Hasil untuk "Ophthalmology"

A. Esteva, Katherine Chou, Serena Yeung et al.

A decade of unprecedented progress in artificial intelligence (AI) has demonstrated the potential for many fields—including medicine—to benefit from the insights that AI techniques can extract from data. Here we survey recent progress in the development of modern computer vision techniques—powered by deep learning—for medical applications, focusing on medical imaging, medical video, and clinical deployment. We start by briefly summarizing a decade of progress in convolutional neural networks, including the vision tasks they enable, in the context of healthcare. Next, we discuss several example medical imaging applications that stand to benefit—including cardiology, pathology, dermatology, ophthalmology–and propose new avenues for continued work. We then expand into general medical video, highlighting ways in which clinical workflows can integrate computer vision to enhance care. Finally, we discuss the challenges and hurdles required for real-world clinical deployment of these technologies.

Carol L. Schroeder, Kristopher M Schroeder

Cocaine is a naturally occurring compound indigenous to the Andes Mountains, West Indies, and Java. It was the first anesthetic to be discovered and is the only naturally occurring local anesthetic; all others are synthetically derived. Cocaine was introduced into Europe in the 1800s following its isolation from coca beans. Sigmund Freud, the noted Austrian psychoanalyst, used cocaine on his patients and became addicted through self-experimentation. In the latter half of the 1800s, interest in the drug became widespread, and many of cocaine's pharmacologic actions and adverse effects were elucidated during this time. In the 1880s, Koller introduced cocaine to the field of ophthalmology, and Hall introduced it to dentistry

P. Gain, R. Jullienne, Zhiguo He et al.

M. Marmor, U. Kellner, T. Lai et al.

C. Wilkinson, Frederick L. Ferris, R. Klein et al.

C. Mangione, P. P. Lee, P. Gutierrez et al.

W. Duke-Elder

Weitong Yao, Yujun Li, Huize Sun et al.

IntroductionThe beta-coronavirus SARS-CoV-2 has been revealed to infect mammals and other species, which potentially promotes the virus adaptation to broader species and the emergence of new variants. The host range of different SARS-CoV-2 variants are mainly determined by the affinity of the receptor-binding domain (RBD) of the spike protein to the host receptor angiotensin-converting enzyme 2 (ACE2). Thus, this study aims to elucidate the detailed mechanisms of such dynamic adaptation of indicated SARS-CoV-2 variants.MethodsIn this study, flow cytometry and surface plasmon resonance (SPR) assays were used to assess the binding affinity between RBDs and avian ACE2. Then, infection assays with MLV-based SARS-CoV-2 spike pseudovirus or authentic viruses were performed to verify the avian ACE2 mediated viral entry. Finally, mutagenesis studies were conducted to identify key amino acids of avian ACE2 orthologs and RBDs.ResultsOur previous findings revealed that wild-type SARS-CoV-2 RBD does not bind chicken ACE2. Here, we found that ACE2 orthologs from chicken and mallard were capable to support binding to RBDs of the Alpha, Beta, and Gamma variants, which further enabled the viral entry. On the contrary, the RBD of BA.1 failed to bind avian ACE2. Whereas, a triple-residue reversal mutant (S446G, S496G, H505Y) restored ACE2 binding and enabled efficient viral entry. Additionally, several key residues within RBD were characterized as the determinant of its affinity to avian ACE2.DiscussionOur findings reveal that higher mutation rates in emerging variants might lead to future cross-species receptor usage or even spillover. Understanding such cross-species transmission mechanisms provides new insights to the virological features and potential host range of emerging SARS-CoV-2 variants.

W. Hu, C. Garrison, R. Prasad et al.

Tryptophan (Trp), an essential amino acid obtained through dietary sources, plays a crucial role in various physiological processes. The metabolism of Trp branches into three principal pathways: the serotonin pathway, the kynurenine pathway, and the indole pathway. The kynurenine and serotonin pathways are host pathways while the indole pathway is solely the result of bacterial metabolism. Trp metabolites extend their influence beyond protein biosynthesis to affect a spectrum of pathophysiological mechanisms including, but not limited to, neuronal function, immune modulation, inflammatory responses, oxidative stress regulation, and maintenance of intestinal health. This review focuses on indole derivatives and their impact on vascular health. Trp-containing dipeptides are highlighted as a targeted nutraceutical approach to modulate Trp metabolism, enhance beneficial metabolite production, and mitigate risk factors for vascular diseases. The importance of optimizing Trp intake and dietary strategies to harness the benefits of Trp-derived metabolites for vascular health is underscored, bringing to light the need for further research to refine these therapeutic approaches.

Ludovic Arnould, Salim Khazem, Hugues Ali Mehenni

Visual Language Models (VLMs) are now sufficiently advanced to support a broad range of applications, including answering complex visual questions, and are increasingly expected to interact with images in varied ways. To evaluate them, current benchmarks often focus on specific domains (e.g., reading charts), constructing datasets of annotated real images paired with pre-defined Multiple Choice Questions (MCQs) to report aggregate accuracy scores. However, such benchmarks entail high annotation costs, risk information leakage, and do not clarify whether failures stem from limitations in visual perception, reasoning, or general knowledge. We propose a new evaluation methodology, inspired by ophthalmologic diagnostics, leveraging procedural generation of synthetic images to obtain control over visual attributes and precisely reveal perception failures in VLMs. Specifically, we build collections of images with gradually more challenging variations in the content of interest (e.g., number of objects in a counting task) while holding other visual parameters constant. This diagnostic allows systematic stress testing and fine-grained failure analysis, shifting the focus from coarse benchmarking toward targeted and interpretable assessment of VLM capabilities. Our code is available at https://github.com/byoeval/BYO-EVAL.

Jia Zhou, Chang-Xing Ma

Clinical trials involving paired organs often yield a mixture of unilateral and bilateral data, where each subject may contribute either one or two responses under certain circumstances. While unilateral responses from different individuals can be treated as independent, bilateral responses from the same individual are likely correlated. Various statistical methods have been developed to account for this intra-subject correlation in the bilateral data, and in practice it is crucial to select an appropriate model for accurate inference. Tang et. al. (2012) discussed model selection issues using a variety of goodness-of-fit test statistics for correlated bilateral data for two groups, and Liu and Ma (2020) extended these methods to settings with $g\ge2$ groups. In this work, we investigate the goodness-of-fit statistics for the combined unilateral and bilateral data under different statistical models that address the intra-subject correlation, including the Clayton copula model, in addition to those considered in prior studies. Simulation results indicate that the performance of the goodness-of-fit tests is model-dependent, especially when the sample size is small and/or the intra-subject correlation is high, which is consistent with the findings by Liu and Ma (2020) for purely bilateral data. Applications to real data from otolaryngologic and ophthalmologic studies are included.

Alif Munim, Jun Ma, Omar Ibrahim et al.

Large language models (LLMs) have rapidly advanced in clinical decision-making, yet the deployment of proprietary systems is hindered by privacy concerns and reliance on cloud-based infrastructure. Open-source alternatives allow local inference but often require large model sizes that limit their use in resource-constrained clinical settings. Here, we benchmark two on-device LLMs, gpt-oss-20b and gpt-oss-120b, across three representative clinical tasks: general disease diagnosis, specialty-specific (ophthalmology) diagnosis and management, and simulation of human expert grading and evaluation. We compare their performance with state-of-the-art proprietary models (GPT-5 and o4-mini) and a leading open-source model (DeepSeek-R1), and we further evaluate the adaptability of on-device systems by fine-tuning gpt-oss-20b on general diagnostic data. Across tasks, gpt-oss models achieve performance comparable to or exceeding DeepSeek-R1 and o4-mini despite being substantially smaller. In addition, fine-tuning remarkably improves the diagnostic accuracy of gpt-oss-20b, enabling it to approach the performance of GPT-5. These findings highlight the potential of on-device LLMs to deliver accurate, adaptable, and privacy-preserving clinical decision support, offering a practical pathway for broader integration of LLMs into routine clinical practice.

Jiancheng Ye, Sophie Bronstein, Jiarui Hai et al.

DeepSeek-R1 is a cutting-edge open-source large language model (LLM) developed by DeepSeek, showcasing advanced reasoning capabilities through a hybrid architecture that integrates mixture of experts (MoE), chain of thought (CoT) reasoning, and reinforcement learning. Released under the permissive MIT license, DeepSeek-R1 offers a transparent and cost-effective alternative to proprietary models like GPT-4o and Claude-3 Opus; it excels in structured problem-solving domains such as mathematics, healthcare diagnostics, code generation, and pharmaceutical research. The model demonstrates competitive performance on benchmarks like the United States Medical Licensing Examination (USMLE) and American Invitational Mathematics Examination (AIME), with strong results in pediatric and ophthalmologic clinical decision support tasks. Its architecture enables efficient inference while preserving reasoning depth, making it suitable for deployment in resource-constrained settings. However, DeepSeek-R1 also exhibits increased vulnerability to bias, misinformation, adversarial manipulation, and safety failures - especially in multilingual and ethically sensitive contexts. This survey highlights the model's strengths, including interpretability, scalability, and adaptability, alongside its limitations in general language fluency and safety alignment. Future research priorities include improving bias mitigation, natural language comprehension, domain-specific validation, and regulatory compliance. Overall, DeepSeek-R1 represents a major advance in open, scalable AI, underscoring the need for collaborative governance to ensure responsible and equitable deployment.

Xi Luo, Shixin Xu, Ying Xie et al.

Accurate medical image analysis can greatly assist clinical diagnosis, but its effectiveness relies on high-quality expert annotations Obtaining pixel-level labels for medical images, particularly fundus images, remains costly and time-consuming. Meanwhile, despite the success of deep learning in medical imaging, the lack of interpretability limits its clinical adoption. To address these challenges, we propose TWLR, a two-stage framework for interpretable diabetic retinopathy (DR) assessment. In the first stage, a vision-language model integrates domain-specific ophthalmological knowledge into text embeddings to jointly perform DR grading and lesion classification, effectively linking semantic medical concepts with visual features. The second stage introduces an iterative severity regression framework based on weakly-supervised semantic segmentation. Lesion saliency maps generated through iterative refinement direct a progressive inpainting mechanism that systematically eliminates pathological features, effectively downgrading disease severity toward healthier fundus appearances. Critically, this severity regression approach achieves dual benefits: accurate lesion localization without pixel-level supervision and providing an interpretable visualization of disease-to-healthy transformations. Experimental results on the FGADR, DDR, and a private dataset demonstrate that TWLR achieves competitive performance in both DR classification and lesion segmentation, offering a more explainable and annotation-efficient solution for automated retinal image analysis.

Tariq M Khan, Toufique Ahmed Soomro, Imran Razzak

Retinal imaging has emerged as a powerful, non-invasive modality for detecting and quantifying biomarkers of systemic diseases-ranging from diabetes and hypertension to Alzheimer's disease and cardiovascular disorders but current insights remain dispersed across platforms and specialties. Recent technological advances in optical coherence tomography (OCT/OCTA) and adaptive optics (AO) now deliver ultra-high-resolution scans (down to 5 μm ) with superior contrast and spatial integration, allowing early identification of microvascular abnormalities and neurodegenerative changes. At the same time, AI-driven and machine learning (ML) algorithms have revolutionized the analysis of large-scale retinal datasets, increasing sensitivity and specificity; for example, deep learning models achieve > 90 \% sensitivity for diabetic retinopathy and AUC = 0.89 for the prediction of cardiovascular risk from fundus photographs. The proliferation of mobile health technologies and telemedicine platforms further extends access, reduces costs, and facilitates community-based screening and longitudinal monitoring. Despite these breakthroughs, translation into routine practice is hindered by heterogeneous imaging protocols, limited external validation of AI models, and integration challenges within clinical workflows. In this review, we systematically synthesize the latest OCT/OCT and AO developments, AI/ML approaches, and mHealth/Tele-ophthalmology initiatives and quantify their diagnostic performance across disease domains. Finally, we propose a roadmap for multicenter protocol standardization, prospective validation trials, and seamless incorporation of retinal screening into primary and specialty care pathways-paving the way for precision prevention, early intervention, and ongoing treatment of life-threatening systemic diseases.

Hamidreza Amiri, Samira Peiravi, Seyedeh sara rezazadeh shojaee et al.

Abstract Background Nowadays, Artificial intelligence (AI) is one of the most popular topics that can be integrated into healthcare activities. Currently, AI is used in specialized fields such as radiology, pathology, and ophthalmology. Despite the advantages of AI, the fear of human labor being replaced by this technology makes some students reluctant to choose specific fields. This meta-analysis aims to investigate the knowledge and attitude of medical, dental, and nursing students and experts in this field about AI and its application. Method This study was designed based on PRISMA guidelines. PubMed, Scopus, and Google Scholar databases were searched with relevant keywords. After study selection according to inclusion criteria, data of knowledge and attitude were extracted for meta-analysis. Result Twenty-two studies included 8491 participants were included in this meta-analysis. The pooled analysis revealed a proportion of 0.44 (95%CI = [0.34, 0.54], P < 0.01, I2 = 98.95%) for knowledge. Moreover, the proportion of attitude was 0.65 (95%CI = [0.55, 0.75], P < 0.01, I2 = 99.47%). The studies did not show any publication bias with a symmetrical funnel plot. Conclusion Average levels of knowledge indicate the necessity of including relevant educational programs in the student’s academic curriculum. The positive attitude of students promises the acceptance of AI technology. However, dealing with ethics education in AI and the aspects of human-AI cooperation are discussed. Future longitudinal studies could follow students to provide more data to guide how AI can be incorporated into education.

Victoria Y. Wang, Men-Tzung Lo, Ta-Ching Chen et al.

Abstract As an alternative to assessments performed by human experts, artificial intelligence (AI) is currently being used for screening fundus images and monitoring diabetic retinopathy (DR). Although AI models can provide quasi-clinician diagnoses, they rarely offer new insights to assist clinicians in predicting disease prognosis and treatment response. Using longitudinal retinal imaging data, we developed and validated a predictive model for DR progression: AI-driven Diabetic Retinopathy Progression Prediction Algorithm (ADRPPA). In this retrospective study, we analyzed paired retinal fundus images of the same eye captured at ≥ 1-year intervals. The analysis was performed using the EyePACS dataset. By analyzing 12,768 images from 6384 eyes (2 images/eye, taken 733 ± 353 days apart), each annotated with DR severity grades, we trained the neural network ResNeXt to automatically determine DR severity. EyePACS data corresponding to 5108 (80%), 639 (10%), and 637 (10%) eyes were used for model training, validation, and testing, respectively. We further used an independent e-ophtha dataset comprising 148 images annotated with microaneurysms, 118 (75%) and 30 (25%) of which were used for training and validation, respectively. This dataset was used to train the neural network Mask Region–based Convolutional Neural Network (Mask-RCNN) for quantifying microaneurysms. The DR and microaneurysm scores from the first nonreferable DR (NRDR) image of each eye were used to predict progression to referable DR (RDR) in the second image. The area under the receiver operating characteristic curve values indicating our model’s performance in diagnosing RDR were 0.963, 0.970, 0.968, and 0.971 for the trained ResNeXt models with input image resolutions of 256 × 256, 512 × 512, 768 × 768, and 1024 × 1024 pixels, respectively. In the validation of the Mask-RCNN model trained on the e-ophtha dataset resized to 1600 pixels in height, the recall, precision, and F1-score values for detecting individual microaneurysms were 0.786, 0.615, and 0.690, respectively. The best model combination for predicting NRDR-to-RDR progression included the 768-pixel ResNeXt and 1600-pixel Mask-RCNN models; this combination achieved recall, precision, and F1-scores of 0.338 (95% confidence interval [CI]: 0.228–0.451), 0.561 (95% CI: 0.405–0.714), and 0.422 (95% CI: 0.299–0.532), respectively. Thus, deep learning models can be trained on longitudinal retinal imaging data to predict NRDR-to-RDR progression. Furthermore, DR and microaneurysm scores generated from low- and high-resolution fundus images, respectively, can help identify patients at a high risk of NRDR, facilitating timely treatment.

Yepeng Liu, Baosheng Yu, Tian Chen et al.

Retinal image registration plays an important role in the ophthalmological diagnosis process. Since there exist variances in viewing angles and anatomical structures across different retinal images, keypoint-based approaches become the mainstream methods for retinal image registration thanks to their robustness and low latency. These methods typically assume the retinal surfaces are planar, and adopt feature matching to obtain the homography matrix that represents the global transformation between images. Yet, such a planar hypothesis inevitably introduces registration errors since retinal surface is approximately curved. This limitation is more prominent when registering image pairs with significant differences in viewing angles. To address this problem, we propose a hybrid registration framework called HybridRetina, which progressively registers retinal images with global and local deformable transformations. For that, we use a keypoint detector and a deformation network called GAMorph to estimate the global transformation and local deformable transformation, respectively. Specifically, we integrate multi-level pixel relation knowledge to guide the training of GAMorph. Additionally, we utilize an edge attention module that includes the geometric priors of the images, ensuring the deformation field focuses more on the vascular regions of clinical interest. Experiments on two widely-used datasets, FIRE and FLoRI21, show that our proposed HybridRetina significantly outperforms some state-of-the-art methods. The code is available at https://github.com/lyp-deeplearning/awesome-retinal-registration.

David Restrepo, Chenwei Wu, Sebastián Andrés Cajas et al.

Large-scale multi-modal deep learning models have revolutionized domains such as healthcare, highlighting the importance of computational power. However, in resource-constrained regions like Low and Middle-Income Countries (LMICs), limited access to GPUs and data poses significant challenges, often leaving CPUs as the sole resource. To address this, we advocate for leveraging vector embeddings to enable flexible and efficient computational methodologies, democratizing multimodal deep learning across diverse contexts. Our paper investigates the efficiency and effectiveness of using vector embeddings from single-modal foundation models and multi-modal Vision-Language Models (VLMs) for multimodal deep learning in low-resource environments, particularly in healthcare. Additionally, we propose a simple yet effective inference-time method to enhance performance by aligning image-text embeddings. Comparing these approaches with traditional methods, we assess their impact on computational efficiency and model performance using metrics like accuracy, F1-score, inference time, training time, and memory usage across three medical modalities: BRSET (ophthalmology), HAM10000 (dermatology), and SatelliteBench (public health). Our findings show that embeddings reduce computational demands without compromising model performance. Furthermore, our alignment method improves performance in medical tasks. This research promotes sustainable AI practices by optimizing resources in constrained environments, highlighting the potential of embedding-based approaches for efficient multimodal learning. Vector embeddings democratize multimodal deep learning in LMICs, particularly in healthcare, enhancing AI adaptability in varied use cases.

Jamie Burke, Justin Engelmann, Charlene Hamid et al.

Purpose: To develop an open-source, fully-automatic deep learning algorithm, DeepGPET, for choroid region segmentation in optical coherence tomography (OCT) data. Methods: We used a dataset of 715 OCT B-scans (82 subjects, 115 eyes) from 3 clinical studies related to systemic disease. Ground truth segmentations were generated using a clinically validated, semi-automatic choroid segmentation method, Gaussian Process Edge Tracing (GPET). We finetuned a UNet with MobileNetV3 backbone pre-trained on ImageNet. Standard segmentation agreement metrics, as well as derived measures of choroidal thickness and area, were used to evaluate DeepGPET, alongside qualitative evaluation from a clinical ophthalmologist. Results: DeepGPET achieves excellent agreement with GPET on data from 3 clinical studies (AUC=0.9994, Dice=0.9664; Pearson correlation of 0.8908 for choroidal thickness and 0.9082 for choroidal area), while reducing the mean processing time per image on a standard laptop CPU from 34.49s ($\pm$15.09) using GPET to 1.25s ($\pm$0.10) using DeepGPET. Both methods performed similarly according to a clinical ophthalmologist, who qualitatively judged a subset of segmentations by GPET and DeepGPET, based on smoothness and accuracy of segmentations. Conclusions: DeepGPET, a fully-automatic, open-source algorithm for choroidal segmentation, will enable researchers to efficiently extract choroidal measurements, even for large datasets. As no manual interventions are required, DeepGPET is less subjective than semi-automatic methods and could be deployed in clinical practice without necessitating a trained operator.