Automatic characterization of human embryos at day 4 post-insemination from time-lapse imaging using supervised contrastive learning and inductive transfer learning techniques

Payá, Elena; Bori, Lorena; Colomer, Adrián; Meseguer, M.; Naranjo, Valery

doi:10.1016/j.cmpb.2022.106895

Cited by 12 publications

(5 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This challenge is the most addressed regarding XAI. Some studies apply Grad-CAM for visualization [32,42] and observe that key regions that the model relies on seem to be consistent with clinical interpretation.…”

Section: Review On Current Workmentioning

confidence: 86%

Exploring the Role of Explainability in AI-Assisted Embryo Selection

Urcelay,

Hinjos,

Martin-Torres

et al. 2023

Frontiers in Artificial Intelligence and Applications

View full text Add to dashboard Cite

In Vitro Fertilization is among the most widespread treatments for infertility. One of its main challenges is the evaluation and selection of embryo for implantation, a process with large inter- and intra-clinician variability. Deep learning based methods are gaining attention, but its opaque nature compromises their acceptance in the clinical context, where transparency in the decision making is key. In this paper we analyze the current work in the explainability of AI-assisted embryo analysis models, identifying the limitations. We also discuss how these models could be integrated in the clinical context as decision support systems, considering the needs of clinicians and patients. Finally, we propose guidelines for the sake of increasing interpretability and trustworthiness, pushing this technology forward towards established clinical practice.

show abstract

Section: Review On Current Workmentioning

confidence: 86%

Exploring the Role of Explainability in AI-Assisted Embryo Selection

Urcelay,

Hinjos,

Martin-Torres

et al. 2023

Frontiers in Artificial Intelligence and Applications

View full text Add to dashboard Cite

show abstract

“…Our stage classification models also showed that MobileNetV2 pre-trained on ImageNet can give high accuracies on our dataset even with very small amounts of training data for simple classification tasks. Other studies have also used ImageNet for classifying human embryos 7,9,10,12,13,[17][18][19]21,24 , however our work is the first to investigate the performance that can be achieved with very little embryo data. Our data is quite different from the typical images included in ImageNet, which is mainly composed of photographs of everyday objects and animals rather than medical images.…”

Section: Discussionmentioning

confidence: 99%

“…Most previous attempts to train ML models to assess embryos from raw time-lapse data aim to classify embryos based on intermediate outcomes such as manual grading by embryologists or embryo aneuploidy, or early pregnancy outcomes after embryo transfer such as foetal heartbeat 7,[9][10][11][12][13][14][15][16][17][18][19][20][21] . Only a few have been designed with the outcome of live birth [22][23][24] .…”

Section: Main Textmentioning

confidence: 99%

Deep learning pipeline reveals key moments in human embryonic development predictive of live birth in IVF

Mapstone,

Hunter,

Brison

et al. 2023

Preprint

View full text Add to dashboard Cite

Demand for IVF treatment is growing, however success rates remain low partly due to the difficulty in selecting the best embryo to be transferred. Current manual assessments are subjective and can lead to significant inter-operator variability. Deep learning techniques could lead to improved embryo assessment and live birth prediction, however previous attempts neglect early developmental stages and often require vast amounts of data. Here, we demonstrate that even with limited data it is possible to train convolutional neural networks to classify developmental stage at high accuracies and predict live birth from various time-points throughout development. We identify key windows that are optimal for assessing embryo viability and demonstrate the importance of incorporating information from earlier stages. Our outcome predictor models are competitive with, and potentially outperform, human expert selection. The pipeline produced here could lead to an improved, standardised approach to embryo selection compatible with multiple transfer strategies.

show abstract

“…In this case, we can conclude that the blastocyst morphology analysis has more efects than time series analysis on the abnormality detection. Furthermore, Payá et al [40] proposed a supervised contrastive learning framework for grading and anomaly detection of embryos which achieved 0.94 AUC in abnormality detection. Nevertheless, this comparison is superfcial and unreliable because the researchers used diferent databases in terms of size and number of items.…”

Section: Early Diagnosis Of Embryo Anomalymentioning

confidence: 99%

Local Binary Convolutional Neural Networks' Long Short-Term Memory Model for Human Embryos' Anomaly Detection

Einy

Sen

Saygın

et al. 2023

Scientific Programming

View full text Add to dashboard Cite

Accurate selection of embryos with the maximum implementation condition is a necessary step to increase the effectiveness of fertility treatment in in vitro fertilization (IVF). The deep learning algorithms presented high potential for monitoring and visualizing embryo features such as cell numbers and their morphological development in time series manner. Due to the ability of the computer vision and deep learning algorithms, this paper aimed to present a novel deep learning approach to distinguish simultaneous abnormality of embryos in time-lapse systems for detecting live and non-live births in IVF. The approach is composed of local binary convolutional neural network (LBCNN) and long short-term memory (LSTM). The LBCNN improved accuracy of classification by employing deep and local feature sets with lower number of learnable parameters in comparison with a standard convolutional layer. Moreover, LSTM network is employed to analyze temporal information of time-lapse embryos. The results indicate that the proposed approach achieves significant results in ROC analysis (0.98) in 5 days of monitoring compared to state-of-the-art models. In addition, the approach showed compatible results in early diagnosis of abnormality detection (72 hours) with 82.8% accuracy of classification compared to the pretrained well-known convolutional neural network (CNN) models and baseline CNN.

show abstract

Automatic characterization of human embryos at day 4 post-insemination from time-lapse imaging using supervised contrastive learning and inductive transfer learning techniques

Cited by 12 publications

References 33 publications

Exploring the Role of Explainability in AI-Assisted Embryo Selection

Exploring the Role of Explainability in AI-Assisted Embryo Selection

Deep learning pipeline reveals key moments in human embryonic development predictive of live birth in IVF

Local Binary Convolutional Neural Networks' Long Short-Term Memory Model for Human Embryos' Anomaly Detection

Contact Info

Product

Resources

About