Cytoplasmic Strings Analysis in Human Embryo Time-Lapse Videos using Deep Learning Framework

2025-12-10

AI Summary

In embryo viability assessment, manual annotation of cytoplasmic strands (CS) suffers from subjectivity, low efficiency, and susceptibility to visual artifacts. To address this, we introduce the first explainable, biologically validated CS temporal imaging dataset and propose a two-stage deep learning framework for frame-level CS classification and sub-pixel–level fine-grained localization. Methodologically, we design a novel uncertainty-aware contraction embedding (NUCE) loss that jointly enforces confidence-weighted regression and embedding-space compactness; integrate a Transformer backbone with the RF-DETR detection architecture; and adopt human-in-the-loop annotation and temporal sparse labeling strategies. Evaluated on 13,568 frames from real-time time-lapse imaging (TLI) videos, our approach achieves significant F1-score improvement via NUCE and sets a new state-of-the-art in CS localization accuracy—enabling, for the first time, robust automated detection of extremely fine, low-contrast CS structures.

Technology Category

• Computer Vision: Medical and Biological Imaging
• Machine Learning: Calibration & Uncertainty Quantification
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Security and Privacy: Data transparency and provenance

Abstract

Infertility is a major global health issue, and while in-vitro fertilization has improved treatment outcomes, embryo selection remains a critical bottleneck. Time-lapse imaging enables continuous, non-invasive monitoring of embryo development, yet most automated assessment methods rely solely on conventional morphokinetic features and overlook emerging biomarkers. Cytoplasmic Strings, thin filamentous structures connecting the inner cell mass and trophectoderm in expanded blastocysts, have been associated with faster blastocyst formation, higher blastocyst grades, and improved viability. However, CS assessment currently depends on manual visual inspection, which is labor-intensive, subjective, and severely affected by detection and subtle visual appearance. In this work, we present, to the best of our knowledge, the first computational framework for CS analysis in human IVF embryos. We first design a human-in-the-loop annotation pipeline to curate a biologically validated CS dataset from TLI videos, comprising 13,568 frames with highly sparse CS-positive instances. Building on this dataset, we propose a two-stage deep learning framework that (i) classifies CS presence at the frame level and (ii) localizes CS regions in positive cases. To address severe imbalance and feature uncertainty, we introduce the Novel Uncertainty-aware Contractive Embedding (NUCE) loss, which couples confidence-aware reweighting with an embedding contraction term to form compact, well-separated class clusters. NUCE consistently improves F1-score across five transformer backbones, while RF-DETR-based localization achieves state-of-the-art (SOTA) detection performance for thin, low-contrast CS structures. The source code will be made publicly available at: https://github.com/HamadYA/CS_Detection.

Problem

• Automates detection of cytoplasmic strings in embryo videos to improve IVF selection.
• Addresses manual inspection issues by using deep learning for objective analysis.
• Overcomes data imbalance and visual subtlety with a novel loss function.

Innovation

• Two-stage deep learning framework for classification and localization
• Novel Uncertainty-aware Contractive Embedding loss for imbalance
• Transformer backbones and RF-DETR for state-of-the-art detection