Age-related cataract is a leading cause of visual impairment among older adults, significantly affecting quality of life and independence. Accurate severity grading is essential for determining optimal surgical timing and improving clinical outcomes; however, conventional slit-lamp examination relies heavily on subjective assessment, resulting in considerable inter-observer variability. In this study, we propose a progressive ordinal attention network (POA-Net) for automated four-level cataract classification. The proposed method is built on a convolutional neural network backbone and introduces a progressive ordinal cumulative attention mechanism to explicitly capture the continuous progression of disease severity. A spatial gating module is further incorporated to suppress background interference and enhance feature representation of the lens region. In addition, an ordinal prototype projection head with a topology-constrained learning strategy is designed to enforce structured feature distributions consistent with clinical knowledge. Experimental results demonstrate that POA-Net achieves an accuracy of 89.20

Dry eye, a common eye disease globally, poses significant challenges to clinical diagnosis and management due to its complex pathogenesis and high incidence rate. The development of artificial intelligence (AI) technology has provided new opportunities for the analysis and auxiliary diagnosis of dry eye imaging. This expert consensus focuses on the classification and annotation methods of dry eye imaging, in line with the application needs of AI technology. It summarizes the scope and tasks of research on the classification and annotation of dry eye imaging and provides detailed standards for the principles and methods of classification and annotation of major imaging modalities, including lipid layer of the tear film, tear meniscus height, tear film breakup time, corneal fluorescein staining, and meibomian gland images. It also clarifies the tools and processes for classification and annotation. The consensus proposes systematic quality control requirements, including annotation consistency assessment, multi-round review, and data cleaning methods. Finally, the consensus summarizes the current challenges and proposes targeted solutions. The launch of this consensus aims to provide high-quality data support for the development of AI in dry eye, enhance the application effects of AI in dry eye diagnosis, disease monitoring, and personalized treatment, and offer scientific references and technical support for clinical and research applications of AI in the field of dry eye.

Purpose:This study presents an artificial intelligence (AI)-based system for measuring eyeball rotation angles, which is a key symptom in assessing eye disease severity. The system aims to accurately segment the optic disc and macula, and compute the eyeball rotation angle based on these features. Methods:The system consists of three modules: optic disc segmentation, macular segmentation, and measurement. The optic disc segmentation module utilizes the Efficient-UNet3+ network to address sample imbalance and irregular edge detection of the optic disc. The macular segmentation module uses the Efficient-UNet based on Dual Attention network (DA-EUNet) to enhance macular recognition and boundary feature detection while suppressing irrelevant background interference. The measurement module calculates the eyeball rotation angle by locating the centers of the optic disc and macula and determining the angle between the line connecting these centers and the horizontal vector. Results:The proposed method demonstrated high accuracy, with a correlation coefficient of 0.94 compared to expert measurements. Statistical analysis revealed no significant difference between the AI-based measurements and expert assessments (P = 0.26). Conclusions:This system achieves high accuracy and reliability in clinical diagnostics. The segmentation techniques used significantly improve feature recognition and segmentation performance, enabling accurate measurements of eyeball rotation. Translational Relevance:This AI-based system bridges the gap between basic research in medical image processing and clinical care. It provides an automated and reliable tool for ophthalmologists to assess eyeball rotation, which is crucial for diagnosing eye diseases. Eyeball rotation can occur in many eye diseases or systemic diseases, and measuring the eyeball rotation angle has been a challenging issue in clinical practice. By automating this process, the system reduces the clinicians' workload and enhances diagnostic consistency.

Background:Eyelid morphological parameters are crucial for quantitatively assessing eyelid morphology and diagnosing related disorders. However, achieving precise and automated measurement of these parameters remains a challenge. This study aims to develop an automated segmentation model for ocular surface images to accurately segment key anatomical structures and compute critical eyelid metrics. Methods:We propose STB-Net, a novel segmentation model tailored for ocular surface imagery. The baseline model, TB-Net, enhances the TransUNet architecture by integrating a Bottom-up Local Attention Modulation (BLAM) module into its decoder, effectively encoding fine-grained features from shallow layers into high-level semantics. The final STB-Net framework integrates TB-Net with an SRSNetwork. In this setup, a TB-Net is first employed as a reconstruction model to learn complex semantic information. Subsequently, its encoder serves as a dynamic convolution module to generate adaptive parameters for a second, segmentation-oriented TB-Net, thereby boosting segmentation performance through augmented reconstruction-task training. The model automatically computes left, central, and right palpebral fissure heights, palpebral fissure width, and area based on the segmentation results. Results:Experimental evaluation on a local dataset demonstrated the model's high efficacy. For palpebral fissure segmentation, the model achieved a Dice score of 0.9875, Global Accuracy (GA) of 0.9955, and Intersection-over-Union (IoU) of 0.9767. Corneal segmentation performance was equally strong, attaining a Dice of 0.9891, GA of 0.9978, and IoU of 0.9790. Conclusions:The proposed STB-Net model provides a robust and effective solution for the automated segmentation of ocular surface structures and the precise quantification of eyelid morphological parameters. It holds significant promise for enhancing the objectivity and efficiency of clinical diagnoses of eyelid disorders.

ObjectiveEyelid curvature analysis serves as a key morphological indicator in the diagnosis of ophthalmic diseases and postoperative evaluation. This study aims to develop an automated and reproducible image processing method to accurately extract eyelid margin curves from anterior segment images and perform quantitative curvature analysis.MethodsA dual-branch U-Net architecture is proposed, utilizing a shared encoder and task-specific decoders to simultaneously segment the palpebral fissure and corneal regions. Based on the segmentation results, eyelid margin curves were extracted and fitted with second-order polynomials to calculate curvature values.ResultsA total of 130 anterior segment images were collected. In segmentation tasks, the proposed AtDU-Net model achieved an IoU of 0.979 and a Dice coefficient of 0.989. The automatically measured eyelid curvatures showed high consistency with manual annotations, with correlation coefficients of 0.9032 for the upper eyelid and 0.9154 for the lower eyelid. Bland-Altman analysis indicated that over 92% of the samples fell within the limits of agreement, validating the consistency and reliability of the measurements.ConclusionThe proposed method demonstrates superior performance in terms of accuracy, robustness, and consistency with manual measurements. It shows strong potential for clinical applications, providing reliable technical support for eyelid morphological analysis and surgical planning.

Vessel morphology and tube diameter changes in the retina are closely related to diseases such as diabetic retinopathy and glaucoma. Traditional methods of vessel segmentation and tube diameter measurement suffer from insufficient accuracy and low automation. To address these limitations, this paper proposes a retinal arteriovenous segmentation network model based on dual branch multi-scale feature adaptive extraction. The model is designed to tackle the challenge of complex and similar arteriovenous vessel structures by incorporating a multi-scale feature extraction and fusion module, which enhances the capture of contextual vessel information and microvascular features. To ensure the model learns rich semantic information, a dual-branch encoder is implemented to extract multi-scale and residual features. The multi-feature local global attention fusion module then performs feature fusion, integrating these features effectively. In the down-sampling module, an adaptive pooling mechanism is introduced to balance detailed and global information of the arterial vessels. This adaptive learning of hyperparameters helps to mitigate the confusion between arterial and venous vessels. Furthermore, a multi-layer semantic supervision module is incorporated to enhance the model's learning ability and improve segmentation accuracy for arteriovenous vessel features. The method's effectiveness is validated on the DRIVEAV, HRF-AV, and LES-AV datasets, achieving accuracy rates of 94.19%, 99.26%, and 93.10%, respectively. Additionally, an equivalent diameter measurement method based on optic disc and macular localization is proposed. This method accurately calculates the equivalent diameter of arterioles and veins, providing quantitative indices for the auxiliary diagnosis of related diseases and enhancing clinical efficiency.

Glaucoma is a leading cause of irreversible blindness, and accurate cup-to-disc ratio (CDR) measurement is essential for early detection. This study presents an enhanced deep learning–based system for automated CDR estimation and glaucoma screening. We propose an end-to-end framework consisting of three modules: (1) optic cup and disc segmentation using an enhanced dual encoder–decoder network (E-DCoAtUNet), (2) a conditional random field (CRF) post-processing module for boundary refinement, and (3) a measurement module for vertical CDR calculation and glaucoma classification. The model was trained and evaluated on the Drishti-GS dataset and validated on the REFUGE dataset to assess generalizability. The system achieved Dice scores of 97.6

Cataracts are a leading cause of blindness worldwide, making accurate diagnosis and effective surgical planning critical. However, grading the severity of the lens nucleus is challenging because deep learning (DL) models pretrained using ImageNet perform poorly when applied directly to medical data due to the limited availability of labeled medical images and high interclass similarity. Self-supervised pretraining offers a solution by circumventing the need for cost-intensive data annotations and bridging domain disparities. In this study, to address the challenges of intelligent grading, we proposed a hybrid model called nuclear cataract mask encoder network (NCME-Net), which utilizes self-supervised pretraining for the four-class analysis of nuclear cataract severity. A total of 792 images of nuclear cataracts were categorized into the training set (533 images), the validation set (139 images), and the test set (100 images). NCME-Net achieved a diagnostic accuracy of 91.0% on the test set, a 5.0% improvement over the best-performing DL model (ResNet50). Experimental results demonstrate NCME-Net’s ability to distinguish between cataract severities, particularly in scenarios with limited samples, making it a valuable tool for intelligently diagnosing cataracts. In addition, the effect of different self-supervised tasks on the model’s ability to capture the intrinsic structure of the data was studied. Findings indicate that image restoration tasks significantly enhance semantic information extraction.

Pterygium and subconjunctival hemorrhage are two common types of ocular surface diseases that can cause distress and anxiety in patients. In this study, 2855 ocular surface images were collected in four categories: normal ocular surface, subconjunctival hemorrhage, pterygium to be observed, and pterygium requiring surgery. We propose a diagnostic classification model for ocular surface diseases, dual-branch network reinforced by PFM block (DBPF-Net), which adopts the conformer model with two-branch architectural properties as the backbone of a four-way classification model for ocular surface diseases. In addition, we propose a block composed of a patch merging layer and a FReLU layer (PFM block) for extracting spatial structure features to further strengthen the feature extraction capability of the model. In practice, only the ocular surface images need to be input into the model to discriminate automatically between the disease categories. We also trained the VGG16, ResNet50, EfficientNetB7, and Conformer models, and evaluated and analyzed the results of all models on the test set. The main evaluation indicators were sensitivity, specificity, F1-score, area under the receiver operating characteristics curve (AUC), kappa coefficient, and accuracy. The accuracy and kappa coefficient of the proposed diagnostic model in several experiments were averaged at 0.9789 and 0.9681, respectively. The sensitivity, specificity, F1-score, and AUC were, respectively, 0.9723, 0.9836, 0.9688, and 0.9869 for diagnosing pterygium to be observed, and, respectively, 0.9210, 0.9905, 0.9292, and 0.9776 for diagnosing pterygium requiring surgery. The proposed method has high clinical reference value for recognizing these four types of ocular surface images.

Age-related macular degeneration (AMD) ranks third among the most common causes of blindness.As the most conventional and direct method for identifying AMD, color fundus photography has become prominent owing to its consistency, ease of use, and good quality in extensive clinical practice.In this study, a convolutional neural network (CSPDarknet53) was combined with a transformer to construct a new hybrid model, HCSP-Net.This hybrid model was employed to tri-classify color fundus photography into the normal macula (NM), dry macular degeneration (DMD), and wet macular degeneration (WMD) based on clinical classification manifestations, thus identifying and resolving AMD as early as possible with color fundus photography.To further enhance the performance of this model, grouped convolution was introduced in this study without significantly increasing the number of parameters.HCSP-Net was validated using an independent test set.The average precision of HCSP-Net in the diagnosis of AMD was 99.2%, the recall rate was 98.2%, the F1-Score was 98.7%, the PPV (positive predictive value) was 99.2%, and the NPV (negative predictive value) was 99.6%.Moreover, a knowledge distillation approach was also adopted to develop a lightweight student network (SCSP-Net).The experimental results revealed a noteworthy enhancement in the accuracy of SCSP-Net, rising from 94% to 97%, while remarkably reducing the parameter count to a quarter of HCSP-Net.This attribute positions SCSP-Net as a highly suitable candidate for the deployment of resource-constrained devices, which may provide ophthalmologists with an efficient tool for diagnosing AMD.

BackgroundFundus vessel segmentation is vital for diagnosing ophthalmic diseases like central serous chorioretinopathy (CSC), diabetic retinopathy, and glaucoma. Accurate segmentation provides crucial vessel morphology details, aiding the early detection and intervention of ophthalmic diseases. However, current algorithms struggle with fine vessel segmentation and maintaining sensitivity in complex regions. Challenges also stem from imaging variability and poor generalization across multimodal datasets, highlighting the need for more advanced algorithms in clinical practice.MethodsThis paper aims to explore a new vessel segmentation method to alleviate the above problems. We propose a fundus vessel segmentation model based on a combination of double skip connections, deep supervision, and TransUNet, namely DS2TUNet. Initially, the original fundus images are improved through grayscale conversion, normalization, histogram equalization, gamma correction, and other preprocessing techniques. Subsequently, by utilizing the U-Net architecture, the preprocessed fundus images are segmented to obtain the final vessel information. Specifically, the encoder firstly incorporates the ResNetV1 downsampling, dilated convolution downsampling, and Transformer to capture both local and global features, which upgrades its vessel feature extraction ability. Then, the decoder introduces the double skip connections to facilitate upsampling and refine segmentation outcomes. Finally, the deep supervision module introduces multiple upsampling vessel features from the decoder into the loss function, so that the model can learn vessel feature representations more effectively and alleviate gradient vanishing during the training phase.ResultsExtensive experiments on publicly available multimodal fundus datasets such as DRIVE, CHASE_DB1, and ROSE-1 demonstrate that the DS2TUNet model attains F1-scores of 0.8195, 0.8362, and 0.8425, with Accuracy of 0.9664, 0.9741, and 0.9557, Sensitivity of 0.8071, 0.8101, and 0.8586, and Specificity of 0.9823, 0.9869, and 0.9713, respectively. Additionally, the model also exhibits excellent test performance on the clinical fundus dataset CSC, with F1-score of 0.7757, Accuracy of 0.9688, Sensitivity of 0.8141, and Specificity of 0.9801 based on the weight trained on the CHASE_DB1 dataset. These results comprehensively validate that the proposed method obtains good performance in fundus vessel segmentation, thereby aiding clinicians in the further diagnosis and treatment of fundus diseases in terms of effectiveness and feasibility.

Background and objectiveThe prevalence of myopia and high myopia is increasing globally, underscoring the growing importance of diagnosing high myopia-related pathologies. While existing image segmentation models, such as U-Net, UNet++, ResU-Net, and TransUNet, have achieved significant success in medical image segmentation, they still face challenges when dealing with ultra-widefield (UWF) fundus images. This study introduces a novel automatic segmentation algorithm for the optic disc and peripapillary atrophy (PPA) based on UWF fundus images, aimed at assisting ophthalmologists in more accurately diagnosing high myopia-related diseases.MethodsIn this study, we developed a segmentation model leveraging a Transformer-based network structure, complemented by atrous convolution and selective boundary aggregation modules, to elevate the accuracy of segmenting the optic disc and PPA in UWF photography. The atrous convolution module adeptly manages multi-scale features, catering to the variances in target sizes and expanding the deep network's receptive field. Concurrently, the incorporation of the selective boundary aggregation module in the skip connections of the model significantly improves the differentiation of boundary information between segmentation targets. Moreover, the comparison of our proposed algorithm with classical segmentation models like U-Net, UNet++, ResU-Net, and TransUNet highlights its considerable advantages in processing UWF photographs.ResultsThe experimental results show that, compared to the other four models, our algorithm demonstrates substantial improvements in segmenting the optic disc and PPA in UWF photographs. In PPA segmentation, our algorithm improves by 0.8% in Dice, 1.8% in sensitivity, and 1.3% in intersection over union (IOU). In optic disc segmentation, our algorithm improves by 0.3% in Dice, 0.6% in precision, and 0.4% in IOU.ConclusionOur proposed method improves the segmentation accuracy of PPA and optic disks based on UWF photographs, which is valuable for diagnosing high myopia-related diseases in ophthalmology clinics.

Accurately quantifying the height of central serous chorioretinopathy (CSCR) lesion is of great significance for assisting ophthalmologists in diagnosing CSCR and evaluating treatment efficacy. The manual measurement results dominated by single optical coherence tomography (OCT) B-scan image in clinical practice face the dilemma of weak reference, poor reproducibility, and experience dependence. In this context, this paper constructs two schemes: Scheme Ⅰ draws on the idea of ensemble learning, namely, integrating multiple models for locating starting key point in the height direction of lesion in the inference stage, which appropriately improves the performance of a single model. Scheme Ⅱ designs an adaptive gradient threshold (AGT) technique, followed by the construction of cascading strategy, which involves preliminary location of starting key point through deep learning, and then employs AGT for precise adjustment. This strategy not only achieves effective location for starting key point, but also significantly reduces the large appetite of deep learning model for training samples. Subsequently, AGT continues to play a crucial role in locating the terminal key point in the height direction of lesion, further demonstrating its feasibility and effectiveness. Quantitative and qualitative key point location experiments in the height direction of lesion on 1152 samples, as well as the final height measurement display, consistently conveys the superiority of the constructed schemes, especially the cascading strategy, expanding another potential tool for the comprehensive analysis of CSCR.

Background: Virtual reality (VR), widely used in the medical field, may affect future medical training and treatment. Therefore, this study examined VR’s potential uses and research directions in medicine. Methods: Citation data were downloaded from the Web of Science Core Collection database (WoSCC) to evaluate VR in medicine in articles published between 1 January 2012 and 31 December 2023. These data were analyzed using CiteSpace 6.2. R2 software. Present limitations and future opportunities were summarized based on the data. Results: A total of 2143 related publications from 86 countries and regions were analyzed. The country with the highest number of publications is the USA, with 461 articles. The University of London has the most publications among institutions, with 43 articles. The burst keywords represent the research frontier from 2020 to 2023, such as “task analysis”, “deep learning”, and “machine learning”. Conclusion: The number of publications on VR applications in the medical field has been steadily increasing year by year. The USA is the leading country in this area, while the University of London stands out as the most published, and most influential institution. Currently, there is a strong focus on integrating VR and AI to address complex issues such as medical education and training, rehabilitation, and surgical navigation. Looking ahead, the future trend involves integrating VR, augmented reality (AR), and mixed reality (MR) with the Internet of Things (IoT), wireless sensor networks (WSNs), big data analysis (BDA), and cloud computing (CC) technologies to develop intelligent healthcare systems within hospitals or medical centers.

This paper analyzes the current status, technological developments, academic exchange platforms, and future challenges and solutions in the field of intelligent ophthalmology (IO) in China. In terms of technology, significant progress has been made in various areas, including diabetic retinopathy, fundus image analysis, quality assessment of medical artificial intelligence products, clinical research methods, technical evaluation, and industry standards. Researchers continually enhance the safety and standardization of IO technology by formulating a series of clinical application guidelines and standards. The establishment of domestic and international academic exchange platforms provides extensive collaboration opportunities for professionals in various fields, and various academic journals serve as publication platforms for IO research. However, challenges such as technological innovation, data privacy and security, lagging regulations, and talent shortages still pose obstacles to future development. To address these issues, future efforts should focus on strengthening technological research and development, regulatory framework construction, talent cultivation, and increasing patient awareness and acceptance of new technologies. By comprehensively addressing these challenges, IO in China is poised to further lead the industry's development on a global scale, bringing more innovation and convenience to the field of ophthalmic healthcare.

Medical artificial intelligence (AI) and big data technology have rapidly advanced in recent years, and they are now routinely used for image-based diagnosis. China has a massive amount of medical data. However, a uniform criteria for medical data quality have yet to be established. Therefore, this review aimed to develop a standardized and detailed set of quality criteria for medical data collection, storage, annotation, and management related to medical AI. This would greatly improve the process of medical data resource sharing and the use of AI in clinical medicine.

PurposeTo propose a five-category model for the automatic detection of myopic macular lesions to help grassroots medical institutions conduct preliminary screening of myopic macular lesions from limited number of color fundus images.MethodsFirst, 1,750 fundus images of non-myopic retinal lesions and four categories of pathological myopic maculopathy were collected, graded, and labeled. Subsequently, three five-classification models based on Vision Outlooker for Visual Recognition (VOLO), EfficientNetV2, and ResNet50 for detecting myopic maculopathy were trained with data-augmented images, and the diagnostic results of the different trained models were compared and analyzed. The main evaluation metrics were sensitivity, specificity, negative predictive value (NPV), positive predictive value (PPV), area under the curve (AUC), kappa and accuracy, and receiver operating characteristic curve (ROC).ResultsThe diagnostic accuracy of the VOLO-D2 model was 96.60% with a kappa value of 95.60%. All indicators used for the diagnosis of myopia-free macular degeneration were 100%. The sensitivity, NPV, specificity, and PPV for diagnosis of leopard fundus were 96.43, 98.33, 100, and 100%, respectively. The sensitivity, specificity, PPV, and NPV for the diagnosis of diffuse chorioretinal atrophy were 96.88, 98.59, 93.94, and 99.29%, respectively. The sensitivity, specificity, PPV, and NPV for the diagnosis of patchy chorioretinal atrophy were 92.31, 99.26, 97.30, and 97.81%, respectively. The sensitivity, specificity, PPV, and NPV for the diagnosis of macular atrophy were 100, 98.10, 84.21, and 100%, respectively.ConclusionThe VOLO-D2 model accurately identified myopia-free macular lesions and four pathological myopia-related macular lesions with high sensitivity and specificity. It can be used in screening pathological myopic macular lesions and can help ophthalmologists and primary medical institution providers complete the initial screening diagnosis of patients.

The lesion boundary of central serous chorioretinopathy (CSCR) is the guarantee to guide the ophthalmologist to accurately arrange the laser spots, so as to enable this ophthalmopathy to be treated precisely. Currently, the accuracy and rapidity of manually locating CSCR lesion boundary in clinic based on single-modal fundus image are limited by imaging quality and ophthalmologist experience, which is also accompanied by poor repeatability, weak reliability and low efficiency. Consequently, a multi-modal fundus image-based lesion boundary auxiliary location method is developed. Firstly, the initial location module (ILM) is employed to achieve the preliminary location of key boundary points of CSCR lesion area on the optical coherence tomography (OCT) B-scan image, then followed by the joint location module (JLM) created based on reinforcement learning for further enhancing the location accuracy. Secondly, the scanning line detection module (SLDM) is constructed to realize the location of lesion scanning line on the scanning laser ophthalmoscope (SLO) image, so as to facilitate the cross-modal mapping of key boundary points. Finally, a simple yet effective lesion boundary location module (LBLM) is designed to assist the automatic cross-modal mapping of key boundary points and enable the final location of lesion boundary. Extensive experiments show that each module can perform well on its corresponding sub task, such as JLM, which makes the correction rate (CR) of ILM increase to 92.11%, comprehensively indicating the effectiveness and feasibility of this method in providing effective lesion boundary guidance for assisting ophthalmologists to precisely arrange the laser spots, and also opening a new research idea for the automatic location of lesion boundary of other fundus diseases.

Accurately and rapidly measuring the diameter of central serous chorioretinopathy (CSCR) lesion area is the key to judge the severity of CSCR and evaluate the efficacy of the corresponding treatments. Currently, the manual measurement scheme based on a single or a small number of optical coherence tomography (OCT) B-scan images encounters the dilemma of incredibility. Although manually measuring the diameters of all OCT B-scan images of a single patient can alleviate the previous issue, the situation of inefficiency will thus arise. Additionally, manual operation is subject to subjective factors of ophthalmologists, resulting in unrepeatable measurement results. Therefore, an automatic image processing method (i.e., a joint framework) based on artificial intelligence (AI) is innovatively proposed for locating the key boundary points of CSCR lesion area to assist the diameter measurement. Firstly, the initial location module (ILM) benefiting from multitask learning is properly adjusted and tentatively achieves the preliminary location of key boundary points. Secondly, the location task is formulated as a Markov decision process, aiming at further improving the location accuracy by utilizing the single agent reinforcement learning module (SARLM). Finally, the joint framework based on the ILM and SARLM is skillfully established, in which ILM provides an initial starting point for SARLM to narrow the active region of agent, and SARLM makes up for the defect of low generalization of ILM by virtue of the independent exploration ability of agent. Experiments reveal the AI-based method which joins the multitask learning, and single agent reinforcement learning paradigms enable agents to work in local region, alleviating the time-consuming problem of SARLM, performing location task in a global scope, and improving the location accuracy of ILM, thus reflecting its effectiveness and clinical application value in the task of rapidly and accurately measuring the diameter of CSCR lesions.