Ocular blood flow in preterm neonates.

Retinopathy of prematurity (ROP) is a disorder affecting low birthweight, preterm neonates. In the preterm eye, the retina is not fully developed and neovascularization may occur at the margin between the developed vascular retina and undeveloped avascular retina. Without timely treatment by laser or intravitreal anti-vascular endothelial growth factor (VEGF) therapy, this can lead to tractional retinal detachment and blindness. Visualization of the retina in regular examinations by indirect ophthalmoscopy is hence the current standard of care, but the exams are stressful and interpretation of images is subjective. The upregulation of VEGF in ROP would suggest an increase in ocular blood flow. In this report, we evaluate the potential of ultrafast plane-wave Doppler ultrasound (PWU) to detect increased flow velocities in the orbital vessels supplying the eye in a gentle exam with objective findings. We imaged both eyes of 50 low-birthweight preterm neonates using 18 MHz PWU. Flow velocity in the central retinal artery (CRA) and vein (CRV), and the short posterior ciliary arteries were determined and values at each ROP Stage compared. We found significantly increased velocities in the CRA and CRV in Stage 3 ROP eyes, where intervention would be considered. We compared multivariate models for identifying Stage 3 eyes comprised solely of clinical factors, solely of Doppler parameters, and clinical plus Doppler parameters. The respective models provided areas under their respective ROC curves of 0.760, 0.812, and 0.904. PWU Doppler represents a gentle, objective means for identifying neonates at risk for ROP that could complement ophthalmoscopy.

A pilot optical coherence tomography angiography classification of retinal neovascularization in retinopathy of prematurity.

Extraretinal neovascularization is a hallmark of treatment-requiring retinopathy of prematurity (ROP). Optical coherence tomography angiography (OCTA) offers vascular flow and depth information not available from indirect ophthalmoscopy and structural OCT, but OCTA is only commercially available as a tabletop device. In this study, we used an investigational handheld OCTA device to study the vascular flow in and around retinal neovascularization in seven preterm infants with treatment-requiring ROP and contrasted them to images of vascular flow in six infants of similar age without neovascular ROP. We showed stages of retinal neovascularization visible in preterm infants from 32 to 47 weeks postmenstrual age: Intraretinal neovascularization did not break through the internal limiting membrane; Subclinical neovascular buds arose from retinal vasculature with active flow through the internal limiting membrane; Flat neovascularization in aggressive ROP assumed a low-lying configuration compared to elevated extraretinal neovascular plaques; Regressed neovascularization following treatment exhibited decreased vascular flow within the preretinal tissue, but flow persisted in segments of retinal vessels elevated from their original intraretinal location. These findings enable a pilot classification of retinal neovascularization in eyes with ROP using OCTA, and may be helpful in detailed monitoring of disease progression, treatment response and predicting reactivation.

Evaluation of retinal vascularization in retinopathy of prematurity regressed after intravitreal ranibizumab monotherapy or without treatment based on fluorescein angiography.

To investigate the fluorescein angiography (FA) findings and compare the extent of retinal vascularization in retinopathy of prematurity (ROP), recovered after intravitreal ranibizumab (IVR) monotherapy and those regressed spontaneously. Infants with a history of ROP who underwent FA between April 2018 and November 2021 were retrospectively included. The patients were divided into two groups based on whether they had received IVR (IVR group) or had ROP that regressed spontaneously without treatment (untreated group). The differences between the two groups in zone II ROP were also compared, to equalize the subgroups as much as possible in terms of disease severity. FA findings were recorded. The extent of vascularization was measured by the ratio of the distance from the center of the disk to the border of the vascularized zone (DB) and the distance from the center of the disk to the center of the fovea (DF). The width of the persistent avascular retina (PAR) was counted by disc diameters (DD). One hundred and ten eyes of 55 infants were included in the IVR group and 76 eyes of 38 babies in the untreated group. The ratio of abnormal shape of vessels was significantly higher in the IVR group than in the untreated group (50.9% vs. 35.5%; P = 0.038), while the linear choroidal filling pattern, tortuosity of vessels over the posterior pole, dye leakage, anomalous branching of vessels, circumferential vessels, arteriovenous shunt, abnormal capillary bed, and macular abnormalities were similarly. There was a smaller temporal DB/DF ratio (4.48 vs. 4.63; P = 0.003) and greater PAR (2.63 vs. 1.76; P < 0.001) in the IVR group compared to the untreated group. In zone II ROP, the progression of retinal vascularization was significantly larger in the IVR group than that in the untreated group (P = 0.003), while no statistical differences were observed in FA features, the DB/DF ratio, and PAR between the two subgroups. The residual vascular abnormalities and PAR may be common results of ROP regression. The DB/DF ratio of 4.0 temporally and 3.3 nasally could be used as the preliminary indicators for safe retinal vascularization in the completion of ROP regression.

ROP-GAN: an image synthesis method for retinopathy of prematurity based on generative adversarial network.

Objective. Training data with annotations are scarce in the intelligent diagnosis of retinopathy of prematurity (ROP), and existing typical data augmentation methods cannot generate data with a high degree of diversity. In order to increase the sample size and the generalization ability of the classification model, we propose a method called ROP-GAN for image synthesis of ROP based on a generative adversarial network.Approach. To generate a binary vascular network from color fundus images, we first design an image segmentation model based on U2-Net that can extract multi-scale features without reducing the resolution of the feature map. The vascular network is then fed into an adversarial autoencoder for reconstruction, which increases the diversity of the vascular network diagram. Then, we design an ROP image synthesis algorithm based on a generative adversarial network, in which paired color fundus images and binarized vascular networks are input into the image generation model to train the generator and discriminator, and attention mechanism modules are added to the generator to improve its detail synthesis ability.Main results. Qualitative and quantitative evaluation indicators are applied to evaluate the proposed method, and experiments demonstrate that the proposed method is superior to the existing ROP image synthesis methods, as it can synthesize realistic ROP fundus images.Significance. Our method effectively alleviates the problem of data imbalance in ROP intelligent diagnosis, contributes to the implementation of ROP staging tasks, and lays the foundation for further research. In addition to classification tasks, our synthesized images can facilitate tasks that require large amounts of medical data, such as detecting lesions and segmenting medical images.

Ocular Pigmentation Impact on Retinal Versus Choroidal Optical Coherence Tomography Imaging in Preterm Infants.

Purpose: To evaluate the association of fundus pigmentation with the visibility of retinal versus choroidal layers on optical coherence tomography (OCT) in preterm infants. Methods: For infants enrolled in BabySTEPS, ophthalmologists recorded fundus pigmentation (blond, medium, or dark) at the first retinopathy of prematurity (ROP) examination. Bedside OCT imaging was performed at each examination, and a masked grader evaluated all OCT scans from both eyes of each infant for visibility (yes/no) of all retinal layers and of the chorio-scleral junction (CSJ). Multivariable logistic regression was used to assess associations between fundus pigmentation and visibility of all retinal layers and CSJ, controlling for potential confounders (i.e., birth weight, gestational age, sex, OCT system, pupil size, and postmenstrual age at imaging). Results: Among 114 infants (mean birth weight, 943 grams; mean gestational age, 27.6 weeks), 43 infants (38%) had blond, 56 infants (49%) had medium, and 15 infants (13%) had dark fundus pigmentation. Of 1042 scans, all retinal layers were visible in 977 (94%) and CSJ in 895 (86%). Pigmentation was not associated with retinal layer visibility (P = 0.49), but medium and dark pigmentation were associated with decreased CSJ visibility (medium: odds ratio [OR] = 0.34, P = 0.001; dark: OR = 0.24, P = 0.009). For infants with dark pigmentation, retinal layer visibility increased (OR = 1.87 per week; P ≤ 0.001) and CSJ visibility decreased (OR = 0.78 per week; P = 0.01) with increasing age. Conclusions: Although fundus pigmentation was not associated with the visibility of all retinal layers on OCT, darker pigmentation decreased CSJ visibility, and this effect increased with age. Translational Relevance: The ability of bedside OCT to capture retinal layer microanatomy in preterm infants, regardless of fundus pigmentation, may represent an advantage over fundus photography for ROP telemedicine.

Multivariate Models to Diagnose Early Referral-Warranted Retinopathy of Prematurity With Handheld Optical Coherence Tomography.

Purpose: The purpose of this study was to create multivariate models predicting early referral-warranted retinopathy of prematurity (ROP) using non-contact handheld spectral-domain optical coherence tomography (OCT) and demographic data. Methods: Between July 2015 and February 2018, infants ≤1500 grams birth weight or ≤30 weeks gestational age from 2 academic neonatal intensive care units were eligible for this study. Infants were excluded if they were too unstable to participate in ophthalmologic examination (2), had inadequate image quality (20), or received prior ROP treatment (2). Multivariate models were created using demographic variables and imaging findings to identify early referral-warranted ROP (referral-warranted ROP and/or pre-plus disease) by routine indirect ophthalmoscopy. Results: A total of 167 imaging sessions of 71 infants (45% male infants, gestational age 28.2+/-2.8 weeks, and birth weight 995.6+/-292.0 grams) were included. Twelve of 71 infants (17%) developed early referral-warranted ROP. The area under the receiver operating characteristic curve (AUC) was 0.94 for the generalized linear mixed model (sensitivity = 95.5% and specificity = 80.7%) and 0.83 for the machine learning model (sensitivity = 91.7% and specificity = 77.8%). The strongest variables in both models were birth weight, image-based Vitreous Opacity Ratio (an estimate of opacity density), vessel elevation, and hyporeflective vessels. A model using only birth weight and gestational age yielded an AUC of 0.68 (sensitivity = 77.3% and specificity = 63.4%), and a model using only imaging biomarkers yielded 0.88 (sensitivity = 81.8% and specificity = 84.8%). Conclusions: A generalized linear mixed model containing handheld OCT biomarkers can identify early referral-warranted ROP. Machine learning produced a less optimal model. Translational Relevance: With further validation, this work may lead to a better-tolerated ROP screening tool.

Retinopathy of prematurity detection: a retrospective quality improvement project before-after implementation of retinal digital imaging for screening.

Screening of retinopathy of prematurity (ROP) was modified in a level-3 neonatal intensive care unit by the introduction of a wide-field retinal imaging. The aim of this study was to evaluate whether retinopathy of prematurity (ROP) diagnosis was improved or not compared to previously used binocular indirect ophthalmoscopy (BIO). This was a retrospective, uncontrolled, quality improvement project. Records of consecutive premature newborns screened for ROP over two 1-year periods were reviewed. Systemic factors potentially influencing the occurrence of ROP were investigated using uni- and multivariable linear regression followed by stepwise forward regression. ROP screening was performed by ophthalmologists using BIO in 2014, and digital wide-field retinal imaging (Panocam™ pro) in 2019. Records of N = 297 patients were analyzed (N = 159 in 2014 and N = 138 in 2019). The proportion of ROP diagnosed at any stage, over the total number of neonates screened, was significantly higher in 2019 (n = 46/138, 33.1%) compared to 2014 (n = 11/159, 6.9%) (p < 0.0001). Most neonates presented with mild forms of ROP during both 1-year periods analyzed. After adjustment for all parameters influencing ROP occurrence, the variables contributing independently to the diagnosis of any stage of ROP were birth weight (p = 0.002), duration of mechanical ventilation (p = 0.028) and wide-field fundus camera-assisted screening (p < 0.001). CONCLUSION: After adjusting for many recognized systemic factors influencing the development of ROP, screening by wide-field digital retinal imaging was independently associated with higher ROP detection. WHAT IS KNOWN: • No consensus has been reached to replace binocular indirect ophthalmoscopy by retinal imaging for ROP screening. • Diagnostic accuracy and high sensitivity and specificity has been reported for wide-field digital imaging. WHAT IS NEW: • The introduction of wide-field imaging for ROP screening in at level-3 reference center was independently associated to higher ROP detection.

Effective field of view of wide-field fundus photography in the Stanford University Network for Diagnosis of Retinopathy of Prematurity (SUNDROP).

Five-field 130° wide-angle imaging is the standard of care for retinopathy of prematurity (ROP) screening with an ideal hypothetical composite field-of-view (FOV) of 180°. We hypothesized that in many real-world scenarios the effective composite FOV is considerably less than ideal. This observational retrospective study analyzed the effective FOV of fundus photos of patients screened for ROP as part of the Stanford University Network for Diagnosis of Retinopathy of Prematurity (SUNDROP) initiative. Five fundus photos were selected from each eye per image session. Effective FOV was defined as the largest circular area centered on the optic disc that encompassed retina in each of the four cardinal views. Seventy-three subjects were analyzed, 35 without ROP and 34 with ROP. Mean effective FOV was 144.55 ± 6.62° ranging from 130.00 to 153.71°. Effective FOV was not correlated with the presence or absence of ROP, gestational age, birth weight, or postmenstrual age. Mean effective FOV was wider in males compared to females. Standard five-field 130° fundus photos yielded an average effective FOV of 144.54° in the SUNDROP cohort. This implies that an imaging FOV during ROP screening considerably less than the hypothetical ideal of 180° is sufficient for detecting treatment warranted ROP.

Telemedicina en retinopatía del prematuro: Atravesando fronteras en la salud visual pediátrica. Estudio TELEROP / Telemedicine in retinopathy of prematurity: Crossing borders in pediatric vision health. TELEROP study

Objetivo general Evaluar la precisión y validez de imágenes vía teléfono inteligente en comparación con el sistema RetCam® para el diagnóstico de retinopatía del prematuro (ROP). Metodología Estudio observacional, de corte longitudinal y enmascarado realizado en el hospital Dr. Elías Santana. Se incluyeron infantes con peso al nacer ≤ 1.500 g, edad gestacional ≤ 30 semanas y/o pacientes expuestos a factores de riesgo o complicaciones ligadas a ROP. Estos sujetos fueron tamizados empleando imágenes vía teléfono inteligente o RetCam®, ambos comparados con la fondoscopía convencional. Se utilizó la clasificación de ICROP para la estadificación. Los resultados principales analizados fueron sensibilidad, especificidad, valores predictivos e índice kappa. Resultados Se obtuvieron 915 imágenes (n = 121), distribuidas en grupo teléfono inteligente (50,4%) y grupo RetCam® (49,6%) entre agosto del 2020 y marzo del 2021. Los sujetos que presentaron ROP tuvieron menor edad gestacional (30,2 sem ± 2.8), peso al nacer (1.361 g ± 398) y mayor exposición a oxigenoterapia (12,8 días ± 11,3). El grupo RetCam® presentó sensibilidad = 80%, especificidad = 78%, valor predictivo positivo = 90% e índice kappa = 0,70. El grupo teléfono inteligente presentó sensibilidad = 88%, especificidad = 90%, valor predictivo positivo = 93,75% e índice kappa = 0,81. Conclusiones Ambos métodos diagnósticos fueron precisos para identificar ROP. El grupo teléfono inteligente obtuvo resultados superiores con una excelente resolución, representando un método costo-efectivo para crear un impacto global en la reducción de la ceguera prevenible en población pediátrica (AU)

General objective To evaluate the accuracy and validity of images with smartphone compared to the RetCam® system for the diagnosis of retinopathy of prematurity (ROP). Methodology Observational, longitudinal and masked study carried out at the Dr. Elías Santana hospital. Infants with birth weight ≤ 1500 g, gestational age ≤ 30 weeks and/or patients exposed to risk factors or complications linked to ROP were included. These subjects were screened using images with smartphone or RetCam®, both compared to conventional fundoscopy. The ICROP classification was used for staging. The main results analyzed were sensitivity, specificity, positive predictive values and kappa index. Results 915 images (n = 121) were obtained, distributed in smartphone group (50.4%) and RetCam® group (49.6%) between August 2020 and March 2021. Subjects with ROP had lower gestational age (30.2 sem ± 2.8), birth weight (1361 g ± 398), and greater exposure to oxygen therapy (12.8 days ± 11.3). The RetCam® group presented sensitivity = 80%, specificity = 78%, positive predictive value = 90% and kappa index = 0.70. The smartphone group presented sensitivity = 88%, specificity = 90%, positive predictive value = 93.75% and kappa index = 0.81. Conclusions Both diagnostic methods were accurate to identify ROP. The smartphone group obtained superior results with excellent resolution, representing a cost-effective method to create a global impact on reducing preventable blindness in the pediatric population (AU)

Evaluation of Ocular Biometric and Optical Coherence Tomography Parameters in Preterm Children Without Retinopathy of Prematurity.

Purpose: To evaluate and compare biometric and optical coherence tomography parameters of ocular structures in preterm children without retinopathy of prematurity with term children. Methods: A cross-sectional, comparative study was carried out from 2018 to 2019. In this study, 124 eyes of 62 preterm children were compared with 132 eyes of 66 term children aged between 7 and 9 years. Preterm children were born at 28 to 32 weeks with a birth weight of less than 2 kg with no ocular abnormalities, and term children were delivered at 37 or greater weeks and had a birth weight of 2 kg or more. All children had standardized eye examinations, and ocular measurements using the anterior and posterior segment optical coherence tomography and laser interferometry. Results: Significant differences were found between the term and preterm children for horizontal corneal diameter: median, 12.2 mm (interquartile range [IQR], 0.4) versus median, 12.1 mm (IQR, 0.6; P < 0.005); axial length median, 23.03 mm (IQR, 1.10 mm) versus median, 22.88 mm (IQR, 1.35 mm; P = 0.017); global retinal nerve fiber layer thickness: mean ± standard deviation, 106.54 ± 10.23 µm versus mean ± standard deviation, 103.65 ± 10.178 µm (P = 0.024); temporal retinal nerve fiber layer thickness: median, 76 µm (IQR, 16 µm) vs median, 74 µm (IQR, 14 µm; P = 0.012); and the angle opening distance at 750 µm nasal: mean ± standard deviation, 0.815 ± 0.23 mm vs mean ± standard deviation, 0.749 ± 0.21 mm (P = 0.016). No significant differences were found for other anterior segment and angle parameters. Conclusions: Preterm children with no retinopathy of prematurity have smaller eyes and thinner retinal nerve fiber layers than their term counterparts. The long-term effects of interrupted ocular growth in preterm children should be further studied into adulthood. Translational Relevance: Preterm children maybe more predisposed to certain eye conditions because they have smaller eyes, and thus should be further monitored clinically.

MHANet: A hybrid attention mechanism for retinal diseases classification.

With the increase of patients with retinopathy, retinopathy recognition has become a research hotspot. In this article, we describe the etiology and symptoms of three kinds of retinal diseases, including drusen(DRUSEN), choroidal neovascularization(CNV) and diabetic macular edema(DME). In addition, we also propose a hybrid attention mechanism to classify and recognize different types of retinopathy images. In particular, the hybrid attention mechanism proposed in this paper includes parallel spatial attention mechanism and channel attention mechanism. It can extract the key features in the channel dimension and spatial dimension of retinopathy images, and reduce the negative impact of background information on classification results. The experimental results show that the hybrid attention mechanism proposed in this paper can better assist the network to focus on extracting thr fetures of the retinopathy area and enhance the adaptability to the differences of different data sets. Finally, the hybrid attention mechanism achieved 96.5% and 99.76% classification accuracy on two public OCT data sets of retinopathy, respectively.

Development of 3D Printed Smartphone-Based Multi-Purpose Fundus Camera (MultiScope) for Retinopathy of Prematurity.

Retinopathy of Prematurity (ROP) is a prominent source of low vision and blindness in preterm babies. Wide-Field Digital Retinal Imaging (WFDRI) systems acquire accurate digital images which are very useful for identification, documentation and transmitting the various retinal diseases. This telemedicine technique has potential for an alternative tool for Binocular Indirect Ophthalmoscopy (BIO) in ROP screening, but it is very expensive and accessibility for poor communities is limited. Capabilities of good illumination, high resolution camera and processing speed of the modern smartphones are being identified as a substitute. Potential applications of 3D printing is that it provides a severe impact in medical field, especially in ophthalmology sector. Competences of 3D printing are very useful for the development of retinal camera from any smartphone with the help of 3D printable devices. The primary aim of this study is to develop a handheld 3D printed smartphone-based multi-purpose fundus camera for ROP screening. The secondary aim is to check the feasibility and compare the digital fundus images obtained from the developed fundus camera against the commercial RetCam imaging. The proposed cost effective and remote reading device is an alternative to WFDRI for ROP screening and can improve the potential of ROP care for low resource communities.

Comparison of heads up three dimensional visualization system to conventional microscope in retinopathy of prematurity related tractional retinal detachment.

To report the outcomes, advantages and disadvantages of a heads-up three-dimensional (3D) visualization system compared to the conventional microscope in pediatric tractional retinal detachment (TRD) surgery secondary to advanced stage retinopathy of prematurity (ROP). Medical records of patients with ROP stage 4 or 5 who underwent surgery for tractional retinal detachment at King Khaled Eye Specialist Hospital between September 2017 and July 2019 were identified and reviewed. Eyes were divided into 2 groups, eyes that underwent surgery with a 3D heads-up platform (3D group) and eyes that underwent surgery with a conventional microscope (conventional group). Data were collected on neonatal history, visual acuity, intraoperative complications and success rates between groups.Eighteen eyes of 14 patients who underwent surgical repair of TRD related to ROP. Postoperative outcomes were compared between 10 eyes (7 patients) in the 3D group and 8 eyes (7 patients) in the conventional group There was no statistically significant difference in success rate between both groups (75% conventional group vs 70% 3D group). Partial or complete reattachment was achieved in 7 eyes in 3D group compared to 6 eyes in conventional group. Lower postmenstrual age at the time of the first surgery and presence of retinal breaks were associated with poorer surgical outcome. Heads up 3D visualization system is feasible in tractional retinal detachment related to ROP with similar success rate and no increased risk of complications when compared to conventional microscope. This system may be advantageous in advanced pediatric tractional retinal detachment surgeries.

Assistive Framework for Automatic Detection of All the Zones in Retinopathy of Prematurity Using Deep Learning.

Retinopathy of prematurity (ROP) is a potentially blinding disorder seen in low birth weight preterm infants. In India, the burden of ROP is high, with nearly 200,000 premature infants at risk. Early detection through screening and treatment can prevent this blindness. The automatic screening systems developed so far can detect "severe ROP" or "plus disease," but this information does not help schedule follow-up. Identifying vascularized retinal zones and detecting the ROP stage is essential for follow-up or discharge from screening. There is no automatic system to assist these crucial decisions to the best of the authors' knowledge. The low contrast of images, incompletely developed vessels, macular structure, and lack of public data sets are a few challenges in creating such a system. In this paper, a novel method using an ensemble of "U-Network" and "Circle Hough Transform" is developed to detect zones I, II, and III from retinal images in which macula is not developed. The model developed is generic and trained on mixed images of different sizes. It detects zones in images of variable sizes captured by two different imaging systems with an accuracy of 98%. All images of the test set (including the low-quality images) are considered. The time taken for training was only 14 min, and a single image was tested in 30 ms. The present study can help medical experts interpret retinal vascular status correctly and reduce subjective variation in diagnosis.

Dome-shaped macula in premature infants visualized by handheld spectral-domain optical coherence tomography.

PURPOSE: To describe dome-shaped macula and associated clinical findings in premature infants. METHODS: This prospective, observational cohort study included a consecutive sample of premature infants screened for retinopathy of prematurity (ROP) with 9-month follow-up. Handheld spectral domain optical coherence tomography (SD-OCT) was performed at the time of ROP screening. Images were assessed for dome-shaped macula, cystoid macular edema, epiretinal membrane, vitreous bands, and punctate hyperreflective vitreous opacities. Dome height measurements were performed in a subset of images. Teller visual acuity and cycloplegic refraction were performed at an adjusted age of 8-10 months. RESULTS: Of 37 infants (74 eyes; 49% male; mean gestational age 27.8 ± 3.2 weeks; mean birth weight 949 ± 284 g), 24/37 (65%) demonstrated dome-shaped macula in at least one eye (13 both eyes, 5 right eye only, and 6 left eye only). Of the 74 eyes, 26 (35%) could be reliably measured, with a mean dome height of 139.0 ± 72.3 µm (range, 54-369 µm). Presence of dome-shaped macula was associated with a diagnosis of ROP (P = 0.02; OR, 3.03; 95% CI, 1.18-7.82) and pre-plus or plus disease (P = 0.02; OR, 4.20; 95% CI, 1.05-16.78). Infants with dome-shaped macula had lower birth weight compared with those without (877 vs 1081 g; P = 0.04). No associations with other demographics, OCT findings, and 9-month refractive outcomes were found. CONCLUSIONS: Dome-shaped macula was frequently identified by handheld SD-OCT in premature infants, especially those with lower birth weight and severe ROP. The long-term clinical significance of this finding is unknown.

Key factors in a rigorous longitudinal image-based assessment of retinopathy of prematurity.

To describe a database of longitudinally graded telemedicine retinal images to be used as a comparator for future studies assessing grader recall bias and ability to detect typical progression (e.g. International Classification of Retinopathy of Prematurity (ICROP) stages) as well as incremental changes in retinopathy of prematurity (ROP). Cohort comprised of retinal images from 84 eyes of 42 patients who were sequentially screened for ROP over 6 consecutive weeks in a telemedicine program and then followed to vascular maturation or treatment, and then disease stabilization. De-identified retinal images across the 6 weekly exams (2520 total images) were graded by an ROP expert based on whether ROP had improved, worsened, or stayed the same compared to the prior week's images, corresponding to an overall clinical "gestalt" score. Subsequently, we examined which parameters might have influenced the examiner's ability to detect longitudinal change; images were graded by the same ROP expert by image view (central, inferior, nasal, superior, temporal) and by retinal components (vascular tortuosity, vascular dilation, stage, hemorrhage, vessel growth), again determining if each particular retinal component or ROP in each image view had improved, worsened, or stayed the same compared to the prior week's images. Agreement between gestalt scores and view, component, and component by view scores was assessed using percent agreement, absolute agreement, and Cohen's weighted kappa statistic to determine if any of the hypothesized image features correlated with the ability to predict ROP disease trajectory in patients. The central view showed substantial agreement with gestalt scores (κ = 0.63), with moderate agreement in the remaining views. Of retinal components, vascular tortuosity showed the most overall agreement with gestalt (κ = 0.42-0.61), with only slight to fair agreement for all other components. This is a well-defined ROP database graded by one expert in a real-world setting in a masked fashion that correlated with the actual (remote in time) exams and known outcomes. This provides a foundation for subsequent study of telemedicine's ability to longitudinally assess ROP disease trajectory, as well as for potential artificial intelligence approaches to retinal image grading, in order to expand patient access to timely, accurate ROP screening.

Automatic Staging for Retinopathy of Prematurity With Deep Feature Fusion and Ordinal Classification Strategy.

Retinopathy of prematurity (ROP) is a retinal disease which frequently occurs in premature babies with low birth weight and is considered as one of the major preventable causes of childhood blindness. Although automatic and semi-automatic diagnoses of ROP based on fundus image have been researched, most of the previous studies focused on plus disease detection and ROP screening. There are few studies focusing on ROP staging, which is important for the severity evaluation of the disease. To be consistent with clinical 5-level ROP staging, a novel and effective deep neural network based 5-level ROP staging network is proposed, which consists of multi-stream based parallel feature extractor, concatenation based deep feature fuser and clinical practice based ordinal classifier. First, the three-stream parallel framework including ResNet18, DenseNet121 and EfficientNetB2 is proposed as the feature extractor, which can extract rich and diverse high-level features. Second, the features from three streams are deeply fused by concatenation and convolution to generate a more effective and comprehensive feature. Finally, in the classification stage, an ordinal classification strategy is adopted, which can effectively improve the ROP staging performance. The proposed ROP staging network was evaluated with per-image and per-examination strategies. For per-image ROP staging, the proposed method was evaluated on 635 retinal fundus images from 196 examinations, including 303 Normal, 26 Stage 1, 127 Stage 2, 106 Stage 3, 61 Stage 4 and 12 Stage 5, which achieves 0.9055 for weighted recall, 0.9092 for weighted precision, 0.9043 for weighted F1 score, 0.9827 for accuracy with 1 (ACC1) and 0.9786 for Kappa, respectively. While for per-examination ROP staging, 1173 examinations with a 4-fold cross validation strategy were used to evaluate the effectiveness of the proposed method, which prove the validity and advantage of the proposed method.

Avascular island in the vascularised retina in zone II ROP.

We describe an atypical form of avascular island demarcated by a ridge in the vascularised retina of retinopathy of prematurity. Fundus fluorescein angiography showed an unusual finding of an avascular island demarcated by shunt vessels in vascularised retina posterior to the ridge. There is a need to document this finding as this could throw some light on the aetiopathogenesis of the disease.

Early angiographic signs of retinopathy of prematurity requiring treatment.

BACKGROUND/OBJECTIVES: Fluorescein angiography (FA) has been a pivotal tool to study the pathophysiology of retinopathy of prematurity (ROP) in vivo. We examined the course of ROP using FA in order to assess the predictive value of angiographic features. SUBJECTS/METHODS: This is an observational retrospective cohort multi-center study of eyes screened for ROP with binocular indirect ophthalmoscope and with FA. All infants undergoing screening examination for ROP who had retinal vasculature limited to Zone I and posterior Zone II vascularization underwent FA between 31 and 34 weeks postmenstrual age. RetCam fundus imaging and video digital fluorescein angiography were performed in the neonatal intensive care units. Masked grading of the FA images was retrospectively conducted by two ROP expert ophthalmologists. Ten criteria that describe retinovascular and choroidal features on FA were used to assess their predictive value for development of treatment-requiring ROP. RESULTS: A total of 98 eyes of 56 patients were included for this study. FAs of eyes of premature infants show a wide range of features either at the junction between the vascular and avascular retina and posteriorly to that. Among the angiographic features evaluated, leakage, shunts and hyperfluorescent lesions at the junction between vascular and avascular zone were predictive of the development of treatment-requiring ROP (p < 0.05), but findings in the posterior vascularized retina were not. CONCLUSIONS: FA can add to our understanding of the evolution of vascular abnormalities in the course of ROP and can help predict which eyes will go on to treatment.