Current Opinion in Ophthalmology最新文献

Purpose of review: This review examines the emerging role of artificial intelligence (AI) in the detection and management of pediatric retinal diseases, with a focus on systems that achieve expert-level performance in classifying fundus abnormalities. We highlight models developed for clinical application as assistive or autonomous tools with the potential to enhance early identification and referral, streamline care delivery, and improve access to care.

Recent findings: AI systems, have shown high diagnostic accuracy in identifying retinal pathology associated with retinopathy of prematurity, pediatric myopia, diabetic retinopathy, and retinoblastoma. Many of these systems have potential utility in real-world screening, supplementing clinical decision-making, and guiding early intervention. In addition, autonomous AI systems can increase access in low-resource, remote settings and areas where access to pediatric ophthalmologists is limited. Ongoing advances include integration with nonmydriatic fundus photography, smartphone-based imaging, and image-free diagnostic modalities, further expanding reach and applicability.

Summary: AI holds transformative promise for pediatric retina care by providing scalable, accurate, and accessible screening solutions. These systems have demonstrated to enhance clinical expertise, minimize interobserver variability, facilitate timely referrals and support decision-making. As integration of algorithms into clinical and community settings is established, AI is poised to become an essential component of pediatric ophthalmology, improving early detection and reducing the global burden of preventable childhood blindness.

Purpose of review: With the rise of 'oculomics' and the application of advanced artificial intelligence techniques in healthy ageing, retinal imaging, the only way we can directly visualize the microvascular circulation, is expanding beyond ophthalmology into broader systemic health monitoring. The purpose of this review is to summarize recent advances in this rapidly evolving field and assess the opportunities, challenges, and future directions of the use of oculomics in translating into real-world clinical use.

Recent findings: Retinal imaging modalities, such as color fundus photography, optical coherence tomography (OCT), OCT angiography (OCTA), and wide-field imaging, are increasingly integrated with deep learning algorithms to detect, predict, and manage a broad spectrum of systemic diseases, including cardiovascular, cerebrovascular, renal, metabolic, and neurodegenerative disorders, as well as less commonly studied conditions. While research in more established areas is beginning to address clinical translation and implementation, significant challenges remain before these technologies can be reliably adopted in long-term, real-world healthcare settings.

Summary: Artificial intelligence applied to retinal imaging has matured from proof-of-concept classifiers to externally validated, occasionally regulated tools that noninvasively profile systemic conditions. Multiplexed foundation models and multimodal transformers herald a shift toward holistic 'oculomics' platforms, yet prospective multicenter trials, equitable performance auditing, and health-economic evaluations are essential before widescale clinical adoption.

Purpose of review: The current article provides an overview of the utility of artificial intelligence approaches to aid in the design, recruitment, execution, and dissemination of ophthalmic clinical trials.

Recent findings: Within the last decade, artificial intelligence has heralded a new age for ophthalmology, with novel applications habitually appearing within the literature. Though clinical trials are considered the gold standard for driving evidence-based practice, remarkably few studies have examined the potential for machine learning to augment the clinical trial pipeline. Clinical trials within ophthalmology often do not reach planned endpoints due to insufficient enrolment, cost overruns, and can lack reliability from unblinded outcome assessors. Ones that do, frequently take longer to enroll patients than intended. Artificial intelligence-based approaches have recently been shown to be effective in identifying eligible clinical trial participants using both imaging and text data.

Summary: Given the key role of clinical trials in the advancement of ophthalmic clinical practice, trialists should consider the potential for artificial intelligence-powered tools to enhance the design, recruitment, and delivery of future studies.

Purpose of review: To highlight the ocular manifestations associated with COVID-19 and review the evidence surrounding proposed associations with corneal transplant rejection and uveitis reactivation.

Recent findings: SARS-CoV-2 has been associated with various ocular manifestations, most commonly conjunctivitis. Less frequently, more severe complications such as uveitis, retinal vascular occlusions, and neuro-ophthalmic syndromes have been reported. These manifestations are theorized to occur secondary to both direct viral invasion as well as secondary systemic inflammation and immune dysregulation. While case reports suggest a possible link between COVID-19 infection and vaccination and corneal graft rejection, large cohort studies and meta-analyses have not confirmed an association. In contrast, retrospective data suggest a slight increase in uveitis recurrence following COVID-19 infection and vaccination in patients with prior uveitis.

Summary: Ocular involvement in COVID-19 ranges from mild conjunctivitis to rare, vision-threatening inflammatory and vascular complications. While case reports have suggested temporal links between COVID-19 infection or vaccination and corneal transplant rejection or uveitis reactivation, current evidence does not support an association with graft rejection, and only a slight increase in uveitis recurrence has been observed in certain high-risk populations. With longer follow-up and prospective analyses, a clearer understanding of underlying mechanisms and risk factors may emerge to better guide ophthalmic management during and beyond the pandemic.

Purpose of review: Alongside the development of large language models (LLMs) and generative artificial intelligence (AI) applications across a diverse range of clinical applications in Ophthalmology, this review highlights the importance of evaluation of LLM applications by discussing evaluation metrics commonly adopted.

Recent findings: Generative AI applications have demonstrated encouraging performance in clinical applications of Ophthalmology. Beyond accuracy, evaluation in the form of quantitative and qualitative metrics facilitate a more nuanced assessment of LLM output responses. Several challenges limit evaluation including the lack of consensus on standardized benchmarks, and limited availability of robust and curated clinical datasets.

Summary: This review outlines the spectrum of quantitative and qualitative evaluation metrics adopted in existing studies, highlights key challenges in LLM evaluation, to catalyze further work towards standardized and domain-specific evaluation. Robust evaluation to effectively validate clinical LLM applications is crucial in closing the gap towards clinical integration.

Purpose of review: Photobiomodulation (PBM) is a noninvasive therapy utilising low-level light energy to stimulate cellular processes, modulate inflammatory pathways, enhance mitochondrial activity, promote tissue regeneration. With growing interest in PBM as a potential treatment modality, this review synthesises current evidence and highlights challenges of implementing PBM across various ophthalmic conditions.

Recent findings: Current ophthalmic applications of PBM can be categorised into established and exploratory therapies, differentiated primarily by the attainment of regulatory approval. Established applications of PBM include the treatment of dry eye disease and nonexudative age-related macular degeneration, while the use of PBM is still largely exploratory in conditions such as diabetic macular oedema and retinopathy of prematurity. Regardless of the level of regulatory authorisation, the application of PBM in each ophthalmic condition presents with distinct challenges requiring further research for comprehensive validation.

Summary: While PBM holds promise as a novel therapeutic option, its long-term efficacy and safety remains to be fully established. Standardised treatment guidelines and larger randomised controlled trials are essential to optimise its use in future clinical practice.

Purpose of review: Neovascular glaucoma (NVG) is a severe secondary glaucoma precipitated by ocular ischemia and abnormal neovascularization, resulting in elevated intraocular pressure (IOP) and vision loss if not promptly addressed. This study evaluates recent advances in both medical and surgical management of NVG, focusing on strategies that integrate anti-vascular endothelial growth factor (VEGF) therapy, retinal ablation, and evolving surgical techniques.

Recent findings: Anti-VEGF agents remain central to NVG treatment, with newer agents and combination regimens showing sustained neovascular suppression, alongside panretinal photocoagulation as an additional cornerstone in reducing neovascular drive. Glaucoma drainage devices continue to have prominence for their ability to bypass fibrotic outflow pathways, while trabeculectomy augmented with mitomycin C continues to offer a viable option in select cases. Cyclodestructive procedures, including micropulse transscleral cyclophotocoagulation and endoscopic cyclophotocoagulation, provide alternative surgical avenues for refractory cases. Novel studies including lipidomic analyses present novel metabolic pathways that are potentially implicated in NVG pathogenesis, suggesting future targets beyond VEGF.

Summary: Timely recognition and comprehensive treatment - encompassing IOP control, ischemic drive reduction, and inflammation management - remain critical in the management of NVG. As research illuminates additional molecular targets and refines surgical interventions, the promise of a more personalized, biomarker-driven approach to NVG management continues to grow.

Purpose of review: Many ocular neovascular diseases, including diabetic retinopathy (DR) and diabetic macular edema (DME), are significant complications that can worsen during pregnancy, posing risks to vision. Antivascular endothelial growth factor (anti-VEGF) agents represent the mainstay of DR and DME treatment in nonpregnant populations. However, their use in pregnant populations remain controversial due to concerns over fetal safety. This review explores the available literature on anti-VEGF use during pregnancy, evaluating maternal and fetal outcomes, and incorporating lessons learned from their recent use in treating retinopathy of prematurity (ROP).

Recent findings: Although human data on the safety of anti-VEGF agents during pregnancy remain limited, available studies suggest that systemic absorption is minimal, and concerns regarding potential effects on fetal angiogenesis largely theoretical. Case reports and retrospective studies have not consistently demonstrated harm. Furthermore, several studies examining long-term systemic effects of anti-VEGF use in treating premature infants with ROP reveal no association with adverse neurodevelopmental outcomes when compared to untreated ROP or ROP treated with laser photocoagulation alone.

Summary: While the use of anti-VEGF agents in pregnancy remains controversial, recent evidence suggests they should not be categorically excluded as a treatment option. Until more causal studies emerge, treatment decisions should be individualized, balancing disease severity, maternal visual function, and fetal safety.

Purpose of review: To review the role of microbial cultures in managing endophthalmitis following cataract surgery and intravitreal injections.

Recent findings: The endophthalmitis vitrectomy study (EVS) continues to guide the management of endophthalmitis including its recommendation for intraocular vitreous sampling. Vitreous cultures have better diagnostic yield than aqueous cultures. Advances in culture techniques have improved the efficiency and sensitivity of pathogen identification. While culture results are useful for prognostication, microbiology data rarely changes management in endophthalmitis following cataract surgery or intravitreal injections. Endophthalmitis managed without microbial cultures had similar outcomes to eyes managed with microbial cultures.

Summary: Despite advances in culture techniques, most institutions still rely on gram stain and microbial cultures, often requiring several days for results. Initial administration of empiric antibiotics occurs before microbiology data becomes available, limiting the utility of cultures in clinical management. Despite concerns for antibiotic resistance, the large majority of postcataract surgery and postintravitreal injection endophthalmitis pathogens are susceptible to vancomycin and ceftazidime. Given this, obtaining intraocular specimens can be deferred if it poses a delay to prompt treatment with intravitreal antibiotics. Emerging culture techniques hold promise for more rapid and sensitive pathogen identification, which could change the landscape of culture guidelines and improve clinical usefulness.