Multitasking Embedding for Embryo Blastocyst Grading Prediction (MEmEBG)

Nahid Khoshk Angabini; Mohsen Tajgardan; Mahesh Madhavan; Zahra Asghari Varzaneh; Reza Khoshkangini; Thomas Ebner

cs.CVApr 14, 2026

Multitasking Embedding for Embryo Blastocyst Grading Prediction (MEmEBG)

Authors: Nahid Khoshk Angabini, Mohsen Tajgardan, Mahesh Madhavan, Zahra Asghari Varzaneh, Reza Khoshkangini, Thomas Ebner

Organizations: ⋆Department of Computer Science and Media Technology, Malmö University, Malmö, Sweden · ◦Faculty of Electrical and Computer Engineering, Qom University of Technology, Iran · †Kepler Universitätsklinikum, Linz, Austria

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.

Figures & tables

Panel of Figure 1 — Fig. 1 : Example blastocyst image with segmented components: (a) original day-5 blastocyst, (b-d) corresponding masks highlighting the trophectoderm (TE), inner cell mass (ICM), and zona pellucida (ZP) regions.

Fig. 2 : The conceptual view of the proposed multitask embedding approach

	TE		ICM		EXP
Method	f-score & std	p-value	f-score & std	p-value	f-score & std	p-value
STL-TE	$0.60\pm 0.03$	$<0.05$	—	—	—	—
STL-ICM	—	—	$0.64\pm 0.03$	0.1	—	—
STL-EXP	—	—	—	—	$0.72\pm 0.04$	0.05
MTL-TE	$0.64\pm 0.02$	—	—	—	—	—
MTL-ICM	—	—	$0.63\pm 0.12$	—	—	—

TABLE I: Performance comparison between single-task learning (STL) and multitask learning (MTL) models for score grading prediction. Statistical significance was evaluated using the

5\times 2

cross-validated paired t-test.

Fig. 3 : The validation accuracy over the training process for ICM, TE, and EXP.

Fig. 4 : The confusion matrix from three different tasks.

Class	Precision	Recall	F1-score	Support
0	0.75	0.73	0.75	16
1	0.80	0.87	0.76	32
2	0.89	0.67	0.76	12
Accuracy			0.76	50
Macro Avg	0.78	0.76	0.76	50
Weighted Avg	0.78	0.76	0.76	50

TABLE II : Expansion Classification Performance

Class	Precision	Recall	F1-score	Support
A	0.80	0.84	0.85	37
B	0.50	0.38	0.44	13
Accuracy			0.76	50
Macro Avg	0.65	0.61	0.64	50
Weighted Avg	0.72	0.74	0.73	50

TABLE III : ICM Classification Performance.

Class	Precision	Recall	F1-score	Support
A	0.76	0.61	0.68	31
B	0.52	0.68	0.60	19
Accuracy			0.66	50
Macro Avg	0.64	0.65	0.64	50
Weighted Avg	0.67	0.66	0.65	50

TABLE IV : TE Classification Performance.

Automated Grading

Automated grading uses machine learning and language models to score student work against answer keys or rubrics and provide feedback.

See all 89 papers

Blasto-Net: An Explainable Multi-Task Learning for Blastocyst Segmentation, Grading, and Implantation Prediction
Jun 24, 2026Zahra Asghari Varzaneh, Reza Khoshkangini, Magnus Johnsson +2Clinical Outcome Prediction Multi-Task Learning
Department of Computer Science and Media Technology, Malmö University, Malmö, Sweden · Research Environment of Computer Science, Kristianstad University, Sweden · Kepler University Hospital, Krankenhausstr, Linz, Austria +1
Predicting Blastocyst Formation in IVF: Integrating DINOv2 and Attention-Based LSTM on Time-Lapse Embryo Images
Apr 14, 2026Zahra Asghari Varzaneh, Niclas Wölner-Hanssen, Reza Khoshkangini +2Microscopy Image Analysis Recurrent Neural Networks
Sustainable Digitalisation Research Center, Department of Computer Science and Media Technology, Malmö University, Malmö, Sweden · School of Information Technology, Halmstad University, Halmstad, Sweden · Kepler Universitätsklinikum, Linz, Austria +1
EMBRACE: A Multi-task Framework for Comprehensive Quality Assessment in Cleavage-stage Embryo
Jul 11, 2026Anwar Hussain Sofi, Jung-Hua Wang, Ming-Jer Chen +4Multi-Task Learning Microscopy Image Analysis
International Master Program in Applied Artificial Intelligence, National Taiwan Ocean University, Keelung, 202301, Taiwan · AI Research Center, National Taiwan Ocean University, Keelung 20224, Taiwan · Keelung, 202301, Taiwan +4

Jun 24, 2026J/K move · Enter open · S save

Blasto-Net: An Explainable Multi-Task Learning for Blastocyst Segmentation, Grading, and Implantation Prediction

Zahra Asghari Varzaneh, Reza Khoshkangini, Magnus Johnsson, Thomas Ebner+1

Department of Computer Science and Media Technology, Malmö University, Malmö, Sweden · Research Environment of Computer Science, Kristianstad University, Sweden · Kepler University Hospital, Krankenhausstr, Linz, Austria+1

Clinical Outcome Prediction Multi-Task Learning Medical Image Segmentation Explainable Medical Image Analysis

Open paper AskPDF arXiv

This study introduces Blasto-Net, a multi-task deep learning model for comprehensive blastocyst analysis. The proposed model performs three tasks simultaneously in a single forward pass: segmentation of the ZP, TE, and ICM compartments, morphological grading, and implantation outcome prediction. Accurate blastocyst analysis in in vitro fertilization (IVF) is challenging. The compartments often have similar textures but very different structures. To address these challenges, Blasto-Net employs an EfficientNet-B3 encoder with a UNet-style decoder enhanced by the Convolutional Block Attention Module (CBAM) and a novel Edge-Aware Attention Module (EAAM) to effectively capture both semantic and boundary information. To handle distinct compartment topologies, the network employs specialized segmentation heads and a composite region- and boundary-based loss. Additionally, Grad-CAM++ visualizations are used to verify the anatomical consistency of the model's predictions. Evaluated on a public HMC blastocyst dataset, Blasto-Net achieves Dice scores of 94.93%, 91.60%, and 88.82% for ICM, ZP, and TE, respectively, alongside an implantation F1-score of 80.0%. These results demonstrate that Blasto-Net offers an accurate, interpretable, and efficient solution for automated blastocyst assessment, with strong potential to support clinical decision-making in IVF.

Abstract

Automated Grading

Explore similar work

Blasto-Net: An Explainable Multi-Task Learning for Blastocyst Segmentation, Grading, and Implantation Prediction

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

EMBRACE: A Multi-task Framework for Comprehensive Quality Assessment in Cleavage-stage Embryo