Pith. sign in

REVIEW 3 major objections 6 minor 39 references

One multi-task network can segment cytoplasmic fragments and grade cleavage-stage embryo morphology from a single static image.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.5

2026-07-14 00:27 UTC pith:5VFHGPOW

load-bearing objection Solid applied multi-task embryo morphology paper with honest ablation and cautious framing; the main soft spot is unvalidated single-annotator labels, not the architecture or the feasibility claim as stated. the 3 major comments →

arxiv 2607.10093 v1 pith:5VFHGPOW submitted 2026-07-11 cs.CV

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

Anwar Hussain Sofi , Jung-Hua Wang , Ming-Jer Chen , Tsung-Hsien Lee , Yu-Chiao Yi , Ming-Kuan Lin , Yi-Chung Lai This is my paper
classification cs.CV
keywords Cleavage-stage embryoMulti-task learningFragmentation segmentationBlastomere symmetryCytoplasmic fragmentationEmbryo morphologyMedical image segmentationIn vitro fertilization
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Cleavage-stage IVF embryo review depends on three linked visual judgments: where cytoplasmic fragments sit, whether the embryo is at the two-cell or four-cell stage, and how evenly the blastomeres are sized. Human scoring of those features is noisy, especially when fragments are small, scattered, or low-contrast. This paper claims that a single shared-encoder model, EMBRACE, can output a pixel-level fragmentation mask together with t2/t4 stage and three-class symmetry grades from one static microscopy image, and that a compact multi-task design is enough to make that joint assessment feasible. On a held-out set of 914 annotated images the model reaches Dice 0.781 for fragments, near-perfect stage discrimination, and strong agreement on symmetry. The result matters because it turns fragmentation from an opaque image-level score into a spatially inspectable mask that can be checked against the photo, while still delivering the embryo-level morphology labels embryologists use together.

Core claim

EMBRACE shows that joint multi-task learning from static cleavage-stage images is feasible: a shared ResNet-50 encoder with concatenation-based multi-scale fusion, a U-Net-style decoder, and two classification heads can produce a cytoplasmic-fragmentation mask (Dice 0.781, IoU 0.677), t2/t4 stage labels (accuracy 0.995, macro-F1 0.994, AUC 1.000), and blastomere-symmetry grades (balanced accuracy 0.901, macro-F1 0.907, quadratic weighted kappa 0.859) on a 914-image held-out test set, with the compact baseline outperforming more complex auxiliary variants.

What carries the argument

EMBRACE: a shared ResNet-50 backbone plus Concatenation-Based Multi-Scale Feature Fusion (C-MSFF) that feeds a U-Net-style segmentation decoder and two task-specific heads, trained with a weighted compound loss that emphasizes sparse, boundary-sensitive fragmentation while balancing stage and symmetry classification.

Load-bearing premise

The pixel masks drawn by hand and the symmetry categories cut from published size-ratio thresholds are treated as reliable ground truth even though embryo morphology is known to vary between observers and no inter-annotator agreement is reported on this set.

What would settle it

An external multi-center test set with independent double annotation of the same images, reporting inter-observer Dice on fragmentation and kappa on symmetry, that either reproduces the reported Dice/IoU and class-wise metrics or shows a large drop once labels and imaging conditions leave the original two-center pool.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Fragmentation assessment can be reviewed as an overlay mask rather than only as an image-level score.
  • Stage and symmetry can be produced in the same forward pass as the mask, avoiding separate pipelines that may disagree.
  • A compact shared-encoder design can outperform more elaborate multi-task add-ons for this morphology suite.
  • External and prospective validation remain required before any clinical decision-support use.
  • Failure analysis can prioritize residual segmentation error on small, low-contrast, or boundary-ambiguous fragments.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the compact multi-task balance holds externally, labs could audit model disagreements by comparing predicted masks and symmetry labels side-by-side with the original photo rather than trusting a single viability score.
  • The same shared-encoder pattern could be stress-tested on other sparse embryo structures (e.g., multinucleation or zona irregularities) to see whether joint local-plus-global heads remain preferable to separate models.
  • Publishing inter-annotator Dice and kappa on the same LabelMe polygons would turn the weakest label assumption into a measurable ceiling for how high segmentation and symmetry metrics can reasonably go.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. EMBRACE is a multi-task deep learning framework that jointly predicts a pixel-wise cytoplasmic-fragmentation mask, a binary t2/t4 developmental-stage label, and a three-class blastomere-symmetry grade from a single static cleavage-stage embryo microscopy image. The architecture uses a shared ImageNet-pretrained ResNet-50 encoder, a concatenation-based multi-scale feature-fusion (C-MSFF) module, a U-Net-style segmentation decoder, and two task-specific classification heads, optimized with a fixed-weight multi-task loss that prioritizes a compound segmentation objective. After inclusion/exclusion filtering, 9,137 annotated images from two IVF centers were stratified 80/10/10 into train/validation/held-out test sets. On the 914-image test set the compact EXP1 configuration reports Dice 0.781 / IoU 0.677 for fragmentation, accuracy 0.995 / macro-F1 0.994 / AUC 1.000 for stage, and balanced accuracy 0.901 / macro-F1 0.907 / quadratic weighted kappa 0.859 for symmetry. A six-variant ablation (EXP1–EXP6) under a fixed split shows that added modules (Siamese consistency, uncertainty weighting, cross-task attention, CORAL ordinal regression, domain-adversarial learning, ASPP/deep supervision) do not improve the multi-task balance and often degrade segmentation. The authors frame the work as a research-oriented decision-support tool and explicitly call for external and prospective validation before clinical use.

Significance. If the reported held-out metrics hold under more rigorous label-quality and external validation, the paper would demonstrate a practical, spatially inspectable multi-output alternative to single-endpoint embryo AI models. Combining pixel-level fragmentation localization with stage and symmetry grading in one forward pass is clinically aligned with how embryologists actually score cleavage-stage embryos, and the large annotated set (9,137 images with pixel-level masks) plus a controlled ablation that favors a compact baseline over auxiliary complexity are genuine strengths. The work is therefore of clear interest to medical-image multi-task learning and reproductive-medicine AI, provided the ground-truth reliability and generalizability gaps are closed.

major comments (3)
  1. Sections 3.2 and 5 (and the abstract’s own caveat on observer variability): the central feasibility claim rests on single-annotator LabelMe fragmentation polygons and on degree-of-symmetry categories obtained by applying Yang et al. stage-level thresholds (Eq. 3) to instantaneous blastomere-area ratios on static frames. No inter-annotator agreement, consensus protocol, or multi-rater subset is reported for the 9,137-image set. Residual errors already concentrate on low-contrast fragments and the symmetric/discarded boundary (Fig. 2c–f, Fig. 3b, Table 7). Without a reliability estimate, the held-out Dice/accuracy figures measure consistency with one labeling process rather than clinically stable morphology; this is load-bearing for the multi-task feasibility claim and must be quantified (e.g., multi-rater subset with Dice/kappa) or the claim must be explicitly scoped to “agreement with th
  2. Section 3.1 / Table 2 and Discussion limitation 1: CSMUH images were added specifically to enrich minority asymmetric and discarded classes rather than as an independent external test set. The paper correctly notes this, yet the abstract and conclusions still present the 914-image held-out numbers as the primary evidence of feasibility. A true center-held-out or leave-one-center-out evaluation (or at least fully disaggregated per-center metrics for all three tasks, not only the partial EXP5 Dice numbers) is required before the multi-center framing can support generalizability claims.
  3. Section 3.6 and Table 4/5: no confidence intervals, bootstrap estimates, or repeated-run statistics are provided for any primary metric, despite acknowledged class imbalance and sparse foregrounds. For a multi-task medical-imaging claim whose strongest numbers approach saturation (stage Acc 0.995, AUC 1.000), uncertainty quantification is necessary to judge whether the EXP1 advantage over EXP2–EXP6 is stable and whether minority-class symmetry performance (asymmetric F1 0.881) is reliable.
minor comments (6)
  1. Fig. 1 caption and surrounding text: the figure is referenced as “Fig.1 1”; fix the duplicated numeral.
  2. Section 3.2 Eq. (2)–(3): the stage-level score is written with the same symbol as the instantaneous score (s_sym^CS(i)); the subsequent DoS thresholds mix s_sym and S_sym notation inconsistently. Clarify symbols.
  3. Section 3.3: images are first resampled to 512 imes512 then resized to 299 imes299 with aspect-ratio padding; state whether masks undergo the same geometric transforms before nearest-neighbor resize, and whether any center-specific resolution residual remains.
  4. Table 5: several cells are “NA” and EXP5 reports two Dice rows without a clear single-row summary; make the per-hospital versus pooled reporting uniform and complete IoU/F1 columns where possible.
  5. Section 3.5 / Table 1: the compound segmentation loss lists five terms but does not give relative weights among them; a short statement of how they were combined (equal, tuned, etc.) would aid reproducibility.
  6. References [9] DOI/link appears mismatched (text cites Paternot et al. 2009 but the URL points to a different identifier); verify bibliographic accuracy.

Circularity Check

0 steps flagged

No circularity: held-out Dice/accuracy/kappa are empirical measurements against fixed annotations, not forced by construction, fitted parameters, or self-referential definitions.

full rationale

EMBRACE is an empirical multi-task CNN paper. The central claims are measured performance numbers (Dice 0.781/IoU 0.677, stage Acc 0.995, symmetry BalAcc 0.901) on a stratified held-out test split of 914 images after training on 7309 images with fixed loss weights (α=1.0, β=γ=0.5) and a compound segmentation loss. These metrics do not reduce by construction to any free parameter or definition inside the paper: the model is optimized then scored against independently prepared LabelMe masks and stage/symmetry labels. Symmetry categories (Eq. 3) adapt thresholds from the external Yang et al. BlastAssist distributions (means ± SD), not from a self-citation or uniqueness theorem by the present authors. Ablation variants (EXP1–EXP6) are trained and evaluated under the identical split and protocol; selecting the compact EXP1 baseline on validation does not tautologically produce the test numbers. No equation equates a claimed prediction to a fitted input, no load-bearing uniqueness result is imported from overlapping authors, and no known empirical pattern is merely renamed. Label reliability and inter-observer variability are validity concerns outside the circularity criteria. The derivation chain is therefore self-contained experimental evaluation.

Axiom & Free-Parameter Ledger

4 free parameters · 4 axioms · 2 invented entities

The central feasibility claim rests on standard deep-learning components, clinical morphology assumptions drawn from the IVF literature, hand-chosen multi-task loss weights, and symmetry thresholds adapted from a prior paper. No new physical entities are postulated; the main invented construct is the EMBRACE architecture itself and the three-category DoS labeling rule.

free parameters (4)
  • multi-task loss weights α, β, γ = α=1.0, β=0.5, γ=0.5
    Empirically set to 1.0, 0.5, 0.5 to prioritize segmentation; not derived from first principles and affect joint optimization.
  • symmetry score thresholds for DoS categories = cs2: 0.12 / 0.22; cs4: 0.2 / 0.29
    Adapted from Yang et al. mean ± SD stage-level scores and applied to single static frames (Eq. 3); choice of cut-points defines the three-class labels the model is scored against.
  • fragmentation probability threshold = 0.5
    Fixed at 0.5 for binarizing the segmentation head output; standard but still a free operating-point choice.
  • dropout probability and classification projection dimension = p=0.3, dim=256
    p=0.3 dropout and 2048→256 projection are architectural hyperparameters chosen by the authors.
axioms (4)
  • domain assumption Visible t2 and t4 blastomere topologies plus pixel-level cytoplasmic fragments are sufficient and clinically meaningful targets for static-image multi-task assessment.
    Stated in Introduction and Section 3.1; excludes tPNf and full morphokinetics by design.
  • domain assumption Expert LabelMe polygon annotations and Yang-derived symmetry categories constitute reliable supervised ground truth despite known observer variability.
    Sections 3.2 and 5; no inter-annotator agreement reported.
  • ad hoc to paper Shared ResNet-50 encoder with C-MSFF and task-specific heads can learn both local boundary-sensitive and global morphology features without fatal gradient conflict when losses are fixed-weighted.
    Core architectural premise of EMBRACE; ablation tests variants but the premise itself is design choice.
  • standard math Stratified 80/10/10 split with seed 42 on the joint stage–symmetry label yields an unbiased held-out estimate of generalization within the curated two-center pool.
    Standard ML practice (Section 3.1); does not address external centers or devices.
invented entities (2)
  • EMBRACE multi-task architecture (C-MSFF + U-Net decoder + dual classification heads) no independent evidence
    purpose: Jointly produce fragmentation mask, t2/t4 label, and three-class symmetry grade from one static image.
    Named framework introduced by the paper; components are standard but the specific assembly and loss recipe are new to this work.
  • Three-category degree-of-symmetry (DoS) labeling rule for static frames no independent evidence
    purpose: Convert continuous blastomere-area CV into symmetric / discarded / asymmetric supervision.
    Extension of Yang et al. thresholds to single images (Eq. 3); defines the classification target.

pith-pipeline@v1.1.0-grok45 · 27166 in / 3537 out tokens · 24781 ms · 2026-07-14T00:27:49.178957+00:00 · methodology

0 comments
read the original abstract

Cleavage-stage embryo assessment in in vitro fertilization requires the integrated interpretation of cytoplasmic fragmentation, developmental stage, and blastomere symmetry. However, conventional visual assessment is affected by observer variability, particularly when fragmented regions are small, irregular, or low contrast. This study presents EMBRACE, a multi-task deep learning framework for jointly performing cytoplasmic-fragmentation segmentation, t2/t4 developmental-stage classification, and blastomere-symmetry grading from static cleavage-stage embryo microscopy images. EMBRACE combines a shared ResNet-50 backbone, a concatenation-based multi-scale feature-fusion (C-MSFF) module, a U-Net-style segmentation decoder, and two task-specific classification heads. After predefined inclusion and exclusion criteria, 9,137 annotated embryo images were divided into 7,309 training, 914 validation, and 914 held-out test images. On the held-out test set, EMBRACE achieved a Dice coefficient of 0.781 and an intersection over union of 0.677 for fragmentation segmentation. Developmental-stage classification achieved an accuracy of 0.995, macro-F1 of 0.994, and AUC of 1.000. Blastomere-symmetry grading achieved a balanced accuracy of 0.901, macro-F1 of 0.907, and quadratic weighted kappa of 0.859. These findings support the feasibility of combining spatially inspectable fragmentation localization with embryo-level morphology assessment in a single framework. External and prospective validation is required before clinical deployment.

Figures

Figures reproduced from arXiv: 2607.10093 by Anwar Hussain Sofi, Jung-Hua Wang, Ming-Jer Chen, Ming-Kuan Lin, Tsung-Hsien Lee, Yi-Chung Lai, Yu-Chiao Yi.

Figure 1
Figure 1. Figure 1: EMBRACE framework. The model receives a preprocessed cleavage [PITH_FULL_IMAGE:figures/full_fig_p009_1.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

39 extracted references · 16 canonical work pages

  1. [1]

    Selection b ased on morphological assessment of oocytes and embryos at different stages of preimplantation development: a review

    Ebner T, Moser M, Sommergruber M, Tews G. Selection b ased on morphological assessment of oocytes and embryos at different stages of preimplantation development: a review. Hum Reprod Update. 2003;9(3):251–262. https://doi.org/10.1093/humupd/dmg021

    work page doi:10.1093/humupd/dmg021 2003

  2. [2]

    The Istanbul consensus workshop on embryo assessment: proceedings of an expert meeting

    Alpha Scientists in Reproductive Medicine; ESHRE Special Interest Group of Embryology. The Istanbul consensus workshop on embryo assessment: proceedings of an expert meeting. Hum Reprod. 2011;26(6):1270–1283. https://doi.org/10.1093/humrep/der037

    work page doi:10.1093/humrep/der037 2011

  3. [3]

    Human embryo fragmentation in vitro and its implications for pregnancy and implantation

    Alikani M, Cohen J, Tomkin G, Garrisi GJ, Mack C, Scott RT. Human embryo fragmentation in vitro and its implications for pregnancy and implantation. Fertil Steril . 1999;71(5):836 –842. https://doi.org/10.1016/S0015-0282(99)00092-8

    work page doi:10.1016/s0015-0282(99)00092-8 1999

  4. [4]

    Human embryos with unevenly sized blastomeres have lower pregnancy and implantation rates: indications for aneuploidy and multinucleation

    Hardarson T, Hanson C, Sjögren A, Lundin K. Human embryos with unevenly sized blastomeres have lower pregnancy and implantation rates: indications for aneuploidy and multinucleation. Hum Reprod. 2001;16(2):313–318. https://doi.org/10.1093/humrep/16.2.313

    work page doi:10.1093/humrep/16.2.313 2001

  5. [5]

    The influence of Day 2 blastomere symmetry on blastocyst grade and ploidy status

    Tain DCX, Lim MSR, Ng BL, Hamm ond E, Wong PS. The influence of Day 2 blastomere symmetry on blastocyst grade and ploidy status. Fertil Reprod . 2019;1(2):115 –118. https://doi.org/10.1142/S2661318219500117

    work page doi:10.1142/s2661318219500117 2019

  6. [6]

    Timing of cell division in human cleavage-stage embryos is linked with blastocyst formation and quality

    Cruz M, G arrido N, Herrero J, Pérez -Cano I, Muñoz M, Meseguer M. Timing of cell division in human cleavage-stage embryos is linked with blastocyst formation and quality. Reprod Biomed Online. 2012;25(4):371–381. https://doi.org/10.1016/j.rbmo.2012.06.017

    work page doi:10.1016/j.rbmo.2012.06.017 2012

  7. [7]

    Cleavage anomalies in early human embryos and survival after prolonged culture in vitro

    Alikani M, Calderon G, Tomkin G, Garrisi J, Kokot M, Cohen J. Cleavage anomalies in early human embryos and survival after prolonged culture in vitro. Hum Reprod. 2000;15(12):2634–2643. https://doi.org/10.1093/humrep/15.12.2634

    work page doi:10.1093/humrep/15.12.2634 2000

  8. [8]

    Proposed guidelines on the nomenclature and annotation of dynamic human embryo monitoring by a time-lapse user group

    Ciray HN, Campbell A, Agerholm IE, Aguilar J, Chamayou S, Esbert M, et al. Proposed guidelines on the nomenclature and annotation of dynamic human embryo monitoring by a time-lapse user group. Hum Reprod. 2014;29(12):2650–2660. https://doi.org/10.1093/humrep/deu278

    work page doi:10.1093/humrep/deu278 2014

  9. [9]

    Intra- and interobserver analysis in the morphological assessment of early -stage embryos

    Paternot G, Devroe J, Debrock S, D’Hooghe TM, Spiessens C. Intra- and interobserver analysis in the morphological assessment of early -stage embryos. Reprod Biomed Online . 2009;19(6):836–844. https://doi.org/10.1186/1477-7827-7-105

    work page doi:10.1186/1477-7827-7-105 2009

  10. [10]

    Inter- and intra-observer variability of time-lapse annotations

    Sundvall L, Ingerslev HJ, Breth Knudsen U, Kirkegaard K. Inter- and intra-observer variability of time-lapse annotations. Hum Reprod. 2013;28(12):3215–3221. https://doi.org/10.1093/humrep/det366

    doi:10.1093/humrep/det366 2013

  11. [11]

    Ar tificial intelligence techniques for embryo and oocyte classification

    Manna C, Nanni L, Lumini A, Pappalardo S. Ar tificial intelligence techniques for embryo and oocyte classification. Reprod Biomed Online . 2013;26(1):42 –49. https://doi.org/10.1016/ j.rbmo.2012.09.015

    2013

  12. [12]

    Development of an artificial intelligence-based assessment model for prediction of embryo viability using static images captured by optical light microscopy during IVF

    VerMilyea M, Hall JMM, Diakiw SM, Johnston A, Nguyen T, Perugini D, et al. Development of an artificial intelligence-based assessment model for prediction of embryo viability using static images captured by optical light microscopy during IVF. Hum Reprod . 2020;35(4):770 –784. https://doi.org/10.1093/humrep/deaa013

    work page doi:10.1093/humrep/deaa013 2020

  13. [13]

    Development of deep learning algorithms for predicting blastocyst formation and quality by time -lapse monitoring

    Liao Q, Zhang Q, Feng X, et al. Development of deep learning algorithms for predicting blastocyst formation and quality by time -lapse monitoring. Commun Biol . 2021;4:415. https://doi.org/10.1038/s42003-021-01937-1

    work page doi:10.1038/s42003-021-01937-1 2021

  14. [14]

    BlastAssist : a deep learning pipeline to measure interpretable features of human embryos

    Yang HY, Leahy BD, Jang WD, Wei D, Kalma Y, Rahav R, et al. BlastAssist : a deep learning pipeline to measure interpretable features of human embryos. Hum Reprod . 2024;39(4):698 –708. https://doi.org/10.1093/humrep/deae024

    work page doi:10.1093/humrep/deae024 2024

  15. [15]

    U -Net: convolutional networks for biomedical image segmentation

    Ronneberger O, Fischer P, Brox T. U -Net: convolutional networks for biomedical image segmentation. Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015. LNCS

    2015

  16. [16]

    https://doi.org/10.1007/978-3-319-24574-4_28

    2015:234–241. https://doi.org/10.1007/978-3-319-24574-4_28

    doi:10.1007/978-3-319-24574-4_28 2015

  17. [17]

    V -Net: fully convolutional neural networks for volumetric medical image segmentation

    Milletari F, Navab N, Ahmadi SA. V -Net: fully convolutional neural networks for volumetric medical image segmentation. 2016 Fourth International Conference on 3D Vision . 2016:565 –571. https://doi.org/10.1109/3DV.2016.79

    doi:10.1109/3dv.2016.79 2016

  18. [18]

    The Lovász -Softmax loss: a tractable surrogate for the optimization of the intersection -over-union measure in neural networks

    Berman M, Triki AR, Blaschko MB. The Lovász -Softmax loss: a tractable surrogate for the optimization of the intersection -over-union measure in neural networks. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition . 2018:4413 –4421. https://doi.org/10.1109/CVPR.2018.00464

    doi:10.1109/cvpr.2018.00464 2018

  19. [19]

    Boundary loss for highly unbalanced segmentation

    Kervadec H, Bouchtiba J, Desrosiers C, Granger E, Dolz J, Ayed IB. Boundary loss for highly unbalanced segmentation. Med Image Anal . 2021;67:101851. https://doi.org/10.1016/j.media. 2020.101851

    doi:10.1016/j.media 2021

  20. [20]

    Multitask learning

    Caruana R. Multitask learning. Mach Learn . 1997;28:41 –75. https://doi.org/10.1023/ A:1007379606734

    1997

  21. [21]

    Which tasks should be learned together in multi -task learning? Proceedings of the 37th International Conference on Machine Learning

    Standley T, Zamir A, Chen D, Guibas L, Malik J, Savarese S. Which tasks should be learned together in multi -task learning? Proceedings of the 37th International Conference on Machine Learning. PMLR. 2020;119:9120–9132

    2020

  22. [22]

    Gradient surgery for multi-task learning

    Yu T, Kumar S, Gupta A, Levine S, Hausman K, Finn C. Gradient surgery for multi-task learning. Advances in Neural Information Processing Systems. 2020;33

    2020

  23. [23]

    Diabetic Macular Edema Detection Using End-to-End Deep Fusion Model and Anatomical Landmark Visualization on an Edge Computing Device

    Wang T, Chen Y, Chen J, Liu J, Wu P, Chang S, Lee Y, Su K and Chen C . Diabetic Macular Edema Detection Using End-to-End Deep Fusion Model and Anatomical Landmark Visualization on an Edge Computing Device. Front. Med. 2022; 9:851644. doi: 10.3389/fmed.2022.851644

    work page doi:10.3389/fmed.2022.851644 2022

  24. [24]

    Deep residual learning for image recognition

    He K, Zhang X, Ren S, Sun J. Deep residual learning for image recognition. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition . 2016:770 –778. https://doi.org/10.1109/CVPR.2016.90

    doi:10.1109/cvpr.2016.90 2016

  25. [25]

    Feature pyramid networks for object detection

    Lin TY, Dollár P, Girshick R, He K, Hariharan B, Belongie S. Feature pyramid networks for object detection. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition . 2017:936–944. https://doi.org/10.1109/CVPR.2017.106

    doi:10.1109/cvpr.2017.106 2017

  26. [26]

    Learning from imbalanced data

    He H, Garcia EA. Learning from imbalanced data. IEEE Trans Knowl Data Eng . 2009;21(9):1263–1284. https://doi.org/10.1109/TKDE.2008.239

    doi:10.1109/tkde.2008.239 2009

  27. [27]

    Focal loss for dense object detection

    Lin TY, Goyal P, Girshick R, He K, Dollár P. Focal loss for dense object detection. Proceedings of the IEEE International Conference on Computer Vision . 2017:2980 –2988. https://doi.org/10.1109/ICCV.2017.324

    doi:10.1109/iccv.2017.324 2017

  28. [28]

    Scikit -learn: machine learning in Python

    Pedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O, et al. Scikit -learn: machine learning in Python. J Mach Learn Res. 2011;12:2825–2830

    2011

  29. [29]

    The use of morphokinetics as a predictor of embryo implantation

    Meseguer M, Herrero J, Tejera A, Hilligsøe KM, Ramsing NB, Remohí J. The use of morphokinetics as a predictor of embryo implantation. Hum Reprod . 2011;26(10):2658 –2671. https://doi.org/10.1093/humrep/der256

    work page doi:10.1093/humrep/der256 2011

  30. [30]

    Time -lapse parameters as predictors of blastocyst development and pregnancy outcome in embryos from good prognosis patients: a prospective cohort study

    Kirkegaard K, Kesmodel U, Hindkjæ r J, Ingerslev H. Time -lapse parameters as predictors of blastocyst development and pregnancy outcome in embryos from good prognosis patients: a prospective cohort study. Hum Reprod . 2013;28(10):2643 –2651. https://doi.org/ 10.1093/humrep/det300

    work page doi:10.1093/humrep/det300 2013

  31. [31]

    Signature verification using a Siamese time delay neural network

    Bromley J, Guyon I, LeCun Y, Säckinger E, Shah R. Signature verification using a Siamese time delay neural network. Advances in Neural Information Processing Systems. 1993;6

    1993

  32. [32]

    Cross -stitch networks for multi -task learning

    Misra I, Shrivastava A, Gupta A, Hebert M. Cross -stitch networks for multi -task learning. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2016:3994–4003. https://doi.org/10.1109/CVPR.2016.433

    doi:10.1109/cvpr.2016.433 2016

  33. [33]

    CBAM: Convolutional Block Attention Module

    Woo S, Park J, L ee JY, Kweon IS. CBAM: Convolutional Block Attention Module. European Conference on Computer Vision. 2018:3–19. https://doi.org/10.1007/978-3-030-01234-2_1

    doi:10.1007/978-3-030-01234-2_1 2018

  34. [34]

    Rank consistent ordinal regression for neural networks with application to age estimation

    Cao W, Mirjalili V, Raschka S. Rank consistent ordinal regression for neural networks with application to age estimation. Pattern Recognit Lett . 2020;140:325 –331. https://doi.org/10.1016/ j.patrec.2020.11.008

    2020

  35. [35]

    Domain-adversarial training of neural networks

    Ganin Y, Ustinova E, Ajakan H, Germain P, Larochelle H, Laviolette F, Marchand M, Lempitsky V. Domain-adversarial training of neural networks. J Mach Learn Res. 2016;17(59):1–35

    2016

  36. [36]

    Rethinking atrous convoluti on for semantic image segmentation

    Chen LC, Papandreou G, Schroff F, Adam H. Rethinking atrous convoluti on for semantic image segmentation. arXiv:1706.05587. 2017. https://doi.org/10.48550/arXiv.1706.05587

    Pith/arXiv arXiv doi:10.48550/arxiv.1706.05587 2017

  37. [37]

    Deeply -supervised nets

    Lee CY, Xie S, Gallagher P, Zhang Z, Tu Z. Deeply -supervised nets. Proceedings of the 18th International Conference on Artificial Intelligence and Statistics. PMLR. 2015;38:562–570

    2015

  38. [38]

    Weakly supervised deep learning in radiology

    Misera L, Müller-Franzes G, Truhn D, Kather JN. Weakly supervised deep learning in radiology. Radiology. 2024;312(1):e232085. https://doi.org/10.1148/radiol.232085

    work page doi:10.1148/radiol.232085 2024

  39. [39]

    Medical image analysis method and device

    Wang TY, Deng M, Lee Y, LIU J. Medical image analysis method and device. US Patent 11610306, 2023

    2023