Multitasking Embedding for Embryo Blastocyst Grading Prediction (MEmEBG)

arXiv cs.AI / 4/16/2026

💬 OpinionSignals & Early TrendsIdeas & Deep AnalysisModels & Research

共有:

Key Points

The paper addresses subjectivity and lack of standardization in current IVF blastocyst grading by targeting TE (trophectoderm), ICM (inner cell mass), and expansion (EXP) assessment from day-5 embryo images.
It proposes a multitask embedding-based approach that learns discriminative image representations using a pretrained ResNet-18 with an added embedding layer, aiming to improve separation of visually similar structures.
The method jointly performs automated region identification (TE and ICM) and grade prediction, leveraging both biological/physical cues from the images.
The study reports experimental results suggesting the approach can deliver more reliable and consistent blastocyst quality assessment, particularly on limited datasets.
Overall, it presents a research prototype intended to support robust quality assurance in embryo evaluation workflows used in IVF clinics.

Abstract

Reliable evaluation of blastocyst quality is critical for the success of in vitro fertilization (IVF) treatments. Current embryo grading practices primarily rely on visual assessment of morphological features, which introduces subjectivity, inter-embryologist variability, and challenges in standardizing quality assurance. In this study, we propose a multitask embedding-based approach for the automated analysis and prediction of key blastocyst components, including the trophectoderm (TE), inner cell mass (ICM), and blastocyst expansion (EXP). The method leverages biological and physical characteristics extracted from images of day-5 human embryos. A pretrained ResNet-18 architecture, enhanced with an embedding layer, is employed to learn discriminative representations from a limited dataset and to automatically identify TE and ICM regions along with their corresponding grades, structures that are visually similar and inherently difficult to distinguish. Experimental results demonstrate the promise of the multitask embedding approach and potential for robust and consistent blastocyst quality assessment.