DMS-Net:Dual-Modal Multi-Scale Siamese Network for Binocular Fundus Image Classification

Guohao Huo; Hao Tang; Ruiting Dai; Zibo Lin; Zitong Wang

REVIEW 4 major objections 5 minor 1 cited by

DMS-Net:Dual-Modal Multi-Scale Siamese Network for Binocular Fundus Image Classification

T0 review · 4 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read A weight-sharing Siamese network that fuses left- and right-eye features tops the ODIR-5K binocular fundus benchmark.

desk verdict A plausible Siamese architecture with internally consistent ablations, but the SOTA claim is unverifiable because the evaluation protocol and external comparisons are missing. read the letter →

arxiv 2504.18046 v3 pith:ZYFVYT4S submitted 2025-04-25 cs.CV cs.AI

Guohao Huo , Zibo Lin , Zitong Wang , Ruiting Dai , Hao Tang This is my paper · ORCID

classification cs.CVcs.AI

keywords binocularfundusimageclassificationSiamesenetworkcross-modalattentionmulti-scalepoolingODIR-5Kophthalmicdiagnosticrobotdiseasedeeplearning

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

This paper tries to establish that binocular fundus image classification is better done by reading the two eyes together than by classifying each eye separately. It proposes DMS-Net, a weight-sharing Siamese ResNet-152 that extracts features from the left and right fundus images in parallel, then applies four attention-and-alignment modules to combine them. On the ODIR-5K benchmark the model reaches an accuracy of 82.9%, recall of 84.5%, and a kappa coefficient of 83.2%, which the authors report as state-of-the-art. The clinical motivation is that pathologies often appear asymmetrically or symmetrically across the two eyes, so explicit modeling of both contrast and correlation could support ophthalmic diagnostic robots.

What carries the argument

The load-bearing mechanism is a set of paired-image feature interactions stacked on a Siamese ResNet-152. OSIM, a multi-scale adaptive pooling plus spatial attention module, aggregates lesion features at several resolutions; CASFM, a bidirectional cross-attention module with learned residual weights, builds a modality-agnostic shared representation of both fundus structures; CCAM, a cross-attention-driven contrastive alignment module, highlights asymmetric lesion differences between eyes; and CIAM, a cross-attention-driven integrative alignment module, reinforces symmetric interocular patterns. Together these modules turn two independent feature maps into one fused representation that the classifier uses to predict a single disease label per patient.

What would settle it

Take the released code and a fixed public split of ODIR-5K, rerun the full model and ablations across several random seeds, and check whether the accuracy gap over the RFB baseline and the CASFM ablation gap persist; if either collapses to within run-to-run variation, the main performance claim is not supported.

Watch

Extended reading notes

Core claim

The paper's central claim is that explicitly modeling inter-eye relations improves binocular fundus classification beyond what monocular models or simple feature concatenation achieve. The authors argue that asymmetric lesion patterns, such as hemorrhage in one eye, and symmetric co-evolving patterns, such as cup-to-disc changes in both eyes, are complementary diagnostic signals, and they design modules to capture each. DMS-Net combines a weight-sharing Siamese ResNet-152 backbone with OSIM for multi-scale pooling, CASFM for cross-eye semantic fusion, and the parallel CCAM and CIAM modules for contrastive and integrative alignment. Reported results on ODIR-5K show the complete model at 82.9% accuracy, 84.5% recall, 83.2% kappa, and 0.973 AUC, with ablations attributing the largest accuracy drop to removing CASFM.

Load-bearing premise

The reported performance rests on an unreported division of the ODIR-5K data into training and test sets and a single training run, so the small accuracy gaps over comparison models could reflect which patients were used for testing rather than the architecture alone.

Editorial extensions

If this is right

On the ODIR-5K benchmark, DMS-Net reports higher accuracy, recall, kappa, F1, and AUC than every backbone and module variant the paper compares against.
Removing CASFM costs the most performance, indicating that cross-eye semantic fusion is the main contributor among the proposed modules.
Removing either CCAM or CIAM lowers accuracy, supporting the claim that asymmetric and symmetric lesion information are complementary signals.
The OSIM module outperforms ASPP, SPPF, SPP, simSPPF, and RFB in the same backbone setup, suggesting multi-scale pooling with spatial attention is well suited to diffuse fundus lesions.
Deeper ResNet backbones improve results monotonically, so backbone capacity remains a relevant factor even with the Siamese pairing.

Reading between the lines

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

The same contrast-and-integration design could be applied to longitudinal same-eye image pairs, where the two inputs are the same retina at two time points and the contrast signal becomes disease progression.
Adapting the fused representation to output per-eye risk scores, instead of one patient-level label, would make the model more directly useful for asymmetric-disease staging.
Because the reported gains over the RFB baseline are small, an independent replication with a fixed public split and multiple seeds would test whether the module improvements generalize beyond the paper's experimental setup.

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.

Desk Editor's Note

read the letter

Colleague,

The short version: DMS-Net is a sensible engineering combo of Siamese ResNet-152 with multi-scale pooling and attention modules, and its ablations are internally consistent. But the paper claims state-of-the-art on ODIR-5K without reporting the split, seeds, or any comparison to published binocular classifiers. Those numbers are currently unverifiable, and the edge over RFB (0.5% accuracy) is well within plausible run-to-run noise.

What is new: the particular stack (OSIM + CASFM + CCAM + CIAM) is not in the cited prior work, and the ablation tables show each module helps when removed. That is decent evidence the architecture is doing something. The writing is clear, and the figures are fine. The paper is honest in that it says code will be released, but that is a promise, not a result.

Where it is soft: the elephant is the evaluation. Section IV describes preprocessing and CutMix, but never says whether the split is patient-level, whether augmentation is done before or after splitting, how many seeds, or if the task is multi-class or multi-label. If CutMix is applied on the pooled data before partition, augmented samples built from test images can leak into training. That alone could explain the gains. The paper also compares only to its own backbone and module variants; the abstract's 'state-of-the-art' has no external reference. The reported differences between OSIM and RFB are tiny, and with no confidence intervals the selectivity of the module is not demonstrated.

I would not say the architecture is wrong. The ablation pattern is sensible and the idea of explicitly modeling asymmetric versus symmetric interocular pathology has merit. But as written, the empirical claim is a set of numbers in need of verification. I would send it to a serious referee with the explicit demand that the authors release the exact split, seeds, and code, and add comparisons to at least the Zeng et al. and Nirthika et al. methods they cite. If the numbers hold under a defined protocol, this is a reasonable incremental contribution to the binocular fundus literature.

Who it is for: researchers working on paired-eye diagnosis, especially with Siamese networks, will find the module design worth a look. The framework is plausible but the headline result should not be trusted until the method section is complete.

Recommendation: merit peer review, but conditional on the authors fixing the evaluation transparency. Not a desk reject.

Referee Report

4 major / 5 minor

Summary. The manuscript introduces DMS-Net, a Siamese-network architecture for classifying paired left/right fundus photographs from ODIR-5K. It combines a weight-shared ResNet-152 backbone with four proposed modules (OSIM, CASFM, CCAM, CIAM) and reports accuracy 82.9%, recall 84.5%, and Cohen's kappa 83.2%, claiming state-of-the-art performance on the ODIR-5K benchmark. Three ablation tables compare backbone choices, module removal, and alternative multi-scale pooling modules.

Significance. If the reported results are reproducible and the comparisons are fair, the paper offers a modular, well-motivated architecture for binocular fundus classification. The strength is the internal consistency of the ablations: removing CASFM, CCAM, or CIAM degrades most metrics, and OSIM generally outperforms ASPP/SPP-style variants. However, the headline state-of-the-art claim cannot be assessed without an external comparison and a specified evaluation protocol, so the current significance is moderate.

major comments (4)

[Section IV / V] The evaluation protocol is not specified. Section IV describes preprocessing and augmentation but never states how ODIR-5K is split into training, validation, and test sets, whether the split is at patient level (to prevent left/right images of the same patient from appearing in both training and test), how many random seeds or runs are averaged, or the exact task formulation (single-label eight-class versus multi-label). CutMix is described as applied to illumination-corrected images (Section IV, Figures 4-5) with no statement about its order relative to the split; if augmentation is performed before partitioning, synthetic samples can bridge train and test. Because the abstract's numbers are the central claim, this missing protocol is load-bearing and must be supplied before the results can be evaluated.
[Section V, Tables I-III] The label 'state-of-the-art' is unsupported: all comparisons are against DMS-Net variants (backbones, ablated modules, or replacement pooling modules). No previously published binocular fundus classifier or ODIR-5K result is included as an external baseline. To substantiate state-of-the-art, the authors should add published methods on ODIR-5K (and, if available, other binocular datasets) evaluated under the same protocol.
[Table III] The claimed superiority of OSIM over the strongest alternative is within plausible noise. OSIM achieves accuracy 0.829 and kappa 0.832 versus RFB with 0.824 and 0.828, differences of 0.5 and 0.4 percentage points, respectively. Without multiple seeds, confidence intervals, or significance tests, these differences do not establish that OSIM outperforms RFB. The conclusion's statement that OSIM improves kappa by 0.8% over the best baseline is also inconsistent with these numbers.
[Section V] No implementation details are reported: optimizer, learning-rate schedule, batch size, input resolution, number of training epochs, and class-imbalance handling beyond 'class distribution preservation' are absent. This further impedes reproducibility, especially given the statement that code and processed data will be released subsequently.

minor comments (5)

[Section III-C, Eq. (12)] T_left and T_right are used in the residual connection but never defined; specify their relationship to F'_left/right, and also define W^O.
[Conclusion, Section VI] 'Improving Kappa coefficient by 0.8% over the best baseline' conflicts with Table III, where the best baseline RFB has kappa 0.828 and OSIM has 0.832; please correct the percentage.
[References [5] and [6]] References [5] and [6] are identical; one of them appears to be a duplicate citation that should be replaced or removed.
[Section III heading] 'THEPROPSOEDMETHOD' should be 'THE PROPOSED METHOD'.
[Section IV] The relation between Figures 3-5 and the final training set is unclear; specify the augmentation probability and mixing coefficient for CutMix, and clarify which classes correspond to the eight abbreviations.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the reported results are empirical benchmark measurements; unsupported SOTA labeling and missing split details are reproducibility/correctness concerns, not circular reasoning.

full rationale

The paper's central claim is an empirical accuracy measurement on ODIR-5K (Acc 82.9%, Recall 84.5%, Kappa 83.2%), obtained by training the proposed DMS-Net and evaluating it, with ablations in Tables I-III. No equation defines a predicted quantity in terms of the reported metric. OSIM is constructed from standard pooling and spatial attention operations (Eqs. 1-7), CASFM from pooling, cross-attention and residual fusion (Eqs. 8-13), and CCAM/CIAM are described as cross-attention guided dense blocks; none are fitted to the reported test numbers. The paper does not invoke a self-citation chain or uniqueness theorem to force the architecture choice, and no fitted parameter is renamed as a prediction. The absence of a described train/validation/test split, the lack of confidence intervals, and the omission of external state-of-the-art comparisons undermine the 'state-of-the-art' label and reproducibility, but these are verification/correctness issues, not circularity. Per the hard rules, unsupported claims and missing protocol details do not constitute circular derivation, so the honest finding is no significant circularity.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The paper introduces no new physical or conceptual entities beyond neural network modules. The free parameters are the learned weights of the proposed modules, and the key domain assumptions are label correctness, augmentation fidelity, transferability of pretrained features, and representativeness of the undisclosed data split.

free parameters (4)

Learnable modality-adaptive weights lambda_left and lambda_right = not reported
Used in Equation 9 to blend max-pooled and average-pooled features in CASFM; fitted on ODIR-5K.
Adaptive fusion parameters alpha_left/right and beta_left/right = not reported
Used in Equation 12 for parameter-adaptive residual connections in CASFM; fitted during training.
Positional embedding parameters P0 and Wp = not reported
Introduced in Equation 10 to compensate for spatial positional deviations in binocular features; learned parameters.
All trained weights of OSIM, CCAM, CIAM, and the classifier = not reported
The convolution, attention, and classification weights learned from data; standard for deep models but not enumerated.

assumptions (4)

domain assumption ODIR-5K ground-truth labels are clinically accurate.
The paper uses physician diagnostic keywords as labels without independent adjudication or quality audit.
domain assumption CutMix augmentation preserves the original class label for augmented pairs.
Section IV-A states that CutMix integrates local regions of homogeneous samples while preserving original class labels; if this fails, training labels are noisy.
domain assumption Pretrained ImageNet weights transfer usefully to fundus images.
The backbone is a pretrained ResNet-152; the paper provides no analysis of domain shift between ImageNet and fundus photographs.
domain assumption The unstated train/validation/test split is representative.
No split or seed protocol is reported, so all performance numbers depend on this implicit and unverified choice.

how reviews work

0 comments

Cite this review

Pith. "Pith review of DMS-Net:Dual-Modal Multi-Scale Siamese Network for Binocular Fundus Image Classification." pith.science (2026). https://pith.science/paper/ZYFVYT4S

@misc{pith2026250418046,
  author       = {Pith},
  title        = {Pith review of: DMS-Net:Dual-Modal Multi-Scale Siamese Network for Binocular Fundus Image Classification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZYFVYT4S}},
  note         = {Machine review of arXiv:2504.18046}
}

read the original abstract

Ophthalmic diseases pose a significant global health burden. However, traditional diagnostic methods and existing monocular image-based deep learning approaches often overlook the pathological correlations between the two eyes. In practical medical robotic diagnostic scenarios, paired retinal images (binocular fundus images) are frequently required as diagnostic evidence. To address this, we propose DMS-Net-a dual-modal multi-scale siamese network for binocular retinal image classification. The framework employs a weight-sharing siamese ResNet-152 architecture to concurrently extract deep semantic features from bilateral fundus images. To tackle challenges like indistinct lesion boundaries and diffuse pathological distributions, we introduce the OmniPool Spatial Integrator Module (OSIM), which achieves multi-resolution feature aggregation through multi-scale adaptive pooling and spatial attention mechanisms. Furthermore, the Calibrated Analogous Semantic Fusion Module (CASFM) leverages spatial-semantic recalibration and bidirectional attention mechanisms to enhance cross-modal interaction, aggregating modality-agnostic representations of fundus structures. To fully exploit the differential semantic information of lesions present in bilateral fundus features, we introduce the Cross-Modal Contrastive Alignment Module (CCAM). Additionally, to enhance the aggregation of lesion-correlated semantic information, we introduce the Cross-Modal Integrative Alignment Module (CIAM). Evaluation on the ODIR-5K dataset demonstrates that DMS-Net achieves state-of-the-art performance with an accuracy of 82.9%, recall of 84.5%, and a Cohen's kappa coefficient of 83.2%, showcasing robust capacity in detecting symmetrical pathologies and improving clinical decision-making for ocular diseases. Code and the processed dataset will be released subsequently.

Figures

Figures reproduced from arXiv: 2504.18046 by the authors.

**Figure 2.** The architecture of DMS-Net. SPP4×4(x) = interp avgpool(4,4)(x),size = H × W (2) In the formula, x ∈ R H×W×C denotes the high-level semantic features extracted by the ResNet152 network. Here, interp represents bilinear interpolation mode = bilinear to restore the feature map to its original dimensions H × W. This is followed by the integration of Global Max Pooling to enhance feature responses in salient lesion … view at source ↗

**Figure 5.** The first row displays four pairs of original fundus [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗

Figures from the paper (2 more)

**Figure 3.** Figure 3: Comparison of raw data versus augmented data [PITH_FULL_IMAGE:figures/full_fig_p006_3.png]

**Figure 4.** Figure 4: The first row displays four pairs of original fun [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]

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

Model Confidence-Guided Multi-Image Fusion of Fundus Images for Diabetic Retinopathy Diagnosis
eess.IV 2026-07 conditional novelty 5.0 of 10

Confidence-guided transformer fusion of multiple fundus images yields higher balanced accuracy and sensitivity at matched coverage than quality-cascade or single-image baselines on mBRSET and BRSET.

Reference graph

Works this paper leans on

39 extracted references · 37 canonical work pages · cited by 1 Pith paper

[1]

World Health Organization,World report on vision, Geneva, 2019

work page 2019
[2]

Global prevalence and major risk factors of diabetic retinopathy,

J. W. J. Yau, S. L. Rogers, R. Kawasaki, E. L. Lamoureux, J. W. Kowalski, T. Bek, S. J. Chen, J. M. Dekker, A. Fletcher, J. Grauslund, S. Haffner, R. F. Hamman, M. K. Ikram, T. Kayama, B. E. K. Klein, R. Klein, S. Krishnaiah, K. Mayurasakorn, J. P. O’Hare, T. J. Orchard, M. Porta, M. Rema, M. S. Roy, T. Sharma, J. Shaw, H. Taylor, J. M. Tielsch, R. Varma,...

work page 2012
[3]

Pivotal trial of an autonomous ai-based diagnostic system for detection of diabetic retinopathy in primary care offices,

M. D. Abràmoff, P. T. Lavin, M. Birch, N. Shah, and J. C. Folk, “Pivotal trial of an autonomous ai-based diagnostic system for detection of diabetic retinopathy in primary care offices,”NPJ digital medicine, vol. 1, no. 1, p. 39, 2018

work page 2018
[4]

Development and validation of a deep learning system for diabetic retinopathy and related eye diseases using retinal images from multiethnic populations with diabetes,

D. S. W. Ting, C. Y .-L. Cheung, G. Lim, G. S. W. Tan, N. D. Quang, A. Gan, H. Hamzah, R. Garcia-Franco, I. Y . San Yeo, S. Y . Lee et al., “Development and validation of a deep learning system for diabetic retinopathy and related eye diseases using retinal images from multiethnic populations with diabetes,”Jama, vol. 318, no. 22, pp. 2211–2223, 2017

work page 2017
[5]

Multi-modal deep learning and its applications in oph- thalmic artificial intelligence,

L. Xirong, “Multi-modal deep learning and its applications in oph- thalmic artificial intelligence,”Medical Journal of Peking Union Medical College Hospital, vol. 12, no. 5, pp. 602–607, 2021

work page 2021
[6]

Multi-modal deep learning and its applications in ophthalmic artificial intelligence,

——, “Multi-modal deep learning and its applications in ophthalmic artificial intelligence,”Medical Journal of Peking Union Medical College Hospital, vol. 12, no. 5, pp. 602–607, 2021

work page 2021
[7]

Global causes of blindness and distance vision impairment 1990–2020: a systematic review and meta-analysis,

S. R. Flaxman, R. R. Bourne, S. Resnikoff, P. Ackland, T. Braithwaite, M. V . Cicinelli, A. Das, J. B. Jonas, J. Keeffe, J. H. Kempenet al., “Global causes of blindness and distance vision impairment 1990–2020: a systematic review and meta-analysis,”The Lancet Global Health, vol. 5, no. 12, pp. e1221–e1234, 2017

work page 1990
[8]

A machine learning approach to medical image classification: Detecting age-related macular degeneration in fundus images,

A. García-Floriano, Á. Ferreira-Santiago, O. C. Nieto, and C. Yáñez- Márquez, “A machine learning approach to medical image classification: Detecting age-related macular degeneration in fundus images,”Comput. Electr. Eng., vol. 75, pp. 218–229, 2019

work page 2019

Show all 39 references

[9]

A lightweight cnn for diabetic retinopathy classification from fundus images,

G. S., V . P. Gopi, and P. Palanisamy, “A lightweight cnn for diabetic retinopathy classification from fundus images,”Biomedical Signal Processing and Control, vol. 62, p. 102115, 2020. [Online]. Available: https://www.sciencedirect.com/science/article/pii/ S1746809420302676

2020
[10]

Deep learning architecture based on segmented fundus image features for classification of diabetic retinopathy,

S. Das, K. Kharbanda, S. M, R. Raman, and E. D. D, “Deep learning architecture based on segmented fundus image features for classification of diabetic retinopathy,”Biomedical Signal Processing and Control, vol. 68, p. 102600, 2021. [Online]. Available: https: //www.sciencedire...

2021
[11]

Multi-label classification of fundus images with efficientnet,

J. Wang, L. Yang, Z. Huo, W. He, and J. Luo, “Multi-label classification of fundus images with efficientnet,”IEEE Access, vol. 8, pp. 212 499– 212 508, 2020

2020
[12]

A deep neural network and machine learning approach for retinal fundus image classification,

R. Thanki, “A deep neural network and machine learning approach for retinal fundus image classification,”Healthcare Analytics, vol. 3, p. 100140, 2023. [Online]. Available: https: //www.sciencedirect.com/science/article/pii/S2772442523000072

2023
[13]

Fundus disease image classification based on improved transformer,

H. Yang, J. Chen, and M. Xu, “Fundus disease image classification based on improved transformer,” in2021 International Conference on Neuromorphic Computing (ICNC), 2021, pp. 207–214

2021
[14]

Mil-vt: Multiple instance learning enhanced vision trans- former for fundus image classification,

S. Yu, K. Ma, Q. Bi, C. Bian, M. Ning, N. He, Y . Li, H. Liu, and Y . Zheng, “Mil-vt: Multiple instance learning enhanced vision trans- former for fundus image classification,” inMedical Image Computing and Computer Assisted Intervention – MICCAI 2021, M. de Bruijne, P. C. Cat...

2021
[15]

Signature verification using a siamese time delay neural network,

J. Bromley, I. Guyon, Y . LeCun, E. Säckinger, and R. Shah, “Signature verification using a siamese time delay neural network,” inAdvances in Neural Information Processing Systems 6, [7th NIPS Conference, Denver, Colorado, USA, 1993], J. D. Cowan, G. Tesauro, and J. Alspector,...

1993
[16]

Plant leaves classification: A few-shot learning method based on siamese network,

B. Wang and D. Wang, “Plant leaves classification: A few-shot learning method based on siamese network,”IEEE Access, vol. 7, pp. 151 754– 151 763, 2019

2019
[17]

Primary open-angle glaucoma diagnosis from optic disc photographs using a siamese network,

M. Lin, L. Liu, M. Gordon, M. Kass, F. Wang, S. H. Van Tassel, and Y . Peng, “Primary open-angle glaucoma diagnosis from optic disc photographs using a siamese network,”Ophthalmology Science, vol. 2, no. 4, p. 100209, 2022. [Online]. Available: https: //www.sciencedirect.com/s...

2022
[18]

Automated diabetic retinopathy detection based on binocular siamese-like convolutional neural network,

X. Zeng, H. Chen, Y . Luo, and W. B. Ye, “Automated diabetic retinopathy detection based on binocular siamese-like convolutional neural network,”IEEE Access, vol. 7, pp. 30 744–30 753, 2019

2019
[19]

Siamese network based fine grained classification for diabetic retinopathy grading,

R. Nirthika, S. Manivannan, and A. Ramanan, “Siamese network based fine grained classification for diabetic retinopathy grading,”Biomedical Signal Processing and Control, vol. 78, p. 103874, 2022

2022
[20]

An advanced deep learning method to detect and classify diabetic retinopathy based on color fundus images,

P. Akella and R. Kumar, “An advanced deep learning method to detect and classify diabetic retinopathy based on color fundus images,” Graefes Arch Clin Exp Ophthalmol, vol. 262, pp. 231 – 247, 2024. [Online]. Available: https://doi.org/10.1007/s00417-023-06181-3

2024 doi
[21]

Adaptive multiscale retinal diagnosis: A hybrid trio-model approach for comprehensive fundus multi-disease detection leveraging transfer learning and siamese networks,

Y . S. Inan, “Adaptive multiscale retinal diagnosis: A hybrid trio-model approach for comprehensive fundus multi-disease detection leveraging transfer learning and siamese networks,”CoRR, vol. abs/2405.18449, 2024

2024 arXiv
[22]

Automated fundus ultrasound image classification based on siamese convolutional neural networks with multi-attention,

J. Tan, Y . Dong, and J. Li, “Automated fundus ultrasound image classification based on siamese convolutional neural networks with multi-attention,”BMC Med Imaging, vol. 23, no. 89, 2023. [Online]. Available: https://doi.org/10.1186/s12880-023-01047-w

2023 doi
[23]

Cross-image siamese graph convolutional network for fine-grained image retrieval in diabetic retinopathy,

F. Chen, W. Zhao, and X. Zhou, “Cross-image siamese graph convolutional network for fine-grained image retrieval in diabetic retinopathy,”Biomedical Signal Processing and Control, vol. 92, p. 106045, 2024. [Online]. Available: https://www.sciencedirect.com/ science/article/pii...

2024
[24]

Effective diabetic retinopathy classification with siamese neural network: A strategy for small dataset challenges,

M. Tariq, V . Palade, and Y . Ma, “Effective diabetic retinopathy classification with siamese neural network: A strategy for small dataset challenges,”IEEE Access, vol. 12, pp. 182 814–182 827, 2024

2024
[25]

Madanan, S

M. Madanan, S. S. Gunasekaran, M. A. Mahmoud, J. S. Dhillon, S. Mostafa, and N. N. S. Nidzam,Artificial Intelligence Methods and Image Recognition Techniques in Ophthalmic Robotic Surgery: A Review. Cham: Springer Nature Switzerland, 2025, pp. 345–364. [Online]. Available: htt...

2025 doi
[26]

A. M. Khan, D. Kallogjeri, and J. F. Piccirillo, “51152 efficacy of bimodal visual-olfactory training in patients with covid-19 resultant hyposmia or anosmia using patient-preferred scents (volt trial - visual- olfactory training),”Journal of Clinical and Translational Science...

2021
[27]

An automatic drug injection device with spatial micro-force perception guided by an microscopic image for robot-assisted ophthalmic surgery,

Z. Li, P. Fu, B.-T. Wei, J. Wang, A.-L. Li, M.-J. Li, and G.-B. Bian, “An automatic drug injection device with spatial micro-force perception guided by an microscopic image for robot-assisted ophthalmic surgery,”Frontiers in Robotics and AI, vol. V olume 9 - 2022, 2022. [Onlin...

2022
[28]

Robotic process automation support in telemedicine: Glaucoma screening usage case,

S. Thainimit, P. Chaipayom, N. Sa-arnwong, D. Gansawat, S. Petchyim, and S. Pongrujikorn, “Robotic process automation support in telemedicine: Glaucoma screening usage case,”Informatics in Medicine Unlocked, vol. 31, p. 101001, 2022. [Online]. Available: https: //www.sciencedi...

2022
[29]

New directions for ophthalmic oct – handhelds, surgery, and robotics,

J. Foust, M. McCloud, A. Narawane, R. M. Trout, X. Chen, A.-H. Dhalla, J. D. Li, C. Viehland, M. Draelos, L. Vajzovic, R. P. McNabb, A. N. Kuo, and C. A. Toth, “New directions for ophthalmic oct – handhelds, surgery, and robotics,”Translational Vision Science & Technology, vol...

2025 doi
[30]

Evaluating the accuracy of the ophthalmologist robot for multiple blindness-causing eye diseases: a multicentre, prospective study protocol,

Q. Li, J. Tan, H. Xie, X. Zhang, Q. Dai, Z. Li, L. L. Yan, and W. Chen, “Evaluating the accuracy of the ophthalmologist robot for multiple blindness-causing eye diseases: a multicentre, prospective study protocol,”BMJ Open, vol. 14, no. 3, 2024. [Online]. Available: https://bm...

2024
[31]

Active motion cancellation for robotic optical coherence tomography of moving eyes: A nystagmus phantom study,

H. Pan, C. W. Lim, K. King, R. Guan, and M. Draelos, “Active motion cancellation for robotic optical coherence tomography of moving eyes: A nystagmus phantom study,” in2024 International Symposium on Medical Robotics (ISMR), 2024, pp. 1–7

2024
[32]

Robotic optical coherence tomography of human subjects with posture-invariant head and eye alignment in six degrees of freedom,

M. Draelos, P. Ortiz, A. Narawane, R. P. McNabb, A. N. Kuo, and J. A. Izatt, “Robotic optical coherence tomography of human subjects with posture-invariant head and eye alignment in six degrees of freedom,” in2023 International Symposium on Medical Robotics (ISMR), 2023, pp. 1–7

2023
[33]

Mobile robotic optical coherence tomography system for ophthalmic imaging in clinical environments,

G. Zhou, H. Pan, S. Staudinger, Y . Wang, J. Liu, S. Fleifil, C. Jin, N. Valikodath, R. P. McNabb, A. N. Kuo, and M. Draelos, “Mobile robotic optical coherence tomography system for ophthalmic imaging in clinical environments,” in2025 International Symposium on Medical Robotic...

2025
[34]

Odir-5k,

KOTO, “Odir-5k,” https://www.heywhale.com/mw/dataset/ 5e95e3ede7ec38002d0351f6, 2020

2020
[35]

Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected crfs,

L. Chen, G. Papandreou, I. Kokkinos, K. Murphy, and A. L. Yuille, “Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected crfs,”IEEE Trans. Pattern Anal. Mach. Intell., vol. 40, no. 4, pp. 834–848, 2018

2018
[36]

What is yolov5: A deep look into the internal features of the popular object detector,

R. Khanam and M. Hussain, “What is yolov5: A deep look into the internal features of the popular object detector,”arXiv preprint arXiv:2407.20892, 2024

2024 arXiv
[37]

Spatial pyramid pooling in deep convolutional networks for visual recognition,

K. He, X. Zhang, S. Ren, and J. Sun, “Spatial pyramid pooling in deep convolutional networks for visual recognition,”IEEE Trans. Pattern Anal. Mach. Intell., vol. 37, no. 9, pp. 1904–1916, 2015

1904
[38]

Receptive field block net for accurate and fast object detection,

S. Liu, D. Huang, and Y . Wang, “Receptive field block net for accurate and fast object detection,” inECCV 2018, ser. Lecture Notes in Computer Science, V . Ferrari, M. Hebert, C. Sminchisescu, and Y . Weiss, Eds., vol. 11215. Springer, 2018, pp. 404–419

2018
[39]

Yolov6: A single-stage object detection framework for industrial applications,

C. Li, L. Li, H. Jiang, K. Weng, Y . Geng, L. Li, Z. Ke, Q. Li, M. Cheng, W. Nieet al., “Yolov6: A single-stage object detection framework for industrial applications,”arXiv preprint arXiv:2209.02976, 2022

2022 arXiv

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.

[1] [1]

World Health Organization,World report on vision, Geneva, 2019

work page 2019

[2] [2]

Global prevalence and major risk factors of diabetic retinopathy,

J. W. J. Yau, S. L. Rogers, R. Kawasaki, E. L. Lamoureux, J. W. Kowalski, T. Bek, S. J. Chen, J. M. Dekker, A. Fletcher, J. Grauslund, S. Haffner, R. F. Hamman, M. K. Ikram, T. Kayama, B. E. K. Klein, R. Klein, S. Krishnaiah, K. Mayurasakorn, J. P. O’Hare, T. J. Orchard, M. Porta, M. Rema, M. S. Roy, T. Sharma, J. Shaw, H. Taylor, J. M. Tielsch, R. Varma,...

work page 2012

[3] [3]

Pivotal trial of an autonomous ai-based diagnostic system for detection of diabetic retinopathy in primary care offices,

M. D. Abràmoff, P. T. Lavin, M. Birch, N. Shah, and J. C. Folk, “Pivotal trial of an autonomous ai-based diagnostic system for detection of diabetic retinopathy in primary care offices,”NPJ digital medicine, vol. 1, no. 1, p. 39, 2018

work page 2018

[4] [4]

Development and validation of a deep learning system for diabetic retinopathy and related eye diseases using retinal images from multiethnic populations with diabetes,

D. S. W. Ting, C. Y .-L. Cheung, G. Lim, G. S. W. Tan, N. D. Quang, A. Gan, H. Hamzah, R. Garcia-Franco, I. Y . San Yeo, S. Y . Lee et al., “Development and validation of a deep learning system for diabetic retinopathy and related eye diseases using retinal images from multiethnic populations with diabetes,”Jama, vol. 318, no. 22, pp. 2211–2223, 2017

work page 2017

[5] [5]

Multi-modal deep learning and its applications in oph- thalmic artificial intelligence,

L. Xirong, “Multi-modal deep learning and its applications in oph- thalmic artificial intelligence,”Medical Journal of Peking Union Medical College Hospital, vol. 12, no. 5, pp. 602–607, 2021

work page 2021

[6] [6]

Multi-modal deep learning and its applications in ophthalmic artificial intelligence,

——, “Multi-modal deep learning and its applications in ophthalmic artificial intelligence,”Medical Journal of Peking Union Medical College Hospital, vol. 12, no. 5, pp. 602–607, 2021

work page 2021

[7] [7]

Global causes of blindness and distance vision impairment 1990–2020: a systematic review and meta-analysis,

S. R. Flaxman, R. R. Bourne, S. Resnikoff, P. Ackland, T. Braithwaite, M. V . Cicinelli, A. Das, J. B. Jonas, J. Keeffe, J. H. Kempenet al., “Global causes of blindness and distance vision impairment 1990–2020: a systematic review and meta-analysis,”The Lancet Global Health, vol. 5, no. 12, pp. e1221–e1234, 2017

work page 1990

[8] [8]

A machine learning approach to medical image classification: Detecting age-related macular degeneration in fundus images,

A. García-Floriano, Á. Ferreira-Santiago, O. C. Nieto, and C. Yáñez- Márquez, “A machine learning approach to medical image classification: Detecting age-related macular degeneration in fundus images,”Comput. Electr. Eng., vol. 75, pp. 218–229, 2019

work page 2019