ORViT-DR: Ordinally-Robust Hybrid ViT for Low-Resolution Diabetic Retinopathy Grading
Authors: Soumit Kumar Kundu, Nabil Ashab, Bidhan Biswas, Shahadat Hossain Sohag, Saif Mahmud Parvez, Souvik Kumar Kundu, Zunayed Ahmed Rafi
Organizations: Department of Computer Science and Engineering, Dhaka International University, Dhaka, Bangladesh · KUET, Khulna, Bangladesh · Bangladesh Bank, Dhaka, Bangladesh
Diabetic retinopathy (DR) is one of the main causes of impaired vision. A good and reliable automated grading system can make the screening process safer and more accurate. Because DR stages progress gradually, the task of grading disease severity naturally follows an ordinal structure in which neighboring classes share similar visual characteristics. In this study, ORViT-DR, a hybrid deep learning framework, is designed to improve DR grading from low-resolution retinal images. The proposed approach combines convolutional feature extraction with transformer-based global context modeling through a pre-trained ViT-Hybrid backbone, which integrates BiT-ResNetv2 with a Vision Transformer architecture. The approach is tested on the RetinaMNIST subset of the MedMNISTv2 dataset, which contains 28x28 retinal fundus images annotated with five levels of disease severity. To promote stable training and better feature learning, the training strategy applies progressive layer unfreezing, layer-wise learning rate decay, exponential moving average (EMA) parameter updates, and ensemble-based prediction during inference. Experimental results on the official RetinaMNIST test set show that the proposed method achieves 57.00% classification accuracy, along with a quadratic weighted kappa score of 0.5963 and a macro-F1 score of 0.4293. These results suggest that hybrid CNN-Transformer architectures can provide effective representations for ordinal retinal image analysis.
Figures & tables
Split
Samples
Classes
Image Resolution
Train
1080
5 (No DR, Mild, Moderate, Severe, Proliferative)
28 × 28 (resized to 384 × 384 by the model processor)
Validation
120
Test
400
Table I: Baseline dataset statistics
Figure 1: Class distribution of the RetinaMNIST training split (imbalanced toward No DR).
Figure 2: ORViT-DR whole architecture. It is based on a hybrid CNN–Transformer backbone.
Partial unfreezing of transformer encoder + LayerNorm + head
LLRD
Decay factor 0.85, base LR 3×10−5 , head LR 3×10−4
Regularization
Label smoothing 0.1, dropout in head, gradient clipping (1.0)
MixUp/CutMix
MixUp α=0.3 ; CutMix α=1.0 , randomly sampled per batch
EMA
Decay 0.999 (applied on trainable parameters)
Table II: Training configuration and fine-tuning strategy
Figure 3: Confusion matrices for the best single checkpoint (56.25%) and top-3 uniform ensemble (57.00%).
Figure 4: Training curves produced by the implementation: (a) validation accuracy and (b) train/validation loss across stages.
Configuration
Δ Single + TTA (pp)
Δ Ensemble + TTA (pp)
No LLRD vs. With LLRD
-8.00
-2.25
Table III: Ablation study under a fixed seed setting (seed 42). Results are reported as relative performance changes.
Figure 5: One-vs-rest ROC curves for the top-3 ensemble. The per-class AUCs are shown in the legend.
Figure 6: Representative Grad-CAM visualizations on RetinaMNIST samples. From left to right: original image, Grad-CAM heatmap, and heatmap overlay. Warmer regions indicate a stronger positive contribution to the predicted class.
Diabetic retinopathy (DR) is a leading cause of vision impairment worldwide, and automated grading systems play a crucial role in large-scale screening programs. However, deep learning models often exhibit degraded performance when deployed across datasets acquired under different imaging conditions. This study presents a robust dual-resolution deep learning framework for DR grading that integrates attention-based feature fusion with ordinal regression to improve cross-dataset generalization. The proposed method employs two parallel EfficientNet backbones operating at different spatial resolutions to capture complementary retinal features. A learnable attention mechanism adaptively fuses multi-resolution representations, while an ordinal regression formulation based on the cumulative link model (CORAL) explicitly accounts for the ordered nature of DR severity levels. To mitigate domain discrepancies between datasets, a preprocessing strategy combining circular cropping, contrast enhancement, and histogram matching is applied. The model was trained on the APTOS 2019 dataset and evaluated on both an internal validation split and an external Messidor-2 test set. Experimental results demonstrate strong grading performance, achieving a quadratic weighted kappa (QWK) of 0.88 on the APTOS validation set and 0.68 on the unseen Messidor-2 dataset, indicating improved robustness for cross-dataset DR grading applications.
Afshan Hashmi
Tuwaiq Academy, Tuwaiq Research and Development Centre, Riyadh, Saudi Arabia
Diabetic Retinopathy (DR) progresses as a continuous and irreversible deterioration of the retina, following a well-defined clinical trajectory from mild to severe stages. However, most existing ordinal regression approaches model DR severity as a set of static, symmetric ranks, capturing relative order while ignoring the inherent unidirectional nature of disease progression. As a result, the learned feature representations may violate biological plausibility, allowing implausible proximity between non-consecutive stages or even reverse transitions. To bridge this gap, we propose Directed Ordinal Diffusion Regularization (D-ODR), which explicitly models the feature space as a directed flow by constructing a progression-constrained directed graph that strictly enforces forward disease evolution. By performing multi-scale diffusion on this directed structure, D-ODR imposes penalties on score inversions along valid progression paths, thereby effectively preventing the model from learning biologically inconsistent reverse transitions. This mechanism aligns the feature representation with the natural trajectory of DR worsening. Extensive experiments demonstrate that D-ODR yields superior grading performance compared to state-of-the-art ordinal regression and DR-specific grading methods, offering a more clinically reliable assessment of disease severity. Our code is available on https://github.com/HovChen/D-ODR.
Huangwei Chen, Junhao Jia, Ruocheng Li +7
Zhejiang Key Laboratory of Accessible Perception and Intelligent Systems, College of Computer Science and Technology, Zhejiang University · Hangzhou Pujian Medical Technology Co., Ltd · Hangzhou Dianzi University +1
Diabetic retinopathy (DR) is a leading cause of preventable blindness, and automated fundus image grading can play an important role in large-scale screening. In this work, we investigate three CLIP-based approaches for five-class DR severity grading: (1) a zero-shot baseline using prompt engineering, (2) a hybrid FCN-CLIP model augmented with CBAM attention, and (3) a ranking-aware prompting model that encodes the ordinal structure of DR progression. We train and evaluate on a combined dataset of APTOS 2019 and Messidor-2 (n=5,406), addressing class imbalance through resampling and class-specific optimal thresholding. Our experiments show that the ranking-aware model achieves the highest overall accuracy (93.42%, AUROC 0.9845) and strong recall on clinically critical severe cases, while the hybrid FCN-CLIP model (92.49%, AUROC 0.99) excels at detecting proliferative DR. Both substantially outperform the zero-shot baseline (55.17%, AUROC 0.75). We analyze the complementary strengths of each approach and discuss their practical implications for screening contexts.
Sungjun Cho
Hong Kong University of Science and Technology (HKUST), Hong Kong, China