Advances in ophthalmology practice and research最新文献

Background

Uncorrected refractive error is a major cause of vision impairment worldwide and its increasing prevalent necessitates effective screening and management strategies. Meanwhile, deep learning, a subset of Artificial Intelligence, has significantly advanced ophthalmological diagnostics by automating tasks that required extensive clinical expertise. Although recent studies have investigated the use of deep learning models for refractive power detection through various imaging techniques, a comprehensive systematic review on this topic is has yet be done. This review aims to summarise and evaluate the performance of ocular image-based deep learning models in predicting refractive errors.

Main text

We search on three databases (PubMed, Scopus, Web of Science) up till June 2023, focusing on deep learning applications in detecting refractive error from ocular images. We included studies that had reported refractive error outcomes, regardless of publication years. We systematically extracted and evaluated the continuous outcomes (sphere, SE, cylinder) and categorical outcomes (myopia), ground truth measurements, ocular imaging modalities, deep learning models, and performance metrics, adhering to PRISMA guidelines. Nine studies were identified and categorised into three groups: retinal photo-based (n ​= ​5), OCT-based (n ​= ​1), and external ocular photo-based (n ​= ​3).

For high myopia prediction, retinal photo-based models achieved AUC between 0.91 and 0.98, sensitivity levels between 85.10% and 97.80%, and specificity levels between 76.40% and 94.50%. For continuous prediction, retinal photo-based models reported MAE ranging from 0.31D to 2.19D, and R² between 0.05 and 0.96. The OCT-based model achieved an AUC of 0.79–0.81, sensitivity of 82.30% and 87.20% and specificity of 61.70%–68.90%. For external ocular photo-based models, the AUC ranged from 0.91 to 0.99, sensitivity of 81.13%–84.00% and specificity of 74.00%–86.42%, MAE ranges from 0.07D to 0.18D and accuracy ranges from 81.60% to 96.70%. The reported papers collectively showed promising performances, in particular the retinal photo-based and external eye photo -based DL models.

Conclusions

The integration of deep learning model and ocular imaging for refractive error detection appear promising. However, their real-world clinical utility in current screening workflow have yet been evaluated and would require thoughtful consideration in design and implementation.

Purpose

Methods

Results

Conclusions

Background

High myopia is one of the major causes of visual impairment and has an ever-increasing prevalence, especially in East Asia. It is characterized by excessive axial elongation, leading to various blinding complications that extend beyond mere refractive errors and persist immovably after refractive surgery, presenting substantial public health challenge.

Main text

High myopia-related complications include lens pathologies, atrophic and tractional maculopathy, choroidal neovascularization, peripheral retinal degenerations and retinal detachment, and glaucoma and heightened susceptibility to intraocular pressure (IOP) elevation. Pathological lens changes characteristic of high myopia include early cataractogenesis, overgrowth of lens, weakened zonules, and postoperative capsular contraction syndrome, possibly driven by inflammatory pathogenesis, etc. Dome-shaped macula and cilioretinal arteries are two newly identified protective factors for central vision of highly myopic patients. These patients also face risks of open-angle glaucoma and IOP spike following intraocular surgery. Morphologic alternations of optic nerve in high myopia can complicate early glaucoma detection, necessitating comprehensive examinations and close follow-up. Anatomically, thinner trabecular meshwork increases this risk; conversely lamina cribrosa defects may offer a fluid outlet, potentially mitigating the pressure. Notably, anxiety has emerged as the first recognized extra-ocular complication in high myopia, with an underlying inflammatory pathogenesis that connects visual stimulus, blood and brain.

Conclusions

High myopia induces multiple ocular and potential mental health complications, underscoring the need to develop more effective strategies to improve both physical and emotional well-being of these patients, among which anti-inflammation might possibly represent a promising new target.

Background

Myopia is one of the most common eye diseases globally, and has become an increasingly serious health concern among adolescents. Understanding the factors contributing to the onset of myopia and the strategies to slow its progression is critical to reducing its prevalence.

Main text

Animal models are key to understanding of the etiology of human diseases. Various experimental animal models have been developed to mimic human myopia, including chickens, rhesus monkeys, marmosets, mice, tree shrews, guinea pigs and zebrafish. Studies using these animal models have provided evidences and perspectives on the regulation of eye growth and refractive development. This review summarizes the characteristics of these models, the induction methods, common indicators of myopia in animal models, and recent findings on the pathogenic mechanism of myopia.

Conclusions

Investigations using experimental animal models have provided valuable information and insights into the pathogenic mechanisms of human myopia and its treatment strategies.

Purpose

This study aimed to determine the therapeutic effectiveness of different machines in intense pulsed light (IPL) treatment of meibomian gland dysfunction (MGD).

Methods

213 subjects diagnosed with MGD underwent three sessions of IPL treatment in a control (M22) treatment group or experimental (OPL-I) treatment group and were followed up three to four weeks after each session. Tear breakup time (TBUT), meibomian gland secretion scores (MGSS), meibomian gland meibum scores (MGMS), corneal fluorescein staining (CFS) scores, and the Standard Patient Evaluation of Eye Dryness (SPEED) was used to assess eye dryness signs and symptoms at baseline and follow-up visits.

Results

Two machines had the same working principles except that experimental (OPL-I) group consist of a dual filter system. Both groups showed significant improvements (P ​< ​0.0001) in TBUT, MGSS, MGMS, CFS scores and SPEED scores. Non-inferiority analysis showed no statistically significant differences in any result between the two groups. Various defects appeared on the filter with the extension of usage time. Spectrophotometry showed that light intensity decreased to 93.5% ​± ​0.46% past the first filter.

Conclusions

IPL treatment completed with different machines have the same effect on improving the symptoms and signs of MGD. The dual filter system in the IPL machine reduces light intensity by approximately 6.5% without affecting its therapeutic effect. It is a feasible measure to ensure double safety and has the significance of popularization not only for MGD but also in other IPL treatment scenarios.

Objective

To develop and evaluate a Chinese version of the Symptom Questionnaire for Visual Dysfunctions (CSQVD) to quantify visual dysfunction symptoms in school-age children with various eye diseases, and to explore the relationship between ophthalmological disorders and visual dysfunction symptoms.

Methods

Following standard scale adaptation procedures, the Symptom Questionnaire for Visual Dysfunctions (SQVD) was translated into Chinese (CSQVD). We employed random sampling to survey 198 outpatients aged 7–18 to assess the psychometric properties of the CSQVD. Using the reliable and validated questionnaire, we evaluated the determinants of visual dysfunction symptoms among 406 school-age patients at an eye center. The CSQVD scores were correlated with demographic and clinical variables, including gender, age, eye position, refractive power, and best-corrected visual acuity. Univariate analysis identified potential risk factors, followed by binary logistic regression and multiple linear regression analysis on factors with a P-value <0.05.

Results

The CSQVD scale's critical ratio (CR) values ranged from 6.028 to 10.604. The Cronbach's Alpha coefficient was 0.779, and Spearman-Brown split-half reliability was also 0.779. The I-CVI varied from 0.83 to 1.000, the S-CVI/Ave was 0.857, and the KMO value was 0.821. Multifactorial regression analysis indicated that high myopia (OR ​= ​5.744, 95% CI [1.632, 20.218], P ​= ​0.006) and amblyopia (OR ​= ​9.302, 95% CI [1.878, 46.058], P ​= ​0.006) were significant predictors of CSQVD symptoms. Multiple linear regression analysis showed that BCVA of amblyopic eyes (B ​= ​−5.052, 95% CI [−7.779, 2.325], P ​= ​0.000) and SE power (B ​= ​−0.234, 95% CI [−0.375, 0.205], P ​= ​0.001) significantly affected the CSQVD scale scores.

Conclusions

The Chinese version of the SQVD scale (CSQVD) demonstrates good feasibility, discriminatory power, validity, and reliability in assessing Chinese school-aged children. Furthermore, those who have severe myopia and amblyopia reported more visual dysfunction symptoms.

Purpose

Observational studies have reported positive associations between glaucoma and stroke; however, controversial results exist. Importantly, the nature of the relationship remains unknown since previous studies were not designed to test causality. Therefore, we aimed to investigate the possible causal relationships between glaucoma and stroke.

Methods

Our two-sample Mendelian randomization (MR) encompassed multi-ethnic large-scale genome-wide association studies with more than 20000 cases and 260000 controls for glaucoma, and more than 80000 cases and 630000 controls for stroke. Individual effect estimates for each SNP were combined using the inverse-variance weighted (IVW) method. To avoid potential pleiotropic effects, we adjusted the main results by excluding genetic variants associated with metabolic factors. The weighted median and MR-Egger methods were also used for the sensitivity analysis.

Results

Our MR analysis revealed that glaucoma and its subtypes, including primary open-angle glaucoma and primary angle-closure glaucoma, exhibited no causal role in relation to any stroke (AS), any ischemic stroke (AIS), large-artery atherosclerotic stroke (LAS), small-vessel stroke (SVS), or cardioembolic stroke (CES) across MR analyses (all P ​> ​0.05). The null associations remained robust even after adjusting for metabolic-related traits and were consistent in both the European and Asian populations. Furthermore, reverse MR analyses also did not indicate any significant causal effects of AS, AIS, LAS, or CES on glaucoma risk.

Conclusions

Evidence from our series of causal inference approaches using large-scale population-based MR analyses did not support causal effects between glaucoma and stroke. These findings suggest that the relationship of glaucoma management and stroke risk prevention should be carefully evaluated in future studies. In turn, stroke diagnosis should not be simply applied to glaucoma risk prediction.