Transformer-based segmentation methods exhibit considerable potential in medical image analysis. However, their improved performance often comes with increased computational complexity, limiting their application in resource-constrained medical settings. Prior methods follow two independent tracks: (i) accelerating existing networks via semantic-aware routing, and (ii) optimizing token adapter design to enhance network performance. Despite directness, they encounter unavoidable defects (e.g., inflexible acceleration techniques or non-discriminative processing) limiting further improvements of quality-complexity trade-off. To address these shortcomings, we integrate these schemes by proposing the semantic-aware adapter (SarAdapter), which employs a semantic-based routing strategy, leveraging neural operators (ViT and CNN) of varying complexities. Specifically, it merges semantically similar tokens volume into low-resolution regions while preserving semantically distinct tokens as high-resolution regions. Additionally, we introduce a Mixed-adapter unit, which adaptively selects convolutional operators of varying complexities to better model regions at different scales. We evaluate our method on four medical datasets from three modalities and show that it achieves a superior balance between accuracy, model size, and efficiency. Notably, our proposed method achieves state-of-the-art segmentation quality on the Synapse dataset while reducing the number of tokens by 65.6%, signifying a substantial improvement in the efficiency of ViTs for the segmentation task.

Glaucoma is a leading cause of irreversible yet preventable blindness in working-age populations. Clinical diagnos is currently relies on functional visual field (VF) examinations to evaluate visual function and monitor disease progression, but these tests are time-consuming and require close cooperation with ophthalmologists, limiting their practicality. Although deep learning has shown promise for glaucoma diagnosis, most models have focused on structural changes in fundus and OCT images without linking them to functional VF outcomes. To bridge this gap, the HDMI Laboratory, in collaboration with the Zhongshan Ophthalmic Center of Sun Yat-sen University, organized the STAGE Challenge: Structural-Functional Transition in Glaucoma Assessment. The challenge explores predicting functional VF indicators-mean deviation, sensitivity maps, and pattern deviation probability maps-directly from structural OCT images. A dataset of 401 OCT volumes (each with 256 cross-sectional images) is released with corresponding VF labels and demographic data, along with a standardized evaluation framework to ensure fair comparison. This paper summarizes the methods of the seven finalist teams and analyzes their results. All teams employed dual-branch architectures integrating OCT with tabular data, and those using task-specific OCT models achieved the highest performance. These findings highlight the importance of tailored deep learning strategies for linking structural imaging to functional outcomes. The STAGE Challenge thus establishes the first standardized benchmark with a large, curated dataset for this task, enabling systematic evaluation of algorithms for structure-function analysis in glaucoma, and providing a foundation for future research. Details of the competition are available at http://hdmilab.cn/competition/stage.

BACKGROUND:Despite the expanding use of flow-diverting stents (flow diverters-FDs) for middle cerebral artery (MCA) bifurcation aneurysms, their efficacy remains contentious in this complex anatomy. Current studies report conflicting complete occlusion rates (55-92%) and significant branch stenosis/occlusion risks (8-43%), highlighting unmet needs in stent placement strategy. METHODS:In this retrospective cohort study combining clinical data with computational fluid dynamics (CFD), 20 MCA bifurcation aneurysms (19 patients) treated with FDs were analyzed. Patient-specific models derived from three-dimensional digital subtraction angiography underwent CFD simulations to quantify pre/post-stent hemodynamic changes. Key parameters (inflow rate, wall shear stress, residual flow volume) were compared between aneurysms with complete versus incomplete healing, and the impact of stent placement in branches of varying diameters was analyzed. RESULTS:At a median follow-up of 9 months, complete occlusion was achieved in 55% (11/20) and partial occlusion in 25% (5/20). Though not statistically significant, hemodynamic analysis revealed the daughter artery diameter ratio (DR)-stratified efficacy. Stents with a DR between 0.65 and 0.80 placed in smaller-diameter branches more effectively reduced inflow rate, relative inflow rate, and aneurysm wall shear stress. Conversely, stents with a DR between 0.80 and 0.95 were more effective in larger-diameter branches. CONCLUSION:This study indicates that vessel DR could guide intraoperative decisions for FD placement in MCA bifurcations. Prioritizing smaller branches in certain anatomical profiles may enhance flow remodeling without increasing procedural risks. These findings suggest a hemodynamic basis for moving beyond traditional size-based approaches and could inform more precise stent selection in complex bifurcation interventions. Further validation through controlled trials is needed.

Curvilinear structure segmentation (CSS) plays a vital role in industrial applications, including medical imaging and structural health monitoring. Recently, the strong capacity of the Segment Anything Model (SAM) has inspired its downstream application in CSS tasks. To adapt SAM to CSS tasks, previous methods heavily rely on a certain number of samples and costly pixel-level annotation, which are hard to access for a new scenario. Considering this, the goal of our work is to adapt SAM in a very cost-effective setting where only a single unlabeled image is given. This is far more challenging than the typical supervised, unsupervised, or self-supervised learning manner that needs a large number of training samples. To tackle this problem, we propose a finetuning-free SAM for curvilinear structure segmentation, called curvilinear-aware prompt learning (CaPro), which aims to automatically learn visual prompts via a single unlabeled image. In the first stage, we generate extensive curvilinear structures and oriented sub-curvilinear box annotations. To increase the realism of generated curvilinear structures, we adapt these structures into real image domains via the Fourier Transform using a single real-world unlabeled image. Now, these adapted images can be used to train our oriented sub-curvilinear detector. In the second stage, we propose the curvilinear-aware discrete representation matching to filter those unreliable detection results. Afterward, these reliable detection results can be converted into informative prompts, contributing to the cost-effective SAM adaptation to CSS tasks. Experiments demonstrate the effectiveness of CaPro on medical image and crack segmentation tasks.

Existing multi-modal learning methods on fundus and OCT images mostly require both modalities to be available and strictly paired for training and testing, which appears less practical in clinical scenarios. To expand the scope of clinical applications, we formulate a novel setting, “OCT-enhanced disease recognition from fundus images”, that allows for the use of unpaired multi-modal data during the training phase, and relies on the widespread fundus photographs for testing. To benchmark this setting, we present the first large multi-modal multi-class dataset for eye disease diagnosis, MultiEYE, and propose an OCT-assisted Conceptual Distillation Approach (OCT-CoDA), which employs semantically rich concepts to extract disease-related knowledge from OCT images and leverages them into the fundus model. Specifically, we regard the image-concept relation as a link to distill useful knowledge from OCT teacher model to fundus student model, which considerably improves the diagnostic performance based on fundus images and formulates the cross-modal knowledge transfer into an explainable process. Through extensive experiments on the multi-disease classification task, our proposed OCT-CoDA demonstrates remarkable results and interpretability, showing great potential for clinical application. Our dataset and code are available at https://github.com/xmed-lab/MultiEYE.

Centre-involved Diabetic Macular Edema (ci-DME) is a major cause of vision impairment and can arise at any stage of diabetic retinopathy (DR), with increased prevalence as DR progresses. Early detection is critical for preventing vision loss. Recent advancements in imaging techniques, such as ultra-wide-field color fundus photography (UWF-CFP) and optical coherence tomography (OCT), coupled with the power of deep learning (DL), now enable more precise detection, classification, and grading of DME. The Device-Independent Diabetic Macular Edema Onset Prediction (DIAMOND) Challenge aims to develop DL models capable of predicting whether a patient will develop center-involved diabetic macular edema (ci-DME) within a year using UWF-CFP images, while preventing participants from direct data access to ensure model generalizability in real-world settings. For this challenge, we implemented advanced preprocessing techniques to mitigate out-of-domain data issues and explored multiple DL architectures. Our methodology has shown great promise, positioning our team within the Top 3 of the competition. However, due to the affiliation of some team members with the challenge organizers, we are ineligible for the prize.

Cardiovascular disease (CVD) remains the leading cause of death worldwide, requiring urgent development of effective risk assessment methods for timely intervention. While current research has introduced non-invasive and efficient approaches to predict CVD risk from retinal imaging with deep learning models, the commonly used fundus photographs and Optical Coherence Tomography (OCT) fail to capture detailed vascular features critical for CVD assessment compared with OCT angiography (OCTA) images. Moreover, existing methods typically classify CVD risk only as high or low, without providing a deeper analysis on CVD-related blood factor conditions, thus limiting prediction accuracy and clinical utility. As a result, we propose a novel multi-purpose paradigm of CVD risk assessment that jointly performs CVD risk and CVD-related condition prediction, aligning with clinical experiences. Based on this core idea, we introduce OCTA-CVD, the first OCTA dataset for CVD risk assessment, and a Vessel-Aware Mamba-based Prediction model with Informative Enhancement (VAMPIRE) based on OCTA enface images. Our proposed model aims to extract crucial vascular characteristics through two key components: (1) a Mamba-Based Directional (MBD) Module that captures fine-grained vascular trajectory features and (2) an Information-Enhanced Morphological (IEM) Module that incorporates comprehensive vessel morphology knowledge. Experimental results demonstrate that our method can surpass standard classification backbones, OCTA-based detection methods, and ophthalmologic foundation models. Our codes and the collected OCTA-CVD dataset are available at https://github.com/xmed-lab/VAMPIRE.

Cerebrovascular diseases can occur suddenly and unpredictably, making it crucial to identify high-risk individuals through screening to prevent or mitigate its impact. However, digital subtraction angiography (DSA), the current gold-standard, is difficult to apply to large-scale screening or primary healthcare settings due to its high cost, complex operation, and invasive nature. In contrast, Color Fundus Photography (CFP) can reflect related cerebrovascular diseases through retinal microvascular changes while maintaining low-cost and risk-free advantages. Nevertheless, current CFP image-based methods for predicting cerebrovascular disease mostly focus on pixel-level image features only, ignoring the correlation between arteriovenous morphology, optic disc structure and disease risk. To address this gap, we propose CVGB-Net, a method that integrates a cross-view encoder to fuse high-level semantic features, primarily capturing vascular abnormalities in the retinal vasculature caused by cerebrovascular diseases, with low-level pixel features extracted by the foundation model, RetFound, designed for ocular tasks. The fused cross-view features for each sample are then processed through a graph-based discriminator, which utilizes a graph adapter to link disease-related features across the entire dataset. This approach further enhances the model's ability to differentiate between diseased and healthy cases. To validate our approach, we present a tailored CFP-Cerebrovascular diseases Screening (CCS) dataset with 2,338 expert-diagnosed cases. Experimental results demonstrate the effectiveness of our approach, highlighting its potential for costeffective large-scale cerebrovascular diseases screening. https://github. com/glodxy/CVGB_net.

Different from traditional semi-supervised learning (SSL), semi-supervised medical image segmentation faces two significant challenges: (1) the imbalanced distribution of labeled data causes models to bias towards majority classes; (2) the distribution discrepancy between labeled and unlabeled samples induces confirmation bias in pseudo-labels. Inspired by clinical practice, where experienced doctors utilize intrinsic features from the interior of target organs to clarify ambiguous boundaries and focus on minority classes, we propose a novel boundary-aware dynamic re-weighting network (BDRN). First, we utilize edge filters to generate visually different but semantically aligned views, compelling two sub-networks to learn informative features from organ interiors and boundaries, respectively. Second, we extract the boundary and interior regions using morphological operators and introduce a shape constraint to enhance feature learning. Additionally, a conflict-adversarial module promotes segmentation consistency between different views. Finally, we propose a dynamic re-weighting strategy based on the effective number to improve attention to imbalanced classes. Experiments demonstrate that our method significantly improves segmentation performance, achieving state-of-the-art results on CT and MR images. Ablation studies further confirm the efficacy of boundary consistency constraints and dynamic re-weighting. The segmentation Dice score for minority organs (e.g., esophagus) on the Synapse dataset is improved by 17.1 %, 46.2%, and 49.4% using 10%, 20%, and 40% labeled data, respectively.

Optical coherence tomography angiography (OCTA) shows its great importance in imaging microvascular networks by providing accurate 3D imaging of blood vessels, but it relies upon specialized sensors and expensive devices. For this reason, previous works show the potential to translate the readily available 3D Optical Coherence Tomography (OCT) images into 3D OCTA images. However, existing OCTA translation methods directly learn the mapping from the OCT domain to the OCTA domain in continuous and infinite space with guidance from only a single view, i.e., the OCTA project map, resulting in suboptimal results. To this end, we propose the multi-view Tri-alignment framework for OCT to OCTA 3D image translation in discrete and finite space, named MuTri. In the first stage, we pre-train two vector-quantized variational auto-encoder (VQVAE) by reconstructing 3D OCT and 3D OCTA data, providing semantic prior for subsequent multi-view guidances. In the second stage, our multi-view tri-alignment facilitates another VQVAE model to learn the mapping from the OCT domain to the OCTA domain in discrete and finite space. Specifically, a contrastive-inspired semantic alignment is proposed to maximize the mutual information with the pre-trained models from OCT and OCTA views, to facilitate codebook learning. Meanwhile, a vessel structure alignment is proposed to minimize the structure discrepancy with the pre-trained models from the OCTA project map view, benefiting from learning the detailed vessel structure information. We also collect the first large-scale dataset, namely, OCTA2024, which contains a pair of OCT and OCTA volumes from 846 subjects. Our codes and datasets are available at: https://github.com/xmed-lab/MuTri.

The internal carotid artery (ICA) is a region with a high incidence for small- and medium-sized saccular aneurysms. However, the treatment relies heavily on the surgeon’s experience to achieve optimal outcome. Although the finite element method (FEM) and computational fluid dynamics can predict the postoperative outcomes, due to the computational complexity of traditional methods, there is an urgent need for investigating the fast but versatile approaches related to numerical simulations of flow diverters (FDs) deployment coupled with the hemodynamic analysis to determine the treatment plan. We collected the preoperative and postoperative data from 34 patients (29 females, 5 males; mean age 55.74 ± 9.98 years) who were treated with a single flow diverter for small- to medium-sized intracranial saccular aneurysms on the ICA. The constraint-based virtual deployment (CVD) method is proposed to simulate the FDs expanding outward along the vessel centerline while be constrained by the inner wall of the vessel. The results indicate that there were no significant differences in the reduction rates of wall shear stress and aneurysms neck velocity between the FEM and methods. However, the solution time of CVD was greatly reduced by 98

Purpose: Accurate diagnosis of retinal disease based on optical coherence tomography (OCT) requires scrutiny of both B-scan and en face images. The aim of this study was to investigate the effectiveness of fusing en face and B-scan images for better diagnostic performance of deep learning models. Methods: A multiview fusion network (MVFN) with a decision fusion module to integrate fast-axis and slow-axis B-scans and en face information was proposed and compared with five state-of-the-art methods: a model using B-scans, a model using en face imaging, a model using three-dimensional volume, and two other relevant methods. They were evaluated using the OCTA-500 public dataset and a private multicenter dataset with 2330 cases; cases from the first center were used for training and cases from the second center were used for external validation. Performance was assessed by averaged area under the curve (AUC), accuracy, sensitivity, specificity, and precision. Results: In the private external test set, our MVFN achieved the highest AUC of 0.994, significantly outperforming the other models (P < 0.01). Similarly, for the OCTA-500 public dataset, our proposed method also outperformed the other methods with the highest AUC of 0.976, further demonstrating its effectiveness. Typical cases were demonstrated using activation heatmaps to illustrate the synergy of combining en face and Conclusions: The fusion of en face and B-scan information is an effective strategy for improving the diagnostic accuracy of deep learning models. Translational Relevance: Multiview fusion models combining B-scan and en face images demonstrate great potential in improving AI performance for retina disease diagnosis.

Proper decision-making for endovascular treatment (EVT) is crucial in reducing complications of intracranial aneurysms (IAs) and improving the quality of patients' lives. Electronic medical records (EMRs) possess comprehensive information about patients, which provides substantial data support for EVT decisions. Machine learning (ML) models demonstrate considerable performance in handling the relationship between various clinical indicators and treatment decisions based on EMRs. However, they still face challenges regarding the data scarcity and imbalance distribution of EMRs. To address these issues, we propose a latent distribution- preserving structure that can generate authentic EMRs in a category-aware manner. We first map the EMRs into the latent space to obtain the original prototypes and generate the minority prototypes by augmenting within the latent distributions of the minority categories. The original and synthesized minority prototypes are then reconstructed and synthesized to EMRs. The reconstruction process aligns the original and synthesized data space, promoting high authenticity in data-scarce sample generation. Subsequently, the synthesized and original EMRs are merged to obtain sufficient and balanced training data for ML models. Experiment results demonstrate the effectiveness of our proposed structure in enhancing generative performance across various state-of-the-art tabular GAN and boosts the treatment-decision performance by 2.8%-7.2% in AUC, 3.4%-17.2% in F1. The resources will be publicly available.

Optical Coherence Tomography Angiography (OCTA) is a crucial tool in the clinical screening of retinal diseases, allowing for accurate 3D imaging of blood vessels through non-invasive scanning. However, the hardware-based approach for acquiring OCTA images presents challenges due to the need for specialized sensors and expensive devices. In this paper, we introduce a novel method called TransPro, which can translate the readily available 3D Optical Coherence Tomography (OCT) images into 3D OCTA images without requiring any additional hardware modifications. Our TransPro method is primarily driven by two novel ideas that have been overlooked by prior work. The first idea is derived from a critical observation that the OCTA projection map is generated by averaging pixel values from its corresponding B-scans along the Z-axis. Hence, we introduce a hybrid architecture incorporating a 3D adversarial generative network and a novel Heuristic Contextual Guidance (HCG) module, which effectively maintains the consistency of the generated OCTA images between 3D volumes and projection maps. The second idea is to improve the vessel quality in the translated OCTA projection maps. As a result, we propose a novel Vessel Promoted Guidance (VPG) module to enhance the attention of network on retinal vessels. Experimental results on two datasets demonstrate that our TransPro outperforms state-of-the-art approaches, with relative improvements around 11.4% in MAE, 2.7% in PSNR, 2% in SSIM, 40% in VDE, and 9.1% in VDC compared to the baseline method. The code is available at: https://github.com/ustlsh/TransPro.

Conventional OCT retinal disease classification methods primarily rely on fully supervised learning, which requires a large number of labeled images. However, sometimes the number of labeled images in a private domain is small but there exists a large annotated open dataset in the public domain. In response to this scenario, a new transfer learning method based on sub-domain adaptation (TLSDA), which involves a first sub-domain adaptation and then fine-tuning, was proposed in this study. Firstly, a modified deep sub-domain adaptation network with pseudo-label (DSAN-PL) was proposed to align the feature spaces of a public domain (labeled) and a private domain (unlabeled). The DSAN-PL model was then fine-tuned using a small amount of labeled OCT data from the private domain. We tested our method on three open OCT datasets, using one as the public domain and the other two as the private domains. Remarkably, with only 10% labeled OCT images (~100 images per category), TLSDA achieved classification accuracies of 93.63% and 96.59% on the two private datasets, significantly outperforming conventional transfer learning approaches. With the Gradient-weighted Class Activation Map (Grad-CAM) technique, it was observed that the proposed method could more precisely localize the subtle lesion regions for OCT image classification. TLSDA could be a potential technique for applications where only a small number of images is labeled in a private domain and there exists a public database having a large number of labeled images with domain difference.

Purpose: This study aimed to develop artificial intelligence models for predicting postoperative functional outcomes in patients with rhegmatogenous retinal detachment (RRD). Methods: A retrospective review and data extraction were conducted on 184 patients diagnosed with RRD who underwent pars plana vitrectomy (PPV) and gas tamponade. The primary outcome was the best -corrected visual acuity (BCVA) at three months after the surgery. Those with a BCVA of less than 6/18 Snellen acuity were classified into a vision impairment group. A deep learning model was developed using presurgical predictors, including ultra-widefield fundus images, structural optical coherence tomography (OCT) images of the macular region, age, gender, and preoperative BCVA. A fusion method was used to capture the interaction between different modalities during model construction. Results: Among the participants, 74 (40%) still had vision impairment after the treatment. There were significant differences in age, gender, presurgical BCVA, intraocular pressure, macular detachment, and extension of retinal detachment between the vision impairment and vision non -impairment groups. The multimodal fusion model achieved a mean area under the curve (AUC) of 0.91, with a mean accuracy of 0.86, sensitivity of 0.94, and specificity of 0.80. Heatmaps revealed that the macular involvement was the most active area, as observed in both the OCT and ultra-widefield images. Conclusions: This pilot study demonstrates that artificial intelligence techniques can achieve a high AUC for predicting functional outcomes after RRD surgery, even with a small sample size. Machine learning methods identified The macular region as the most active region. Translational Relevance: Multimodal fusion models have the potential to assist clinicians in predicting postoperative visual outcomes prior to undergoing PPV.

Rupture prediction is crucial for precise treatment and follow-up management of patients with intracranial aneurysms (IAs). Considerable machine learning (ML) methods have been proposed to improve rupture prediction by leveraging electronic medical records (EMRs), however, data scarcity and category imbalance strongly influence performance. Thus, we propose a novel data synthesis method i.e., Transformer-based conditional GAN (TransCGAN), to synthesize highly authentic and category-aware EMRs to address above challenges. Specifically, we first align feature-wise context relationship and distribution between synthetic and original data to enhance synthetic data quality. To achieve this, we first integrate the Transformer structure into GAN to match the contextual relationship by processing the long-range dependencies among clinical factors and introduce a statistical loss to maintain distributional consistency by constraining the mean and variance of the synthesis features. Additionally, a conditional module is designed to assign the category of the synthesis data, thereby addressing the challenge of category imbalance. Subsequently, the synthetic data are merged with the original data to form a large-scale and category-balanced training dataset for IAs rupture prediction. Experimental results show that using TransCGAN's synthetic data enhances classifier performance, achieving AUC of 0.89 and outperforming state-of-the-art resampling methods by 5-33 in F1 score.

Rapid endothelialization is extremely essential for the success of small-diameter tissue-engineered vascular graft (TEVG) (<6 mm) transplantation. However, severe inflammation in situ often causes cellular energy decline of endothelial cells. The cellular energy supply involved in vascular graft therapy remains unclear, and whether promoting energy supply would be helpful in the regeneration of vascular grafts needs to be established. In our work, we generated an AMPK activator (5-aminoimidazole-4-carboxamide ribonucleotide, AICAR) immobilized vascular graft. AICAR-modified vascular grafts were successfully generated by the co-electrospinning technique. In vitro results indicated that AICAR could upregulate energy supply in endothelial cells and reprogram macrophages (MΦ) to assume an anti-inflammatory phenotype. Furthermore, endothelial cells (ECs) co-cultured with AICAR achieved higher survival rates, better migration, and angiogenic capacity than the controls. Concurrently, a rabbit carotid artery transplantation model was used to investigate AICAR-modified vascular grafts at different time points. The results showed that AICAR-modified vascular grafts had higher patency rates (92.9% and 85.7% at 6 and 12 weeks, respectively) than those of the untreated group (11.1% and 0%). In conclusion, AICAR strengthened the cellular energy state and attenuated the adverse effects of inflammation. AICAR-modified vascular grafts achieved better vascular remodeling. This study provides a new perspective on promoting the regeneration of small-diameter vascular grafts.

Glaucoma is a chronic neuro-degenerative condition that is one of the world's leading causes of irreversible but preventable blindness. The blindness is generally caused by the lack of timely detection and treatment. Early screening is thus essential for early treatment to preserve vision and maintain life quality. Colour fundus photography and Optical Coherence Tomography (OCT) are the two most cost-effective tools for glaucoma screening. Both imaging modalities have prominent biomarkers to indicate glaucoma suspects, such as the vertical cup-to-disc ratio (vCDR) on fundus images and retinal nerve fiber layer (RNFL) thickness on OCT volume. In clinical practice, it is often recommended to take both of the screenings for a more accurate and reliable diagnosis. However, although numerous algorithms are proposed based on fundus images or OCT volumes for the automated glaucoma detection, there are few methods that leverage both of the modalities to achieve the target. To fulfil the research gap, we set up the Glaucoma grAding from Multi-Modality imAges (GAMMA) Challenge to encourage the development of fundus & OCT-based glaucoma grading. The primary task of the challenge is to grade glaucoma from both the 2D fundus images and 3D OCT scanning volumes. As part of GAMMA, we have publicly released a glaucoma annotated dataset with both 2D fundus colour photography and 3D OCT volumes, which is the first multi-modality dataset for machine learning based glaucoma grading. In addition, an evaluation framework is also established to evaluate the performance of the submitted methods. During the challenge, 1272 results were submitted, and finally, ten best performing teams were selected for the final stage. We analyse their results and summarize their methods in the paper. Since all the teams submitted their source code in the challenge, we conducted a detailed ablation study to verify the effectiveness of the particular modules proposed. Finally, we identify the proposed techniques and strategies that could be of practical value for the clinical diagnosis of glaucoma. As the first in-depth study of fundus & OCT multi-modality glaucoma grading, we believe the GAMMA Challenge will serve as an essential guideline and benchmark for future research.

Optical Coherence Tomography (OCT) is a novel and effective screening tool for ophthalmic examination. Since collecting OCT images is relatively more expensive than fundus photographs, existing methods use multi-modal learning to complement limited OCT data with additional context from fundus images. However, the multi-modal framework requires eye-paired datasets of both modalities, which is impractical for clinical use. To address this problem, we propose a novel fundus-enhanced disease-aware distillation model (FDDM), for retinal disease classification from OCT images. Our framework enhances the OCT model during training by utilizing unpaired fundus images and does not require the use of fundus images during testing, which greatly improves the practicality and efficiency of our method for clinical use. Specifically, we propose a novel class prototype matching to distill disease-related information from the fundus model to the OCT model and a novel class similarity alignment to enforce consistency between disease distribution of both modalities. Experimental results show that our proposed approach outperforms single-modal, multi-modal, and state-of-the-art distillation methods for retinal disease classification. Code is available at https://github.com/xmed-lab/FDDM.