A Deep Learning Approach for Accurate Discrimination Between Optic Disc Drusen and Papilledema on Fundus Photographs.

BACKGROUND: Optic disc drusen (ODD) represent an important differential diagnosis of papilledema caused by intracranial hypertension, but their distinction may be difficult in clinical practice. The aim of this study was to train, validate, and test a dedicated deep learning system (DLS) for binary classification of ODD vs papilledema (including various subgroups within each category), on conventional mydriatic digital ocular fundus photographs collected in a large international multiethnic population. METHODS: This retrospective study included 4,508 color fundus images in 2,180 patients from 30 neuro-ophthalmology centers (19 countries) participating in the Brain and Optic Nerve Study with Artificial Intelligence (BONSAI) Group. For training and internal validation, we used 857 ODD images and 3,230 papilledema images, in 1,959 patients. External testing was performed on an independent data set (221 patients), including 207 images with ODD (96 visible and 111 buried), provided by 3 centers of the Optic Disc Drusen Studies Consortium, and 214 images of papilledema (92 mild-to-moderate and 122 severe) from a previously validated study. RESULTS: The DLS could accurately distinguish between all ODD and papilledema (all severities included): area under the receiver operating characteristic curve (AUC) 0.97 (95% confidence interval [CI], 0.96-0.98), accuracy 90.5% (95% CI, 88.0%-92.9%), sensitivity 86.0% (95% CI, 82.1%-90.1%), and specificity 94.9% (95% CI, 92.3%-97.6%). The performance of the DLS remained high for discrimination of buried ODD from mild-to-moderate papilledema: AUC 0.93 (95% CI, 0.90-0.96), accuracy 84.2% (95% CI, 80.2%-88.6%), sensitivity 78.4% (95% CI, 72.2%-84.7%), and specificity 91.3% (95% CI, 87.0%-96.4%). CONCLUSIONS: A dedicated DLS can accurately distinguish between ODD and papilledema caused by intracranial hypertension, even when considering buried ODD vs mild-to-moderate papilledema.

Factors associated with dengue prevention and control in two villages in a central Thai province: a retrospective review.

OBJECTIVE: To study the factors associated with dengue prevention and control in Moo 6 (the 6th village) and Moo 7 of Tambon Kaeng-phak-kut, Thaluang District, Lopburi Province. MATERIAL AND METHOD: The authors reviewed the raw data collected by public health officers and village health volunteers (VHVs) as their routine tasks. The authors analyzed the data, 30 dwellings per each village, to compare the demographics, knowledge, attitude, and practice of subjects from Moo 6, a dengue-outbreak community, with that from Moo 7, a control group, as well as larval indices between these 2 studied groups. The present retrospective study is approved by Siriraj Institutional Review Board, Certificate of Approval No. Si393/2012. RESULTS: Both groups of subjects had no statistically significant difference in basic dengue knowledge (p = 0.862), attitude towards dengue prevention and control, practical knowledge (p = 0.457), and actual practice to eliminate Aedes larvae and prevent it laying eggs, except for the practice of managing water container in bathroom or toilet (p = 0.015). On the other hand, dengue incidence and larval indices of both villages were apparently different. CONCLUSION: Although incorrect basic dengue and practical knowledge of subjects from both villages were similar dengue outbreak in Moo 6 of Tambon Kaeng-phak-kut was superior. It may be due to difference in actual practice on larval elimination in water container in bathroom or toilet as well as other factors other than personal factors such as public services, public places, and community surroundings.

A Deep Learning System for Automated Quality Evaluation of Optic Disc Photographs in Neuro-Ophthalmic Disorders.

The quality of ocular fundus photographs can affect the accuracy of the morphologic assessment of the optic nerve head (ONH), either by humans or by deep learning systems (DLS). In order to automatically identify ONH photographs of optimal quality, we have developed, trained, and tested a DLS, using an international, multicentre, multi-ethnic dataset of 5015 ocular fundus photographs from 31 centres in 20 countries participating to the Brain and Optic Nerve Study with Artificial Intelligence (BONSAI). The reference standard in image quality was established by three experts who independently classified photographs as of "good", "borderline", or "poor" quality. The DLS was trained on 4208 fundus photographs and tested on an independent external dataset of 807 photographs, using a multi-class model, evaluated with a one-vs-rest classification strategy. In the external-testing dataset, the DLS could identify with excellent performance "good" quality photographs (AUC = 0.93 (95% CI, 0.91-0.95), accuracy = 91.4% (95% CI, 90.0-92.9%), sensitivity = 93.8% (95% CI, 92.5-95.2%), specificity = 75.9% (95% CI, 69.7-82.1%) and "poor" quality photographs (AUC = 1.00 (95% CI, 0.99-1.00), accuracy = 99.1% (95% CI, 98.6-99.6%), sensitivity = 81.5% (95% CI, 70.6-93.8%), specificity = 99.7% (95% CI, 99.6-100.0%). "Borderline" quality images were also accurately classified (AUC = 0.90 (95% CI, 0.88-0.93), accuracy = 90.6% (95% CI, 89.1-92.2%), sensitivity = 65.4% (95% CI, 56.6-72.9%), specificity = 93.4% (95% CI, 92.1-94.8%). The overall accuracy to distinguish among the three classes was 90.6% (95% CI, 89.1-92.1%), suggesting that this DLS could select optimal quality fundus photographs in patients with neuro-ophthalmic and neurological disorders affecting the ONH.

Through the eyes into the brain, using artificial intelligence.

INTRODUCTION: Detection of neurological conditions is of high importance in the current context of increasingly ageing populations. Imaging of the retina and the optic nerve head represents a unique opportunity to detect brain diseases, but requires specific human expertise. We review the current outcomes of artificial intelligence (AI) methods applied to retinal imaging for the detection of neurological and neuro-ophthalmic conditions. METHOD: Current and emerging concepts related to the detection of neurological conditions, using AI-based investigations of the retina in patients with brain disease were examined and summarised. RESULTS: Papilloedema due to intracranial hypertension can be accurately identified with deep learning on standard retinal imaging at a human expert level. Emerging studies suggest that patients with Alzheimer's disease can be discriminated from cognitively normal individuals, using AI applied to retinal images. CONCLUSION: Recent AI-based systems dedicated to scalable retinal imaging have opened new perspectives for the detection of brain conditions directly or indirectly affecting retinal structures. However, further validation and implementation studies are required to better understand their potential value in clinical practice.

Structural and functional differentiation between compressive and glaucomatous optic neuropathy.

Clinical diagnoses of slow, progressive, painless visual losses with various degrees of visual field (VF) losses and disc atrophy are often confused between suprasellar compressive optic neuropathy (CON) and open-angle glaucomatous optic neuropathy (GON). We plotted the thickness of the peripapillary retinal nerve fiber layer (RNFL) and macular ganglion cell-inner plexiform layer (GCIPL) against the mean deviation (MD) of the VF of 34 eyes of CON at diagnosis, 30 eyes of CON after therapy, 29 eyes of GON, and 60 eyes of healthy controls in a cross-sectional investigation. At diagnosis, a disproportionally early pattern of structural thinning compared with the corresponding VF losses was unique to CON. GON- and CON-specific thinning parameters were generally useful in differentiating GON and CON from moderate to severe MD losses, but early MD losses (0 to - 6 dB) overlapped with GON in a CON-stage specific manner. GON-specific thinning parameters, RNFL in the inferior sector, and inferior to temporal macular GCIPL ratio showed overlap with posttreatment CON in the early MD losses with AUCs of 0.916 (95% CI 0.860-0.971; P < 0.001) and 0.890 (95% CI 0.811-0.968; P < 0.001), respectively. In comparison, CON-specific thinning parameters, superonasal, and inferonasal GCIPL showed overlap with CON at diagnosis for early MD losses. Overall, the nasal-to-temporal macular GCIPL ratio showed good discrimination between CON and GON throughout the MD range, with an AUC of 0.923 (95% CI 0.870-0.976; P < 0.001). Comparing GON with all stages of CON, the cut-point of 0.95 showed the lower nasal-to-temporal GCIPL ratio had a sensitivity of 72% and specificity of 90% for CON. However, the cut-point of 1.10 showed the superior-to-inferior GCIPL ratio had a sensitivity of 60% and specificity of 98% for GON.

The epidemiology and mutation types of Leber's hereditary optic neuropathy in Thailand.

PURPOSE: Leber's hereditary optic neuropathy (LHON), the most common mitochondrial optic neuropathy, causes visual loss, especially in young adults. Due to the absence of epidemiological data in Southeast Asia, we aimed to determine Thai LHON patients' characteristics (demographic data, mutation types, and prognoses) as the first study in this region. METHODS: This retrospective chart review enrolled all Thai LHON patients confirmed by three mitochondrial DNA mutations (G11778A, T14484C, and G3460A) between January 1997 and December 2016. Patients with more than one year of follow-up were included in a visual progression analysis. The Mann-Whitney U-test was applied to compare groups, and prognosis-associated factors were analysed with the generalized estimating equation. RESULTS: In all, 229 patients were enrolled, with only nineteen females. Most mutations were of the G11778A type (91%), with T14484C accounting for the remainder. The age at onset of G11778A (21.9 years; interquartile range [IQR] 14.9, 33.5) was younger than that of T14484C (33.0 years; IQR 19.4, 37.5). Of 45 patients, the T14484C group demonstrated good vision recovery, whereas the G11778A group did not improve (difference in logMAR -0.7 and IQR -1.5, -0.2 versus logMAR 0.0 and IQR -0.3, 0.2, respectively; P value .001). The G11778A mutation, male, and older age were related to poor prognoses. CONCLUSIONS: The leading mutation in Thai LHON patients is the G11778A missense, followed by T14484C, while G3460A was not detected. The vast majority of patients were young adult males. The G11778A mutation, older age, and male gender are associated with poor vision outcomes. Key messageThe G11778A missense mutation is the most common among Thai LHON patients, followed by T14484C, while G3460A was not found. The G11778A mutation, older age, and male gender are associated with poor vision outcomes.

Association between the levels of prostaglandin E2 in tears and severity of dry eye.

AIM: To investigate the relationship between the levels of prostaglandin E2 (PGE2) in tears and dry eye disease severity based on both clinical symptoms and signs. METHODS: Tear samples were collected from 36 non-Sjögren syndrome dry eye patients (10 males and 26 females, mean age 50.11±11.17y). All participants completed the Ocular Surface Disease Index (OSDI) questionnaire and underwent a detailed ophthalmic examination including, tear film breakup time (TBUT), ocular surface fluorescein staining, Schirmer I test, and meibomian gland assessment. The level of PGE2 in tears was measured using enzyme-linked immunosorbent assay (ELISA). The independent associations between tear PGE2 levels and other variables including demographics, OSDI scores, TBUT, Schirmer scores, ocular surface staining scores, and stage of meibomian gland dysfunction (MGD) were evaluated using linear regression analysis. RESULTS: The mean PGE2 level in tears of dry eye patients was 537.85±234.02 pg/mL. The tear PGE2 levels significantly positively correlated with OSDI scores (R=0.608, P<0.001), however, they did not significantly associate with TBUT (R=0.153, P=0.373), Schirmer scores (R=-0.098, P=0.570), ocular surface staining scores (R=0.282, P=0.095), and stage of MGD (R=-0.107, P=0.535). Male sex was significantly negatively correlated with tear PGE2 levels. CONCLUSION: The levels of PGE2 in tears are positively correlated with dry eye symptoms. However, no significant association was found between tear PGE2 levels and the results of other common dry eye diagnostic tests.