in biopsy methods.
Data Source
AI summary
An Artificial Intelligence (AI) based computational system is used to non-invasively estimate the presence of a range of aneuploidies and mosaicism in an image of embryo prior to implantation. Aneuploidies and mosaicism with similar risks of adverse outcomes are grouped and training images are labelled with their group. Separate AI models are trained for each group using the same training dataset and the separate models are then combined, such as by using an Ensemble or Distillation approach to develop a model that can identify a wide range of aneuploidy and mosaicism risks. The AI model for a group is generated by training multiple models including binary models, hierarchical layered models and a multi-class model. In particular the hierarchical layered models are generated by assigning quality labels to images. At each layer the training set is partitioned in the best quality images and other images. The model at that layer is trained on the best quality images, and the other images are passed down to the next layer and the process repeated (so the remaining images are separated into next best quality images and other images). The final model can then be used to non-invasively identify aneuploidy and mosaicism and associated risk of adverse outcomes from an image of an embryo prior to implantation.