Pith. sign in

REVIEW 3 major objections 5 minor 37 references

Are Traditional Deep Learning Model Approaches as Effective as a Retinal-Specific Foundation Model for Ocular and Systemic Disease Detection?

T0 review · 3 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read This paper claims traditional ImageNet-pretrained models match a retina-specific foundation model for ocular disease detection when training data are large, and that the foundation model is superior mainly for systemic disease detection…

desk verdict Systematic head-to-head benchmark of RETFound vs ImageNet-pretrained models with strong external validation; the main caveat is an underspecified internal test split, which is fixable and probably not fatal. read the letter →

arxiv 2501.12016 v1 pith:N2QHI7YB submitted 2025-01-21 cs.CV cs.LG

Samantha Min Er Yew , Xiaofeng Lei , Jocelyn Hui Lin Goh , Yibing Chen , Sahana Srinivasan , Miao-li Chee , Krithi Pushpanathan , Ke Zou
show 21 more authors
Qingshan Hou Zhi Da Soh Cancan Xue Marco Chak Yan Yu Charumathi Sabanayagam E Shyong Tai Xueling Sim Yaxing Wang Jost B. Jonas Vinay Nangia Gabriel Dawei Yang Emma Anran Ran Carol Yim-Lui Cheung Yangqin Feng Jun Zhou Rick Siow Mong Goh Yukun Zhou Pearse A. Keane Yong Liu Ching-Yu Cheng Yih-Chung Tham
This is my paper · ORCID
classification cs.CVcs.LG
keywords RETFoundretinafoundationmodeltransferlearningoculomicssystemicdiseasedetectionfundusphotographydiabeticretinopathycomparison
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 sets out to settle whether a retina-specific foundation model, RETFound, is worth adopting over three standard ImageNet-pretrained deep learning models (ResNet50, ViT-base, SwinV2) for detecting four ocular and three systemic diseases from colour fundus photographs. It fine-tunes all four models on the full training set, on 50% and 20% of it, and on fixed small samples (400, 200, and 100 images; 500 and 250 for diabetic retinopathy), then tests internally and on multiple external datasets. Its central finding is that for ocular diseases with large datasets the standard models match RETFound, while RETFound's advantage appears mainly in systemic disease detection with smaller datasets. The practical stake is whether expensive foundation-model fine-tuning is justified, or whether conventional transfer learning remains a competitive, resource-efficient option.

What carries the argument

RETFound is a masked-autoencoder foundation model pretrained on roughly 900,000 unlabeled colour fundus photographs and 700,000 OCT scans, so its encoder has already learned retina-specific visual structure before fine-tuning. The three traditional models use supervised ImageNet-pretrained weights: ResNet50, ViT-base, and SwinV2. The carrying mechanism of the comparison is a structured fine-tuning grid that varies training data both proportionally (100%, 50%, 20%) and in absolute counts (400, 200, 100 images for binary tasks; 500 and 250 for five-class diabetic retinopathy), followed by internal testing and external validation on population-based and open-source datasets, with performance compared by AUC and Z-tests with Bonferroni correction. This design allows the authors to separate task type and data availability as the conditions under which a retina-specific foundation model helps.

What would settle it

Reproduce the small-sample internal experiments with an explicit, documented partition of SEED and APTOS into disjoint training, validation, and test sets; if RETFound's superiority over traditional models in systemic disease detection disappears or reverses under such a split, the central claim is falsified.

Watch

Extended reading notes

Core claim

For ocular diseases, fine-tuning on full datasets gives statistically comparable internal performance between traditional models and RETFound (AUCs of 0.914–0.965 versus 0.938–0.966). With smaller datasets the comparable trend mostly persists, except for diabetic retinopathy at 100 images per class and glaucoma at 400 images or fewer, where ResNet50 is inferior to RETFound; SwinV2 remains comparable in those low-data settings. For systemic diseases, RETFound does not dominate on full datasets, but with smaller sample sizes it consistently outperforms the traditional models: with 100 images, RETFound exceeds the others for diabetes (AUC 0.648 versus 0.612 for ResNet50 and 0.580 for SwinV2), hypertension (0.705 versus 0.634 and 0.648), and chronic kidney disease (0.822 versus 0.778 for SwinV2 and 0.743 for ViT-base). The paper concludes that the value of a retina-specific foundation model is concentrated in low-data oculomics tasks rather than in well-resourced ocular disease detection.

Load-bearing premise

The internal test results assume the SEED and APTOS images used for fine-tuning are disjoint from the images used for testing, but the paper does not describe an explicit train/test split, so any overlap would invalidate the reported internal comparisons.

Editorial extensions

If this is right

  • If the claim is right, clinics with large labelled ocular datasets gain little diagnostic accuracy from switching to RETFound over cheaper ImageNet-pretrained models.
  • For systemic disease screening from retinal photos with limited labels, RETFound should be the preferred starting point, because its advantage grows as sample size shrinks.
  • SwinV2 emerges as a strong resource-efficient alternative to RETFound for diabetic retinopathy and glaucoma in low-data settings, since it remained statistically comparable there.
  • Computational cost should factor into model choice: the paper reports RETFound peaked at 12.3 GB GPU memory during fine-tuning versus 2.3 GB for ResNet50, and processed images more slowly.
  • Cross-ethnic generalization remains a shared weakness: all models, including RETFound, showed larger AUC drops on external datasets whose ethnic composition differed from the fine-tuning data.

Reading between the lines

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

  • Beyond the paper, the pattern suggests that the value of a domain-specific foundation model scales with how far the target signal departs from natural-image features: gross structural lesions transfer well, whereas subtle vascular changes tied to systemic disease require retina-specific representations.
  • A testable extension would rerun the same fine-tuning grid against newer retina-specific foundation models; the paper's logic predicts the small-sample advantage will narrow as pretraining data become larger and more diverse.
  • The protocol also supplies a practical rule of thumb for deployment: when labelled ocular data exceed a few hundred images per class, the cheaper model may suffice, and compute budgets should then decide the choice.
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 / 5 minor

Summary. The paper reports a large empirical comparison between RETFound, a retina-specific self-supervised foundation model, and three ImageNet-pretrained supervised models (ResNet50, ViT-base, SwinV2) for four ocular disease tasks and three systemic disease tasks. Models were fine-tuned/trained on full, proportional, and small fixed-size training sets, then evaluated on internal test images from SEED and APTOS-2019, with external validation on nine additional datasets (BES, CIEMS, SP2, UKBB, ODIR-5k, PAPILA, GAMMA, IDRiD, MESSIDOR-2). The central claim is that traditional models are mostly comparable to RETFound for ocular disease detection when datasets are large, while RETFound is superior for systemic disease detection when fine-tuning datasets are small (e.g., 100 to 400 images). The paper also reports RETFound's higher GPU memory use and slower inference relative to traditional models.

Significance. If the findings hold, this would be a practically useful benchmark for model selection in retinal image analysis, showing when the added cost of a retina-specific foundation model is justified. The main strength is the unusually broad external validation across multiple independent population-based and open-source datasets, which supports the qualitative trend of the conclusions. The manuscript also provides a transparent comparison of label efficiency and computational resource demands, which are rarely reported together in this literature. The key weakness is that the internal train/validation/test partition is not described, so the reported internal AUC numbers, which appear in the abstract and are used to support several specific claims, cannot currently be fully trusted; this is fixable by clarifying or redoing the split.

major comments (3)
  1. [Methods, 'Model fine-tuning/training'] The manuscript never specifies how the training, validation, and internal test sets were partitioned for SEED and APTOS-2019. The text states that all models were fine-tuned/trained on, and internally tested on, these datasets, and that weights were selected by validation AUC, but it does not state that the internal test set is disjoint from the fine-tuning images or from the validation set used for checkpoint selection. If the same images appear in both training and testing, the internal AUCs in Figures 2–4 and Supplementary Tables 5–11, and the abstract's headline small-sample comparisons (e.g., diabetes 0.648 vs 0.612 at 100 images), would be optimistically biased. Please report the explicit split procedure (e.g., patient-level random split, percentages, and whether the validation set is a subset of the training set) and, if any overlap exists, re-run the internal evaluation on a fully held-out set. The external validations are genuinely independent and support the qualitative trend, so this issue should be fixable without changing the study design.
  2. [Methods, 'Model pre-training'] RETFound's pretraining corpus includes approximately 900,000 unlabeled colour fundus photographs from Moorfields and Kaggle EyePACs, while the diabetic retinopathy task uses APTOS-2019, which is also a Kaggle diabetic-retinopathy dataset. The paper does not address the possibility of image or patient overlap, or even close distributional similarity, between the EyePACs pretraining data and the APTOS fine-tuning/test data. This matters because it could explain part of RETFound's advantage in the DR task as a dataset-specific prior rather than as a general retina-specific foundation-model benefit. Please verify and report whether any overlap exists, and discuss the distributional relationship between EyePACs and APTOS (e.g., image sources, acquisition devices, grading scales). This concern does not directly affect the systemic disease tasks, which are fine-tuned on SEED.
  3. [Methods, 'Evaluation Metrics and Statistical Analysis'] The Bonferroni correction is applied to the three pairwise model comparisons within each sample-size and dataset cell, but the paper then makes global claims such as 'consistently outperformed' and 'superior' across many cells aggregated over tasks, sample sizes, and test sets (see Results sections on DR, glaucoma, and systemic diseases). With dozens or hundreds of cells tested, the family-wise error rate for these aggregate statements is substantially higher than the stated 0.05/3. Please either restrict the global claims to the per-cell significant differences or perform an additional sensitivity analysis that applies a broader correction across the full set of comparisons, and report how many significant differences survive.
minor comments (5)
  1. [Abstract] In the Methods sentence, the phrase 'for each DR severity class, 100 and 50 cases were used' is a sentence fragment; it should be integrated into the preceding parentheses or rephrased as a complete sentence.
  2. [Abstract and Conclusion] The phrase 'the respective merits and limitation of traditional models' should read 'the respective merits and limitations of traditional models' (plural).
  3. [Figure 2 and Figure 3] Consider adding confidence intervals or error bars to the main figures, as the AUC difference between models is often small and the uncertainty is currently only available in the supplementary tables.
  4. [Results, DR detection] In the sentence reporting DR AUC ranges (e.g., 'RETFound consistently outperformed ResNet50 (AUCs ranging from 0.625–0.811)'), clarify that these ranges span the different sample sizes and test sets, so readers do not misinterpret them as a single confidence interval.
  5. [Methods, terminology] The paper explicitly distinguishes 'fine-tuning' for RETFound from 'training' for traditional models, but then uses 'fine-tuning' for all models throughout the results and discussion; please add a note where the terminology shifts to avoid ambiguity.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the paper is a direct empirical benchmark comparing fine-tuned model AUCs on independent test sets.

full rationale

This is an empirical benchmarking study, not a derivation. RETFound, ResNet50, ViT-base, and SwinV2 are fine-tuned on training subsets and then evaluated on internal and external test sets; the central claim (RETFound is comparable for ocular disease with large datasets but superior for systemic disease with small datasets) is a summary of measured AUC differences. No equation is inverted, no parameter is fitted to the reported test AUC, and no uniqueness theorem is invoked. The only author-overlapping citations are the original RETFound paper (ref 4, Zhou et al., Nature 2023) and related ophthalmology-FM reviews; these are used to introduce the model and its pretraining, not as evidence for the paper's comparative conclusions. RETFound's pretraining on ~900,000 unlabeled retinal images is the treatment being tested, not a hidden input: it is a data-prior advantage, and the comparison against ImageNet-pretrained models is exactly the empirical question. The internal test split for SEED and APTOS-2019 is not described, which raises a potential data-leakage/validity concern, but that is a correctness risk, not circularity. External validation on BES, CIEMS, SP2, UKBB, ODIR-5k, PAPILA, GAMMA, IDRiD, and MESSIDOR-2 is independent of fine-tuning and would remain informative even if the internal split were flawed.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

The central comparison rests on the premise that training and internal test sets are disjoint, and that the different pretraining/fine-tuning protocols for RETFound and the traditional models do not bias the comparison. These are modelling and data-split assumptions, not fitted parameters.

assumptions (3)
  • domain assumption Internal test sets are disjoint from training sets
    The paper states models were fine-tuned and internally tested on SEED and APTOS without describing an explicit split. If overlapping, internal AUCs are inflated. Location: Methods, Datasets.
  • domain assumption RETFound's pretraining data do not overlap with the DR fine-tuning/test datasets
    RETFound was pretrained on EyePACs, which is a diabetic retinopathy dataset; APTOS-2019 is used for DR fine-tuning. Any image overlap or strong distributional similarity could inflate RETFound's DR AUC. Location: Methods, Model pre-training.
  • domain assumption The chosen hyperparameters for each model are comparably tuned
    RETFound and traditional models use different learning rates, epochs, and weight decay. If RETFound's settings are better tuned, the comparison may be unfair. Location: Methods, Model fine-tuning/training.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Are Traditional Deep Learning Model Approaches as Effective as a Retinal-Specific Foundation Model for Ocular and Systemic Disease Detection?." pith.science (2026). https://pith.science/paper/N2QHI7YB

@misc{pith2026250112016,
  author       = {Pith},
  title        = {Pith review of: Are Traditional Deep Learning Model Approaches as Effective as a Retinal-Specific Foundation Model for Ocular and Systemic Disease Detection?},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/N2QHI7YB}},
  note         = {Machine review of arXiv:2501.12016}
}
read the original abstract

Background: RETFound, a self-supervised, retina-specific foundation model (FM), showed potential in downstream applications. However, its comparative performance with traditional deep learning (DL) models remains incompletely understood. This study aimed to evaluate RETFound against three ImageNet-pretrained supervised DL models (ResNet50, ViT-base, SwinV2) in detecting ocular and systemic diseases. Methods: We fine-tuned/trained RETFound and three DL models on full datasets, 50%, 20%, and fixed sample sizes (400, 200, 100 images, with half comprising disease cases; for each DR severity class, 100 and 50 cases were used. Fine-tuned models were tested internally using the SEED (53,090 images) and APTOS-2019 (3,672 images) datasets and externally validated on population-based (BES, CIEMS, SP2, UKBB) and open-source datasets (ODIR-5k, PAPILA, GAMMA, IDRiD, MESSIDOR-2). Model performance was compared using area under the receiver operating characteristic curve (AUC) and Z-tests with Bonferroni correction (P<0.05/3). Interpretation: Traditional DL models are mostly comparable to RETFound for ocular disease detection with large datasets. However, RETFound is superior in systemic disease detection with smaller datasets. These findings offer valuable insights into the respective merits and limitation of traditional models and FMs.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 37 canonical work pages

  1. [1]

    RETFound

    1 Are Traditional Deep Learning Model Approaches as Effective as a Retinal-Specific Foundation Model for Ocular and Systemic Disease Detection? Samantha Min Er Yew1,2*, Xiaofeng Lei3*, Jocelyn Hui Lin Goh1,2,4*, Yibing Chen4, Sahana Srinivasan1,2,4, Miao-li Chee4, Krithi Pushpanathan1,2, Ke Zou1,2, Qingshan Hou1,2, Zhi Da Soh4, Cancan Xue4, Marco Chak Yan...

    work page 2019

  2. [2]

    For detection of disease-related VI, the models were fine-tuned (for RETFound)/trained (for traditional models), and internally tested on the CFPs and clinical data from the Singapore Epidemiology of Eye Disease (SEED) study, which comprises of the Singapore Malay Eye study (SIMES), the Singapore Chinese Eye study (SCES) and the Singapore Indian Eye study...

    work page 2019

  3. [3]

    A visual-language foundation model for computational pathology

    Lu MY, Chen B, Williamson DFK, et al. A visual-language foundation model for computational pathology. Nature Medicine 2024; 30(3): 863-74

    work page 2024

  4. [4]

    For the five-class DR detection, we first calculated the class-specific AUC and maximum F1 score, followed by macro-average AUC and macro-average maximum F1 score

    Evaluation Metrics and Statistical Analysis For the detection of ocular and systemic disease tasks (excluding DR detection), we evaluated the models’ performance using area under the receiver operating characteristic curve (AUC) and the maximum F1 score. For the five-class DR detection, we first calculated the class-specific AUC and maximum F1 score, foll...

    work page 2000

  5. [5]

    The CIEMS consisted of Indian participants aged 30-100 years

    In brief, the SEED study consisted of participants aged 40.0-91.3 years of Malay (SIMES), Indian (SINDI), and Chinese (SCES) ethnicities. The CIEMS consisted of Indian participants aged 30-100 years. The BES and CUHK-STDR consisted of Chinese participants aged 50-93 years and 15.2-86.7 years, respectively. The SP2 study consisted of individuals aged 24.6-...

    work page 2019

  6. [6]

    *=applies to ResNet50 model only

    Overview of study design. *=applies to ResNet50 model only. †=applies to ViT-base and SwinV2 models only. MEH-MIDAS= Moorfields Diabetic imAge dataset. SEED= Singapore Epidemiology of Eye Diseases. CIEMS= Central India Eye and Medical study. BES=Beijing Eye study. CUHK-STDR= Chinese University of Hong Kong’s Sight Threatening Diabetic Retinopathy study. O...

    work page 2019

  7. [7]

    Foundation models in ophthalmology

    Chia MA, Antaki F, Zhou Y, Turner AW, Lee AY, Keane PA. Foundation models in ophthalmology. Br J Ophthalmol 2024; 108(10): 1341-8

    work page 2024

  8. [8]

    A foundation model for generalizable disease detection from retinal images

    Zhou Y, Chia MA, Wagner SK, et al. A foundation model for generalizable disease detection from retinal images. Nature 2023; 622(7981): 156-63

    work page 2023

Show all 37 references
  1. [9]

    Development and Validation of a Multimodal Multitask Vision Foundation Model for Generalist Ophthalmic Artificial Intelligence

    Qiu J, Wu J, Wei H, et al. Development and Validation of a Multimodal Multitask Vision Foundation Model for Generalist Ophthalmic Artificial Intelligence. NEJM AI 2024; 1(12): AIoa2300221

    2024

  2. [10]

    On the Opportunities and Risks of Foundation Models2021

    Bommasani R, Hudson DA, Adeli E, et al. On the Opportunities and Risks of Foundation Models2021. https://ui.adsabs.harvard.edu/abs/2021arXiv210807258B (accessed August 01, 2021)

    2021

  3. [11]

    Swin transformer v2: Scaling up capacity and resolution

    Liu Z, Hu H, Lin Y, et al. Swin transformer v2: Scaling up capacity and resolution. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition; 2022

    2022

  4. [12]

    An empirical study of training self-supervised vision transformers

    Chen X, Xie S, He K. An empirical study of training self-supervised vision transformers. Proceedings of the IEEE/CVF international conference on computer vision; 2021

    2021

  5. [13]

    Masked autoencoders are scalable vision learners

    He K, Chen X, Xie S, Li Y, Dollár P, Girshick R. Masked autoencoders are scalable vision learners. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition; 2022

    2022

  6. [14]

    Deep Residual Learning for Image Recognition

    He K, Zhang X, Ren S, Sun J. Deep Residual Learning for Image Recognition. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR); 2016 27-30 June 2016

    2016

  7. [15]

    When do we not need larger vision models? European Conference on Computer Vision; 2025: Springer

    Shi B, Wu Z, Mao M, Wang X, Darrell T. When do we not need larger vision models? European Conference on Computer Vision; 2025: Springer

    2025

  8. [16]

    Comparative Analysis of Vision Transformers and Conventional Convolutional Neural Networks in Detecting Referable Diabetic Retinopathy

    Goh JHL, Ang E, Srinivasan S, et al. Comparative Analysis of Vision Transformers and Conventional Convolutional Neural Networks in Detecting Referable Diabetic Retinopathy. Ophthalmol Sci 2024; 4(6): 100552

    2024

  9. [17]

    A survey on deep learning in medical image analysis

    Litjens G, Kooi T, Bejnordi BE, et al. A survey on deep learning in medical image analysis. Medical Image Analysis 2017; 42: 60-88

    2017

  10. [18]

    These comparisons were conducted across various downstream ocular and systemic disease detection tasks, using varying fine-tuning sample sizes and multiple external test sets

    17 Discussion Our study evaluated and compared the performance, label efficiency, and computational resource demands of RETFound, a retina-specific FM, with commonly used traditional DL model approaches, including ResNet50, ViT-base, and SwinV2. These comparisons were conducte...

    2019

  11. [19]

    Battle of the backbones: A large-scale comparison of pretrained models across computer vision tasks

    Goldblum M, Souri H, Ni R, et al. Battle of the backbones: A large-scale comparison of pretrained models across computer vision tasks. Advances in Neural Information Processing Systems 2024

    2024

  12. [20]

    Cohort Profile: The Singapore Epidemiology of Eye Diseases study (SEED)

    Majithia S, Tham YC, Chee ML, et al. Cohort Profile: The Singapore Epidemiology of Eye Diseases study (SEED). Int J Epidemiol 2021; 50(1): 41-52

    2021

  13. [21]

    Refractive error in central India: the Central India Eye and Medical Study

    Nangia V, Jonas JB, Sinha A, Matin A, Kulkarni M. Refractive error in central India: the Central India Eye and Medical Study. Ophthalmology 2010; 117(4): 693-9

    2010

  14. [22]

    The Beijing Eye Study

    Jonas JB, Xu L, Wang YX. The Beijing Eye Study. Acta Ophthalmol 2009; 87(3): 247-61

    2009

  15. [23]

    OCT Angiography Metrics Predict Progression of Diabetic Retinopathy and Development of Diabetic Macular Edema: A Prospective Study

    Sun Z, Tang F, Wong R, et al. OCT Angiography Metrics Predict Progression of Diabetic Retinopathy and Development of Diabetic Macular Edema: A Prospective Study. Ophthalmology 2019; 126(12): 1675-84

    2019

  16. [24]

    APTOS 2019 Blindness Detection

    Karthik M SD. APTOS 2019 Blindness Detection

    2019

  17. [25]

    FEEDBACK ON A PUBLICLY DISTRIBUTED IMAGE DATABASE: THE MESSIDOR DATABASE

    Decencière E, Zhang X, Cazuguel G, et al. FEEDBACK ON A PUBLICLY DISTRIBUTED IMAGE DATABASE: THE MESSIDOR DATABASE. Image Analysis &amp; Stereology 2014; 33(3):

    2014

  18. [26]

    PAPILA: Dataset with fundus images and clinical data of both eyes of the same patient for glaucoma assessment

    Kovalyk O, Morales-Sánchez J, Verdú-Monedero R, Sellés-Navarro I, Palazón-Cabanes A, Sancho-Gómez J-L. PAPILA: Dataset with fundus images and clinical data of both eyes of the same patient for glaucoma assessment. Scientific Data 2022; 9(1)

    2022

  19. [27]

    GAMMA challenge: Glaucoma grAding from Multi-Modality imAges

    Wu J, Fang H, Li F, et al. GAMMA challenge: Glaucoma grAding from Multi-Modality imAges. Medical Image Analysis 2023; 90: 102938

    2023

  20. [28]

    Cohort Profile: The Singapore Multi-Ethnic Cohort (MEC) study

    Tan KHX, Tan LWL, Sim X, et al. Cohort Profile: The Singapore Multi-Ethnic Cohort (MEC) study. International Journal of Epidemiology 2018; 47(3): 699-j

    2018

  21. [29]

    Nang EEK, Khoo CM, Tai ES, et al. Is There a Clear Threshold for Fasting Plasma Glucose That Differentiates Between Those With and Without Neuropathy and Chronic Kidney Disease?: The Singapore Prospective Study Program. American Journal of Epidemiology 2009; 169(12): 1454-62

    2009

  22. [30]

    Cohort profile: design and methods in the eye and vision consortium of UK Biobank

    Chua SYL, Thomas D, Allen N, et al. Cohort profile: design and methods in the eye and vision consortium of UK Biobank. BMJ Open 2019; 9(2): e025077

    2019

  23. [31]

    A method of comparing the areas under receiver operating characteristic curves derived from the same cases

    Hanley JA, McNeil BJ. A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology 1983; 148(3): 839-43

    1983

  24. [32]

    Insights into Systemic Disease through Retinal Imaging-Based Oculomics

    Wagner SK, Fu DJ, Faes L, et al. Insights into Systemic Disease through Retinal Imaging-Based Oculomics. Transl Vis Sci Technol 2020; 9(2):

    2020

  25. [33]

    A deep-learning system for the assessment of cardiovascular disease risk via the measurement of retinal-vessel calibre

    Cheung CY, Xu D, Cheng CY, et al. A deep-learning system for the assessment of cardiovascular disease risk via the measurement of retinal-vessel calibre. Nat Biomed Eng 2021; 5(6): 498-508

    2021

  26. [34]

    Deep-learning-based cardiovascular risk stratification using coronary artery calcium scores predicted from retinal photographs

    Rim TH, Lee CJ, Tham YC, et al. Deep-learning-based cardiovascular risk stratification using coronary artery calcium scores predicted from retinal photographs. Lancet Digit Health 2021; 3(5): e306-e16

    2021

  27. [35]

    Evaluating a Foundation Artificial Intelligence Model for Glaucoma Detection Using Color Fundus Photographs

    Chuter B, Huynh J, Hallaj S, et al. Evaluating a Foundation Artificial Intelligence Model for Glaucoma Detection Using Color Fundus Photographs. Ophthalmology Science 2025; 5(1)

    2025

  28. [36]

    RETFound-enhanced community-based fundus disease screening: real-world evidence and decision curve analysis

    Zhang J, Lin S, Cheng T, et al. RETFound-enhanced community-based fundus disease screening: real-world evidence and decision curve analysis. npj Digital Medicine 2024; 7(1):

    2024

  29. [39]

    A New Foundation Model for Multimodal Ophthalmic Images: Advancing Disease Detection and Prediction

    Chia MA, Zhou Y, Keane PA. A New Foundation Model for Multimodal Ophthalmic Images: Advancing Disease Detection and Prediction. NEJM AI 2024; 1(12)

    2024

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.