Search

This study quantifies the performance of an international cohort of cornea specialists in image-based differentiation of bacterial and fungal keratitis, identifying significant regional variation and establishing a reference standard for comparison against machine learning models.

PurposeTo develop and evaluate a deep learning system for differentiating between eyes with and without glaucomatous visual field damage (GVFD) and predicting the severity of GFVD from spectral domain OCT (SD OCT) optic nerve head images.DesignEvaluation of a diagnostic technology.ParticipantsA total of 9765 visual field (VF) SD OCT pairs collected from 1194 participants with and without GVFD (1909 eyes).MethodsDeep learning models were trained to use SD OCT retinal nerve fiber layer (RNFL) thickness maps, RNFL en face images, and confocal scanning laser ophthalmoscopy (CSLO) images to identify eyes with GVFD and predict quantitative VF mean deviation (MD), pattern standard deviation (PSD), and mean VF sectoral pattern deviation (PD) from SD OCT data.Main outcome measuresDeep learning models were compared with mean RNFL thickness for identifying GVFD using area under the curve (AUC), sensitivity, and specificity. For predicting MD, PSD, and mean sectoral PD, models were evaluated using R2 and mean absolute error (MAE).ResultsIn the independent test dataset, the deep learning models based on RNFL en face images achieved an AUC of 0.88 for identifying eyes with GVFD and 0.82 for detecting mild GVFD significantly (P < 0.001) better than using mean RNFL thickness measurements (AUC = 0.82 and 0.73, respectively). Deep learning models outperformed standard RNFL thickness measurements in predicting all quantitative VF metrics. In predicting MD, deep learning models based on RNFL en face images achieved an R2 of 0.70 and MAE of 2.5 decibels (dB) compared with 0.45 and 3.7 dB for RNFL thickness measurements. In predicting mean VF sectoral PD, deep learning models achieved high accuracy in the inferior nasal (R2 = 0.60) and superior nasal (R2 = 0.67) sectors, moderate accuracy in inferior (R2 = 0.26) and superior (R2 = 0.35) sectors, and lower accuracy in the central (R2 = 0.15) and temporal (R2 = 0.12) sectors.ConclusionsDeep learning models had high accuracy in identifying eyes with GFVD and predicting the severity of functional loss from SD OCT images. Accurately predicting the severity of GFVD from SD OCT imaging can help clinicians more effectively individualize the frequency of VF testing to the individual patient.

Infectious keratitis is a major global cause of visual impairment and blindness, often affecting marginalized populations. Proper diagnosis of the causative organism is critical, and although culture remains the prevailing diagnostic tool, newer techniques such as in vivo confocal microscopy are helpful for diagnosing fungus and Acanthamoeba. Next-generation sequencing holds the potential for early and accurate diagnosis even for organisms that are difficult to culture by conventional methods. Topical antibiotics remain the best treatment for bacterial keratitis, and a recent review found all commonly prescribed topical antibiotics to be equally effective. However, outcomes remain poor secondary to corneal melting, scarring, and perforation. Adjuvant therapies aimed at reducing the immune response associated with keratitis include topical corticosteroids. The large, randomized, controlled Steroids for Corneal Ulcers Trial found that although steroids provided no significant improvement overall, they did seem beneficial for ulcers that were central, deep or large, non-Nocardia, or classically invasive Pseudomonas aeruginosa; for patients with low baseline vision; and when started early after the initiation of antibiotics. Fungal ulcers often have worse clinical outcomes than bacterial ulcers, with no new treatments since the 1960s when topical natamycin was introduced. The randomized controlled Mycotic Ulcer Treatment Trial (MUTT) I showed a benefit of topical natamycin over topical voriconazole for fungal ulcers, particularly among those caused by Fusarium. MUTT II showed that oral voriconazole did not improve outcomes overall, although there may have been some effect among Fusarium ulcers. Given an increase in nonserious adverse events, the authors concluded that they could not recommend oral voriconazole. Viral keratitis differs from bacterial and fungal cases in that it is often recurrent and is common in developed countries. The Herpetic Eye Disease Study (HEDS) I showed a significant benefit of topical corticosteroids and oral acyclovir for stromal keratitis. HEDS II showed that oral acyclovir decreased the recurrence of any type of herpes simplex virus keratitis by approximately half. Future strategies to reduce the morbidity associated with infectious keratitis are likely to be multidimensional, with adjuvant therapies aimed at modifying the immune response to infection holding the greatest potential to improve clinical outcomes.

ObjectiveTo compare subjective and objective clinical tests used in the screening for hydroxychloroquine retinal toxicity to multifocal electroretinography (mfERG) reference testing.DesignProspective, single-center, case control study.ParticipantsFifty-seven patients with a previous or current history of hydroxychloroquine treatment of more than 5 years' duration.MethodsParticipants were evaluated with a detailed medical history, dilated ophthalmologic examination, color fundus photography, fundus autofluorescence (FAF) imaging, spectral-domain (SD) optical coherence tomography (OCT), automated visual field testing (10-2 visual field mean deviation [VFMD]), and mfERG testing. We used mfERG test parameters as a gold standard to divide participants into 2 groups: those affected by hydroxychloroquine-induced retinal toxicity and those unaffected.Main outcome measuresWe assessed the association of various imaging and psychophysical variables in the affected versus the unaffected group.ResultsFifty-seven study participants (91.2% female; mean age, 55.7±10.4 years; mean duration of hydroxychloroquine treatment, 15.0±7.5 years) were divided into affected (n = 19) and unaffected (n = 38) groups based on mfERG criteria. Mean age and duration of hydroxychloroquine treatment did not differ statistically between groups. Mean OCT retinal thickness measurements in all 9 macular subfields were significantly lower (

ObjectiveTo evaluate the ability of a binocular automated pupillograph to discriminate healthy subjects from those with glaucoma.DesignCross-sectional observational study.ParticipantsBoth eyes of 116 subjects, including 66 patients with glaucoma in at least 1 eye and 50 healthy subjects from the Diagnostic Innovations in Glaucoma Study. Eyes were classified as glaucomatous by repeatable abnormal standard automated perimetry (SAP) or progressive glaucomatous changes on stereophotographs.MethodsAll subjects underwent automated pupillography using the RAPDx pupillograph (Konan Medical USA, Inc., Irvine, CA).Main outcome measuresReceiver operating characteristic (ROC) curves were constructed to assess the diagnostic ability of pupil response parameters to white, red, green, yellow, and blue full-field and regional stimuli. A ROC regression model was used to investigate the influence of disease severity and asymmetry on diagnostic ability.ResultsThe largest area under the ROC curve (AUC) for any single parameter was 0.75. Disease asymmetry (P

ObjectiveTo evaluate changes in peripapillary and macular choroidal thickness and volume after the water-drinking test (WDT) using swept-source optical coherence tomography (SS OCT).DesignProspective, cross-sectional, observational study.ParticipantsFifty-six eyes of 28 healthy volunteers.MethodsParticipants underwent a 3-dimensional optic disc and macula scanning protocol with a prototype SS OCT (Topcon, Inc., Tokyo, Japan) at baseline and 15, 30, 45, and 120 minutes after the start of the WDT. The WDT consisted of drinking 1000 ml of water within 5 minutes. Objective measurements of the choroid were obtained with automated segmentation of the choroidal boundaries.Main outcome measuresChoroidal thickness and volume.ResultsMean age ± standard deviation of participants was 35.6 ± 9.1 years. Intraocular pressure (IOP) increased from 14.9 ± 2.7 mmHg at baseline to a peak of 16.8 ± 3.0 mmHg 15 minutes after the WDT (P