Pith. sign in

REVIEW 3 major objections 4 minor 1 cited by

Merging synthetic and real embryo data for advanced AI predictions

T0 review · 3 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Adding synthetic embryo images to real training data pushes AI classification accuracy from 94.5 to 97 percent, and the gain persists on a different clinic's images.

desk verdict Useful public dataset and plausible synthetic-data augmentation result, but the external validation is uninterpretable as written and the headline gain is selected on the test set. read the letter →

arxiv 2412.01255 v2 pith:ARETPKHZ submitted 2024-12-02 eess.IV cs.CV

Oriana Presacan , Alexandru Dorobantiu , Vajira Thambawita , Michael A. Riegler , Mette H. Stensen , Mario Iliceto , Alexandru C. Aldea , Akriti Sharma This is my paper · ORCID
classification eess.IVcs.CV
keywords embryomorphologyassessmentsyntheticdataaugmentationlatentdiffusionmodelStyleGANcell-stageclassificationassistedreproductivetechnologydeeplearningscarcity
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper claims that adding synthetic embryo images to real training data improves a deep-learning classifier's ability to recognize five developmental stages (2-cell, 4-cell, 8-cell, morula, and blastocyst). The best mix—real images plus synthetic images from both a latent diffusion model and a StyleGAN—raises accuracy to 97 percent on the authors' test set, versus 94.5 percent for real data alone. The same trend appears on an external blastocyst dataset from a different clinic, and training solely on synthetic images still reaches 92 percent accuracy. The authors also present a Turing test in which four embryologists failed to identify 66.6 percent of diffusion-generated images as fake, compared with 25.3 percent for GAN images.

What carries the argument

The mechanism is augmentation with two families of generative models. A latent diffusion model (LDM) iteratively denoises a compressed latent representation, while StyleGAN uses an adversarial generator with a mapping network to control style; each is trained separately per cell stage on 1,000 real images, and the best checkpoint by Fréchet inception distance (FID) produces 5,000 synthetic images per class. These are combined with the real training set to fine-tune VGG16, ResNet50, and ViT classifiers, and the synthetic images are also shown to four embryologists in a Turing test. The argument depends on the hypothesis that two generators with different failure modes yield a more diverse synthetic set than either alone.

What would settle it

Train the classifier on 1,000 real images plus 8,000 extra real frames from the same time-lapse sequences (matching the total volume of the best synthetic mix); if accuracy matches the reported 97 percent, the improvement is data volume rather than synthetic-data content.

Watch

Extended reading notes

Core claim

The central discovery is that synthetic embryo images carry real training signal, not just visual plausibility. Classifiers trained on synthetic images alone reach 92 percent accuracy on real test images, and adding synthetic images to real data improves top accuracy from 94.5 to 97 percent on the authors' held-out test set. The gain is largest when synthetic images come from both a latent diffusion model and a StyleGAN, supporting the claim that different generation processes contribute complementary features and create a more diverse training set. The same pattern appears on an external blastocyst dataset: the best synthetic-augmented configuration reached 84.69 percent accuracy versus 68.88 percent for real-only training.

Load-bearing premise

The external generalization claim rests on treating a 98-image subset of another clinic's blastocyst dataset, labeled only with quality grades, as ground truth for five cell-stage classes.

Editorial extensions

If this is right

  • Clinics with small local datasets could train per-stage generative models to create large synthetic training sets, reducing the need to share sensitive patient images.
  • The combination effect suggests that curating synthetic data from more than one architecture is a general recipe for improving image classifiers, not a quirk of these specific models.
  • The diffusion model's higher deception rate in the Turing test makes LDM-generated images the stronger candidate for applications requiring human-like fidelity.
  • Synthetic-only training reaching 92 percent accuracy implies that a purely synthetic pretraining stage could bootstrap classifiers in settings where no local real data exist.
  • The external-dataset gain, if the label mapping is valid, indicates that synthetic augmentation helps classifiers transfer across clinics and imaging systems.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The reported gain over real-only training does not hold total data volume fixed; comparing against real-and-extra-real at equal volume would separate the effect of synthetic content from the effect of simply having more images.
  • The external validation uses a 98-image subset labeled by blastocyst quality, not cell stage, so the cross-clinic accuracies in Table 5 may be measuring a different classification task than the five-stage claim.
  • Embryologist comments singled out GAN artifacts (unusual dots, flattened cells), so the added value of combining generators likely comes from the diffusion model compensating for GAN-specific flaws; ablating those artifacts would test this.
  • A practical next step is to repeat the augmentation recipe on embryos from several clinics and incubator models with proper five-stage annotation, which would tell whether the cross-clinic benefit holds beyond this single study.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The manuscript trains per-stage StyleGAN and latent diffusion models on 1,000 real embryo images per class (2-cell, 4-cell, 8-cell, morula, blastocyst), generates 5,000 synthetic images per class per model, and then trains VGG16, ResNet50, and ViT classifiers on combinations of real and synthetic images. On an internal held-out set of 100 real images per class, the authors report that adding synthetic images raises accuracy from about 94.5% to 97% for fine-tuned VGG, that synthetic-only training reaches around 92-93%, and that combining both generators outperforms either alone. They also report accuracy on an external 98-image STORK blastocyst subset and a Turing test in which four embryologists judged image realism. The paper releases the real and synthetic datasets, code, and checkpoints.

Significance. If the results are valid, the main contribution is a practical demonstration that synthetic embryo images from two generator families can improve a five-class cell-stage classifier, together with a public dataset release that would be useful for the ART-AI community. The manuscript has real strengths: the internal test split is made at sequence level; VGG results are averaged over five seeds with standard deviations; code, checkpoints, and data are publicly available; and the Turing test is described with per-stage detail. However, two load-bearing parts of the paper need attention before the contribution can be accepted: the external validation labels do not match the classifier's task, and the headline configuration was selected after inspecting test-set numbers across many data combinations. The internal claim that synthetic data helps remains plausible, but the evidence as written is weaker than the abstract suggests.

major comments (3)
  1. [Materials and methods, 'To ensure robustness...' paragraph; Results, Table 5] The external subset is described as 98 labeled images 'annotated by embryologists with quality labels: good-quality, fair-quality, and poor-quality.' These are blastocyst quality grades, not the five cell-stage labels (2-cell, 4-cell, 8-cell, morula, blastocyst) used to train and test the classifiers. No mapping from quality labels to stage labels is provided. If the 98 images are all blastocysts, the 'accuracy' reported in Table 5 is only the blastocyst-class recall of the model on a single-class set; it does not test five-way cell-stage discrimination and cannot support the abstract claim that the improvement 'remained consistent when tested on an external Blastocyst dataset.' Stage labels or a redefinition of the external evaluation task are needed.
  2. [Results, Figures 3-4 and Tables S3-S4] The headline 97% versus 94.5% is the maximum over a large grid of synthetic-data amounts and generator combinations, all evaluated on the same held-out test set of 500 images. The same figures are used to choose the 'best' configuration; no separate validation split is reported. This selection on the test set inflates the observed gain, and the standard deviations of the chosen configuration do not account for the multiple comparisons. The same issue applies to selecting 1000-4000-4000 as the best external configuration in Table 5. The authors should either use a nested or held-out validation split to choose the configuration and then report test accuracy once, or report all configurations with an explicit multiple-testing correction.
  3. [Generative Models, FID checkpoint selection; Discussion, limitations] Checkpoint selection by FID is computed against the same 1,000 real training images per stage, and these same real images are also used for classifier training. The paper explicitly states that it did not measure similarity between real and synthetic images. Under these conditions, synthetic images that memorize or closely copy training images could improve classifier accuracy simply by upweighting the training set, rather than by adding morphological diversity. The synthetic-only result is still informative, but the comparison 'real + synthetic' versus 'real only' needs a control, such as adding the same number of augmented real images or reporting an image-retrieval or duplicate-detection analysis between generated and real training images.
minor comments (4)
  1. [Table 4] The column header 'Pre-trained' is ambiguous, and the rows do not clearly indicate whether the reported numbers refer to from-scratch or fine-tuned training; the caption should state this explicitly.
  2. [Introduction] There is a typo in 'addversarial networks' (should be 'adversarial'), and 'V olvat' appears with an unintended space in several places throughout the text.
  3. [Supplementary Tables S5 and S6] These tables are formatted as long single-line rows without visible column separators, which makes them difficult to read; they should be reformatted with proper separate columns.
  4. [Materials and methods, external dataset description] The exact selection procedure for the 98-image STORK subset, including whether it was random, the stage or quality distribution, and the filtering criteria, is not given; this information should be added.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the classification gains are measured on held-out real and external images, and no fitted parameter is renamed as a prediction.

full rationale

The central claim—that adding synthetic images to real training data improves embryo cell-stage classification—rests on classification accuracy computed on a separate held-out set of 100 real images per stage, not on the images used to train the generative models or on a fitted parameter. The external-validation leg uses 98 images from the STORK blastocyst dataset, which is an independent source; although those images are annotated with quality labels rather than the five cell-stage labels, the reported accuracy is still computed from the model's five-class outputs and is not equivalent to the training inputs by construction. The label mismatch is a validity and interpretability concern, not a circular reduction. The FID-based checkpoint selection is performed against the same 1,000 real training images used for the generative models, but FID is a generation-quality monitor and is not substituted for the classifier accuracy measurements; this may weaken claims about synthetic-image novelty or out-of-sample diversity, but it does not make the reported classification result a fitted quantity. The only self-citation, reference 5 (Riegler et al.), is background context and is not load-bearing. The Discussion explicitly acknowledges that the authors did not assess synthetic-to-real similarity to rule out generative overfitting, which is an honest stated limitation rather than a hidden circular step. No equation in the paper reduces the reported predictions to the training labels or to a quantity fitted from those labels, so no circular step can be exhibited with the required specificity.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The central results rest on standard supervised learning assumptions plus domain-specific choices: label fidelity, central-plane representativeness, FID as an in-sample selection metric, a 15% fragmentation cutoff, and an unexplained mapping of external quality labels to cell-stage labels. The synthetic-data claim also assumes generated images are not memorized copies of training images, which the authors note they did not measure. No new physical or conceptual entities are introduced.

free parameters (5)
  • Number of synthetic images per stage per model = 5,000
    Hand-chosen maximum; authors report diminishing returns above 4,000+4,000 and no statistically significant difference between 2,000+2,000 and 5,000+5,000.
  • Fragmentation cutoff = 15%
    Embryos with fragmentation above 15% were excluded, which simplifies the classification task and may overestimate performance; the authors acknowledge this.
  • Number of real images per stage for generator and classifier training = 1,000
    Out of 1,100 images per stage, 1,000 are used for generative training and also for classifier training, with 100 held out for testing.
  • LDM learning rate and batch size = 2e-6, batch size 16
    Chosen through repository code review and empirical testing rather than a stated systematic selection procedure.
  • StyleGAN initialization = Pre-trained FFHQ weights
    Training from scratch was unsatisfactory, so transfer learning from a face dataset was used; this introduces a domain-shift assumption.
assumptions (5)
  • domain assumption Embryologist cell-stage annotations and timings are accurate ground truth.
    Labels come from clinical annotations at Volvat Spiren and from Gomez et al.; no inter-observer variability is reported for this dataset.
  • domain assumption Central focal-plane frames represent the embryo's developmental stage even when the embryo is not centered or is occluded.
    The methods state that images may be partially occluded but still represent the stage; this is not verified.
  • domain assumption FID computed on the generative training set is a valid criterion for selecting and comparing generative models.
    FID is used for checkpoint selection and to claim diffusion-model superiority, but it is computed against the same real images used to train the generators.
  • domain assumption Synthetic images do not simply copy real training images.
    The authors state that no similarity metric was computed to rule out overfitting or memorization; external improvement is indirect evidence only.
  • ad hoc to paper The external STORK subset's quality labels (good/fair/poor) can serve as cell-stage ground truth.
    No mapping or relabeling procedure is described, making the external evaluation in Table 5 difficult to interpret.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Merging synthetic and real embryo data for advanced AI predictions." pith.science (2026). https://pith.science/paper/ARETPKHZ

@misc{pith2026241201255,
  author       = {Pith},
  title        = {Pith review of: Merging synthetic and real embryo data for advanced AI predictions},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ARETPKHZ}},
  note         = {Machine review of arXiv:2412.01255}
}
read the original abstract

Accurate embryo morphology assessment is essential in assisted reproductive technology for selecting the most viable embryo. Artificial intelligence has the potential to enhance this process. However, the limited availability of embryo data presents challenges for training deep learning models. To address this, we trained two generative models using two datasets-one we created and made publicly available, and one existing public dataset-to generate synthetic embryo images at various cell stages, including 2-cell, 4-cell, 8-cell, morula, and blastocyst. These were combined with real images to train classification models for embryo cell stage prediction. Our results demonstrate that incorporating synthetic images alongside real data improved classification performance, with the model achieving 97% accuracy compared to 94.5% when trained solely on real data. This trend remained consistent when tested on an external Blastocyst dataset from a different clinic. Notably, even when trained exclusively on synthetic data and tested on real data, the model achieved a high accuracy of 92%. Furthermore, combining synthetic data from both generative models yielded better classification results than using data from a single generative model. Four embryologists evaluated the fidelity of the synthetic images through a Turing test, during which they annotated inaccuracies and offered feedback. The analysis showed the diffusion model outperformed the generative adversarial network, deceiving embryologists 66.6% versus 25.3% and achieving lower Frechet inception distance scores.

Figures

Figures reproduced from arXiv: 2412.01255 by the authors.

Figure 1
Figure 1. Pipeline of our proposed system, encompassing the training of generative models, the generation of synthetic data, the training of classification models, and the qualitative assessment conducted by embryologists. latent space representation via a mapping network. This representation is then adjusted to manipulate the generated image’s style at various resolutions, employing a sequence of convolutional layers to enha… view at source ↗
Figure 2
Figure 2. Comparison of real versus synthetic images generated with StyleGAN and LDM models for each of the five embryo classes: 2-cell, 4-cell, 8-cell, morula, and blastocyst. Classification results [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Classification accuracy trends on test data (100 real images) for the VGG model, trained with various combinations of synthetic images generated by LDM and StyleGAN models [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Classification accuracy trends on test data (100 real images) trained with various combinations of real and synthetic data generated by LDM and StyleGAN models. 1,000 real images in addition to the synthetic ones. When training exclusively on real images, an accuracy o…
Figure 5
Figure 5. Figure 5: Accuracy differences between training the VGG model from scratch versus using a pre-trained model, based on the same data combinations as in [PITH_FULL_IMAGE:figures/full_fig_p010_5.png]
Figure 6
Figure 6. Figure 6: Average Turing test results conducted by embryologists, evaluating their accuracy in identifying real and synthetic images. The "Real Images" pie chart displays the proportion of images correctly identified as real versus incorrectly labeled as fake. The "Synthetic Ima…
Figure 7
Figure 7. Figure 7: Annotations by embryologists highlighting perceived inaccuracies in images identified as synthetic. The first two images (a) and (b) depict real images, the following one (c) is generated using the LDM model, and the last three images (d), (e), and (f) are produced wit…
Figure 8
Figure 8. Figure 8: Examples of embryos misclassified by the real-data-only model but correctly classified with real and synthetic data. True (black) and misclassified (red) classes are shown below each image. approach in both FID scores and visual quality. However, this comparison was li…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. BlastDiffusion: A Latent Diffusion Model for Generating Synthetic Embryo Images to Address Data Scarcity in In Vitro Fertilization

    q-bio.QM 2025-06 conditional novelty 5.0 of 10

    A conditional latent diffusion model generates synthetic oocyte images that score closer to real images than a GAN baseline, with better FID, LPIPS, and SSIM.

Reference graph

Works this paper leans on

55 extracted references · 40 canonical work pages · cited by 1 Pith paper

  1. [1]

    Infertility

    World Health Organization. Infertility. https://www.who.int/news-room/fact-sheets/detail/infertility (2024)

    work page 2024

  2. [2]

    ART in Europe, 2019: results generated from European registries by ESHRE†

    The European IVF Monitoring Consortium (EIM) for the European Society of Human Reproduction and Embryology (ESHRE) et al. ART in Europe, 2019: results generated from European registries by ESHRE†. Hum. Reproduction 38, 2321–2338, DOI: 10.1093/humrep/dead197 (2023)

    work page doi:10.1093/humrep/dead197 2023

  3. [3]

    J., Breth Knudsen, U

    Sundvall, L., Ingerslev, H. J., Breth Knudsen, U. & Kirkegaard, K. Inter- and intra-observer variability of time-lapse annotations. Hum. Reproduction 28, 3215–3221 (2013)

    work page 2013

  4. [4]

    & Andershed, A

    Adolfsson, E. & Andershed, A. Morphology vs morphokinetics: A retrospective comparison of interobserver and intra- observer agreement between embryologists on blastocysts with known implantation outcome. JBRA assisted reproduction 22, DOI: 10.5935/1518-0557.20180042 (2018)

    work page arXiv 2018

  5. [5]

    Riegler, M. A. et al. Artificial intelligence in the fertility clinic: status, pitfalls and possibilities. Hum. Reproduction 36, 2429–2442, DOI: 10.1093/humrep/deab168 (2021)

    work page doi:10.1093/humrep/deab168 2021

  6. [6]

    Gliozheni, O. et al. ART in Europe, 2018: results generated from European registries by ESHRE. Hum. Reproduction Open 2022, DOI: 10.1093/hropen/hoac022 (2022)

    work page doi:10.1093/hropen/hoac022 2022

  7. [7]

    & Janssens, R

    Sciorio, R., Aiello, R. & Janssens, R. Considerations on staffing levels for a modern assisted reproductive laboratory. JBRA Assist. Reproduction DOI: 10.5935/1518-0557.20220048 (2022)

    work page arXiv 2022

  8. [8]

    Thirumalaraju, P. et al. Evaluation of deep convolutional neural networks in classifying human embryo images based on their morphological quality. Heliyon 7, e06298, DOI: https://doi.org/10.1016/j.heliyon.2021.e06298 (2021)

    work page doi:10.1016/j.heliyon.2021.e06298 2021

Show all 55 references
  1. [9]

    Danardono, G. B. et al. A homogeneous ensemble of robust pre-defined neural network enables automated annotation of human embryo morphokinetics. J. Reprod. Infertil. 23, 250–256 (2022)

    2022

  2. [10]

    Dimitriadis, I. et al. Deep convolutional neural networks (CNN) for assessment and selection of normally fertilized human embryos. Fertility Steril. 112, e272, DOI: https://doi.org/10.1016/j.fertnstert.2019.07.805 (2019). 2019 ASRM Abstract Issue

    2019 doi

  3. [11]

    Liu, Z. et al. Multi-Task Deep Learning With Dynamic Programming for Embryo Early Development Stage Classification From Time-Lapse Videos. IEEE Access 7, 122153–122163, DOI: 10.1109/ACCESS.2019.2937765 (2019)

    2019

  4. [12]

    Lucio, C. M. et al. CHLOE (FAIRTILITY) can automatically annotate images from time-lapse cultured embryos for Pronucleate (PN) count, Morphokinetics and Ranking according to ploidy and implantation potential, with a strong agreement compared to experienced embryologists. Repro...

    2022 doi

  5. [13]

    & Berntsen, J

    Theilgaard Lassen, J., Fly Kragh, M., Rimestad, J., Nygård Johansen, M. & Berntsen, J. Development and validation of deep learning based embryo selection across multiple days of transfer. Sci. Reports 13, DOI: 10.1038/s41598-023-31136-3 (2023)

    2023 doi

  6. [14]

    Artificial Intelligence in the IVF Laboratory: A Review of Advancements Over the Last Decade.Fertility Steril

    Jiang, V . Artificial Intelligence in the IVF Laboratory: A Review of Advancements Over the Last Decade.Fertility Steril. 120, DOI: 10.1016/j.fertnstert.2023.05.149 (2023)

    2023 doi

  7. [15]

    Bormann, C. et al. Consistency and objectivity of automated embryo assessments using deep neural networks. Fertility Steril. 113, 781–787.e1, DOI: 10.1016/j.fertnstert.2019.12.004 (2020)

    2020 doi

  8. [16]

    Goled, S. Why Transformers Are Increasingly Becoming As Important As RNN And CNN? https://analyticsindiamag.co m/why-transformers-are-increasingly-becoming-as-important-as-rnn-and-cnn/ (2021)

    2021

  9. [17]

    Cruz, M. et al. Embryo quality, blastocyst and ongoing pregnancy rates in oocyte donation patients whose embryos were monitored by time-lapse imaging. J. assisted reproduction genetics 28, 569–73, DOI: 10.1007/s10815-011-9549-1 (2011)

    2011 doi

  10. [18]

    Gilbert, S. F. Early Mammalian Development (Sinauer Associates, 2000), 6th edn. Retrieved 13 May 2022. 15/17

    2000

  11. [19]

    Atlas of human embryology: from oocytes to preimplantation embryos

    European Society of Human Reproduction and Embryology (ESHRE). Atlas of human embryology: from oocytes to preimplantation embryos. https://atlas.eshre.eu/ (2016)

    2016

  12. [20]

    Martins, W. P.et al. Blastocyst vs cleavage-stage embryo transfer: systematic review and meta-analysis of reproductive outcomes. Ultrasound Obstet. &; Gynecol. 49, 583–591, DOI: 10.1002/uog.17327 (2017)

    2017 doi

  13. [21]

    & Ziebe, S

    Lemmen, J., Agerholm, I. & Ziebe, S. Kinetic markers of human embryo quality using time-lapse recordings of ivf/icsi- fertilized oocytes. Reproductive BioMedicine Online 17, 385–391, DOI: https://doi.org/10.1016/S1472-6483(10)60222-2 (2008)

    2008 doi

  14. [22]

    Urcelay, L. et al. Exploring the role of explainability in ai-assisted embryo selection (2023). 2308.02534

    2023 arXiv

  15. [23]

    Salih, M. et al. Embryo selection through artificial intelligence versus embryologists: a systematic review. Hum. reproduction open 2023, hoad031, DOI: 10.1093/hropen/hoad031 (2023)

    2023 doi

  16. [24]

    Khosravi, P. et al. Deep learning enables robust assessment and selection of human blastocysts after in vitro fertilization. NPJ Digit. Med. 2, 21, DOI: 10.1038/s41746-019-0096-y (2019)

    2019 doi

  17. [25]

    K., Thirumalaraju, P., Kandula, H.et al

    Kanakasabapathy, M. K., Thirumalaraju, P., Kandula, H.et al. Adaptive adversarial neural networks for the analysis of lossy and domain-shifted datasets of medical images. Nat. Biomed. Eng. 5, 571–585, DOI: 10.1038/s41551-021-00733-w (2021)

    2021 doi

  18. [26]

    An annotated human blastocyst dataset to benchmark deep learning architectures for in vitro fertilization

    Kromp, F., Wagner, R., Balaban, B.et al. An annotated human blastocyst dataset to benchmark deep learning architectures for in vitro fertilization. Sci. Data 10, 271, DOI: 10.1038/s41597-023-02182-3 (2023)

    2023 doi

  19. [27]

    Gomez, T. et al. A time-lapse embryo dataset for morphokinetic parameter prediction. Data Brief 42, 108258, DOI: https://doi.org/10.1016/j.dib.2022.108258 (2022)

    2022

  20. [28]

    Sheller, M. J. et al. Federated learning in medicine: facilitating multi-institutional collaborations without sharing patient data. Sci. Reports 10, DOI: 10.1038/s41598-020-69250-1 (2020)

    2020 doi

  21. [29]

    Murtaza, H. et al. Synthetic data generation: State of the art in health care domain. Comput. Sci. Rev. 48, 100546, DOI: 10.1016/j.cosrev.2023.100546 (2023)

    2023

  22. [30]

    & Chintala, S

    Radford, A., Metz, L. & Chintala, S. Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks (2016). 1511.06434

    2016 arXiv

  23. [31]

    & Agee, J

    Behara, K., Bhero, E. & Agee, J. T. Skin lesion synthesis and classification using an improved dcgan classifier.Diagnostics 13, DOI: 10.3390/diagnostics13162635 (2023)

    2023 doi

  24. [32]

    P., Sarada, V

    Porkodi, S. P., Sarada, V . & Maik, V . Dcgan for data augmentation in pneumonia chest x-ray image classification. In Mahapatra, R. P., Peddoju, S. K., Roy, S. & Parwekar, P. (eds.)Proceedings of International Conference on Recent Trends in Computing, 129–137 (Springer Nature ...

    2023

  25. [33]

    & Aila, T

    Karras, T., Laine, S. & Aila, T. A Style-Based Generator Architecture for Generative Adversarial Networks (2019). 1812.04948

    2019 arXiv

  26. [34]

    & Ommer, B

    Rombach, R., Blattmann, A., Lorenz, D., Esser, P. & Ommer, B. High-Resolution Image Synthesis with Latent Diffusion Models (2022). 2112.10752

    2022 arXiv

  27. [35]

    & Schwenker, F

    Abdalaziz, A. & Schwenker, F. Generating Synthetic Brain Tumor Data Using StyleGAN3 for Lower Class Enhancement , 73–87 (Springer Nature Switzerland, 2024)

    2024

  28. [36]

    Hong, S. et al. 3D-StyleGAN: A Style-Based Generative Adversarial Network for Generative Modeling of Three- Dimensional Medical Images (2021). 2107.09700

    2021 arXiv

  29. [37]

    Pinaya, W. H. L. et al. Brain imaging generation with latent diffusion models. In Mukhopadhyay, A., Oksuz, I., Engelhardt, S., Zhu, D. & Yuan, Y . (eds.)Deep Generative Models, 117–126 (Springer Nature Switzerland, Cham, 2022)

    2022

  30. [38]

    Ivanovs, M. et al. Synthetic Image Generation With a Fine-Tuned Latent Diffusion Model for Organ on Chip Cell Image Classification. In 2023 Signal Processing: Algorithms, Architectures, Arrangements, and Applications (SPA) , 148–153, DOI: 10.23919/SPA59660.2023.10274460 (2023)

    2023

  31. [39]

    & Scherer, R

    Dirvanauskas, D., Maskeli¯unas, R., Raudonis, V ., Damaševiˇcius, R. & Scherer, R. HEMIGEN: Human Embryo Image Generator Based on Generative Adversarial Networks. Sensors 19, 3578, DOI: 10.3390/s19163578 (2019)

    2019 doi

  32. [40]

    Cao, P. et al. Generative artificial intelligence to produce high-fidelity blastocyst-stage embryo images. Hum. Reproduction 39, 1197–1207, DOI: 10.1093/humrep/deae064 (2024)

    2024 doi

  33. [41]

    Chae, H. J. et al. P-182 high-resolution synthetic embryo images generated by AI. Hum. Reproduction 39, deae108.553, DOI: 10.1093/humrep/deae108.553 (2024). 16/17

    2024 doi

  34. [42]

    Liao, Q. et al. Development of deep learning algorithms for predicting blastocyst formation and quality by time-lapse monitoring. Commun. Biol. 4, 415, DOI: 10.1038/s42003-021-01937-1 (2021)

    2021 doi

  35. [43]

    & Jonaitis, D

    Raudonis, V ., Paulauskaite-Taraseviciene, A., Sutiene, K. & Jonaitis, D. Towards the automation of early-stage human embryo development detection. Biomed. Eng. Online 18, 120, DOI: 10.1186/s12938-019-0738-y (2019)

    2019 doi

  36. [44]

    & Boussommier-Calleja, A

    Canat, G., Duval, A., Gidel-Dissler, N. & Boussommier-Calleja, A. A novel deep learning approach to identify embryo morphokinetics in multiple time lapse systems. Sci. Reports 14, 29016, DOI: 10.1038/s41598-024-80565-1 (2024)

    2024 doi

  37. [45]

    & Fleet, D

    Azizi, S., Kornblith, S., Saharia, C., Norouzi, M. & Fleet, D. J. Synthetic Data from Diffusion Models Improves ImageNet Classification (2023). 2304.08466

    2023 arXiv

  38. [46]

    Zhou, Y ., Sahak, H. & Ba, J. Training on Thin Air: Improve Image Classification with Generated Data (2023). 2305.15316

    2023 arXiv

  39. [47]

    Al–Qerem, A. et al. Synthetic Generation of Multidimensional Data to Improve Classification Model Validity. J. Data Inf. Qual. 15, DOI: 10.1145/3603715 (2023)

    2023 doi

  40. [48]

    Revised guidelines for good practice in IVF laboratories (2015)†

    ESHRE Guideline Group on Good Practice in IVF Labs et al. Revised guidelines for good practice in IVF laboratories (2015)†. Hum. Reproduction 31, 685–686, DOI: 10.1093/humrep/dew016 (2016)

    2015 doi

  41. [49]

    & Köster, M

    Montag, M., Liebenthron, J. & Köster, M. Which morphological scoring system is relevant in human embryo development? Placenta 32, S252–S256, DOI: 10.1016/j.placenta.2011.07.009 (2011)

    2011 doi

  42. [50]

    & Park, H

    Lundin, K. & Park, H. Time-lapse technology for embryo culture and selection. Upsala J. Med. Sci. 125, 1–8, DOI: 10.1080/03009734.2020.1728444 (2020)

    2020

  43. [51]

    & Smith, G

    Sciorio, R. & Smith, G. Embryo culture at a reduced oxygen concentration of 5%: a mini review. Zygote 27, 355–361, DOI: 10.1017/s0967199419000522 (2019)

    2019 doi

  44. [52]

    Louis, C. M. et al. Review of computer vision application in in vitro fertilization: the application of deep learning-based computer vision technology in the world of IVF. J. Assist. Reproduction Genet. 38, 1627–1639, DOI: 10.1007/s10815-021 -02123-2 (2021)

    2021 doi

  45. [53]

    Karras, T. et al. Training Generative Adversarial Networks with Limited Data (2020). 2006.06676

    2020 arXiv

  46. [54]

    & Brox, T

    Ronneberger, O., Fischer, P. & Brox, T. U-Net: Convolutional Networks for Biomedical Image Segmentation. In International Conference on Medical image computing and computer-assisted intervention , 234–241 (Springer, 2015)

    2015

  47. [55]

    & Aila, T

    Karras, T., Laine, S. & Aila, T. Flickr-Faces-HQ Dataset (FFHQ). https://github.com/NVlabs/ffhq-dataset (2019). Author contributions statement O.P., A.S., V .T., A.D., and M.R. conceived the experiments and study design. O.P. conducted the experiments. A.S. conducted the backg...

    2019 arXiv

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.