Evaluation of an Artificial Intelligence System for Retinopathy of Prematurity Screening in Nepal and Mongolia.

Purpose: To evaluate the performance of a deep learning (DL) algorithm for retinopathy of prematurity (ROP) screening in Nepal and Mongolia. Design: Retrospective analysis of prospectively collected clinical data. Participants: Clinical information and fundus images were obtained from infants in 2 ROP screening programs in Nepal and Mongolia. Methods: Fundus images were obtained using the Forus 3nethra neo (Forus Health) in Nepal and the RetCam Portable (Natus Medical, Inc.) in Mongolia. The overall severity of ROP was determined from the medical record using the International Classification of ROP (ICROP). The presence of plus disease was determined independently in each image using a reference standard diagnosis. The Imaging and Informatics for ROP (i-ROP) DL algorithm was trained on images from the RetCam to classify plus disease and to assign a vascular severity score (VSS) from 1 through 9. Main Outcome Measures: Area under the receiver operating characteristic curve and area under the precision-recall curve for the presence of plus disease or type 1 ROP and association between VSS and ICROP disease category. Results: The prevalence of type 1 ROP was found to be higher in Mongolia (14.0%) than in Nepal (2.2%; P < 0.001) in these data sets. In Mongolia (RetCam images), the area under the receiver operating characteristic curve for examination-level plus disease detection was 0.968, and the area under the precision-recall curve was 0.823. In Nepal (Forus images), these values were 0.999 and 0.993, respectively. The ROP VSS was associated with ICROP classification in both datasets (P < 0.001). At the population level, the median VSS was found to be higher in Mongolia (2.7; interquartile range [IQR], 1.3-5.4]) as compared with Nepal (1.9; IQR, 1.2-3.4; P < 0.001). Conclusions: These data provide preliminary evidence of the effectiveness of the i-ROP DL algorithm for ROP screening in neonatal populations in Nepal and Mongolia using multiple camera systems and are useful for consideration in future clinical implementation of artificial intelligence-based ROP screening in low- and middle-income countries.

Retinal ultra-wide-field colour imaging versus dilated fundus examination to screen for sickle cell retinopathy.

PURPOSE: To compare ultra-wide-field colour fundus imaging (UWFI) to dilated fundus examination (DFE) for the screening of sickle cell retinopathy (SCR). DESIGN: This study is a prospective, blinded, multicentre case series. PARTICIPANTS: This study included two groups: an adult group (n=268 eyes) and a paediatric group (n=168 eyes). Sickle cell disease (SCD) types included haemoglobin S homozygous (HbSS), haemoglobin S and C (HbSC) and Hb S with ß-thalassaemia (HbSß-Thal). METHODS: Participants underwent DFE and UWFI. Each eye received three independent grades (1-4), documented by three graders: clinical grader, image grader 1 and image grader 2. Three clinically relevant diagnostic thresholds were determined. Based on these thresholds, the sensitivity, specificity, positive predictive value and negative predictive value for all three graders were calculated relative to each other as reference tests. RESULTS: HbSC was associated with the most advanced SCR grades. When compared to the clinical grader, image grader 1 and image grader 2 consistently detected more SCR and higher SCR grades in both adult and paediatric groups. In both groups, image grader 1 and image grader 2 identified twice as many cases of capillary occlusion/anastomosis than clinical grader. To detect the presence of any proliferative SCR, image grader 1 and image grader 2 had a sensitivity of 82%, 71% in the paediatrics group and 90% and 72% in the adult group. The clinical grader sensitivity was 52% in the paediatrics group and 53% in the adult group. CONCLUSION: The UWFI is a sensitive tool to screen for SCR. It is superior to DFE in detecting capillary occlusion or anastomosis.

Development of Screening Criteria for Retinopathy of Prematurity in Ulaanbaatar, Mongolia, Using a Web-based Data Management System.

PURPOSE: To describe a process for identifying birth weight (BW) and gestational age (GA) screening guidelines in Mongolia. METHODS: This was a prospective cohort study in a tertiary care hospital in Ulaanbataar, Mongolia, of 193 premature infants with GA of 36 weeks or younger and/or BW of 2,000 g or less) with regression analysis to determine associations between BW and GA and the development of retinopathy of prematurity (ROP). RESULTS: As BW and GA decreased, the relative risk of developing ROP increased. The relative risk of developing any stage of ROP in infants born at 29 weeks or younger was 2.91 (95% CI: 1.55 to 5.44; P < .001] compared to older infants. The relative risk of developing any type of ROP in infants with BW of less than 1,200 g was 2.41 (95% CI: 1.35 to 4.29; P = .003] and developing type 2 or worse ROP was 2.05 (95% CI: 0.99 to 4.25; P = .05). CONCLUSIONS: Infants in Mongolia with heavier BW and older GA who fall outside of current United States screening guidelines of GA of 30 weeks or younger and/or BW of 1,500 g or less developed clinically relevant ROP. [J Pediatr Ophthalmol Strabismus. 2020;57(5):333-339.].

Secondary Prevention of Retinoblastoma Revisited: Laser Photocoagulation of Invisible New Retinoblastoma.

PURPOSE: Invisible retinoblastoma tumors are now detected with screening for retinal tumors in at-risk neonates (those inheriting RB1 pathogenic alleles from affected parents) using handheld OCT. Laser photocoagulation is challenging, requiring exact localization of a tumor invisible to indirect ophthalmoscopy and standard imaging. We describe OCT-guided localization and photocoagulation of these invisible tumors with 1-year follow-up. DESIGN: Retrospective, noncomparative, single-institutional, observational case series. PARTICIPANTS: Children with any clinically invisible retinoblastoma tumor that was detected on OCT posterior pole screening. METHODS: OCT revealed round homogeneous invisible tumors within the inner nuclear layer. Software calipers placed beside anatomic retinal landmarks (branched/curved vessels, fovea, or optic disc) mapped the tumor location and extent. A single laser (532 nm) burn flagged the location, and OCT evaluated the tumor-laser burn relationship; laser treatment was then continued in the correct location. Post-laser OCT ensured complete treatment. MAIN OUTCOME MEASURES: Accuracy (frequency of geographic miss and skip areas), effectiveness (recurrence rate), and burden (scar size and characteristics at final follow-up) of laser treatment. RESULTS: Eleven new invisible posterior pole tumors in 7 eyes of 5 children were treated by this technique. Localization and tumor-laser burn relationships were accurate in 11 of 11 tumors (100%, 95% confidence interval [CI], 49.9-100), and all showed swelling and hyper-reflectiveness of the tumor in post-laser OCT. Two photocoagulation sessions (2 weeks apart) were sufficient to successfully manage 9 of 11 tumors (82%, 95% CI, 37.4-100) with resulting permanent flat scars. One tumor (9%, 95% CI, 0.2-50.6) developed OCT-detected subclinical recurrences within 3 months, treated by 1 laser session. No treatment scar showed gliosis, foveal involvement, or retinal traction at 1-year follow-up. Scar expansion occurred in 1 tumor (9%, 95% CI, 0.2-50.6), and all scars (100%, 95% CI, 49.9-100) showed pigmentary changes. CONCLUSIONS: The OCT-guided localization and photocoagulation technique is valuable in achieving precision results in managing invisible new retinoblastoma tumors. This technique shows a potential to improve outcomes of secondary prevention screening for retinoblastoma.

Fluorescein Angiography Findings in Children With Congenital Zika Syndrome.

BACKGROUND AND OBJECTIVE: To evaluate the retinal and vasculature changes in infants with congenital Zika syndrome (CZS) using fluorescein angiography (FA). PATIENTS AND METHODS: This consecutive case series included six infants with CZS. FA and color fundus imaging were performed under general anesthesia in both eyes of all infants using a contact widefield digital imaging system. All color fundus images were obtained using a 130° field of view lens, and the FA images were captured using either a 130° or 80° field of view lens. The immunoglobulin M antibody capture enzyme-linked immunosorbent assay was positive for Zika virus in the cerebrospinal fluid samples of all infants. Other congenital infections were ruled out. RESULTS: The mean ± standard deviation age of the infants at the time of examination was 1.4 years ± 0.1 years (range: 1.3 years to 1.5 years). Contact fundus photographs showed macular abnormalities in seven eyes (58%) and retinal vasculature changes in two eyes (17%). FA detected macular abnormalities in all 12 eyes (100%) and retinal vasculature changes in five eyes (42%). The main retinal vasculature changes were peripheral avascularity in five eyes (42%) and microvasculature abnormalities in three eyes (25%). CONCLUSION: FA may be an important tool for detecting subtle macular and retinal vasculature changes in CZS. [Ophthalmic Surg Lasers Imaging Retina. 2019;50:702-708.].

Vision and visual potential for perifoveal retinoblastoma after optical coherence tomographic-guided sequential laser photocoagulation.

BACKGROUND/AIMS: To assess tumour control, vision and anatomical visual potential in eyes with perifoveal retinoblastoma treated by sequential photocoagulation from the antifoveal tumour edge inwards, avoiding treatment near the fovea. Patients were monitored for tumour control, foveal and perifoveal anatomy at each treatment session by optical coherence tomography (OCT) and treated for amblyopia when the other eye had better vision. METHODS: Eyes with perifoveal retinoblastoma treated between 1 January 2011 and 31 May 2017 with laser therapy after chemotherapy for juxtafoveal (fovea clear of tumour but <3000 µm from tumour edge) or foveolar retinoblastoma (tumour underlying fovea) were retrospectively reviewed for tumour control without recurrence, anatomical success (foveal pit preservation and/or restoration with ≥500 µm perifoveal retina free of tumour and scar) and functional success (acceptable (>0.1 decimal) or good (>0.3 decimal) visual acuity (VA)). RESULTS: Twenty-two eyes (14 juxtafoveal, 8 foveolar tumours) of 20 patients (19 bilateral, 1 familial and 11 females) were included. No juxtafoveal tumour had tumour recurrence, and 13/14 patients showed foveal pit preservation with ≥500 µm of perifoveal retina tumour free. Foveolar tumours had significant worse anatomical outcomes: failure to restore foveal pit or perifoveal retina (8/8, p=0.001) and tumour recurrences (5/8, p=0.001). Functional success with acceptable VA was achieved in 12/14 juxtafoveal and 5/8 foveal tumours eyes (p=0.01). Amblyopia therapy data were insufficient to evaluate impact on VA. CONCLUSIONS: Anatomical visual potential and functional vision were better in juxtafoveal than foveolar retinoblastoma treated with foveal-sparing laser photocoagulation guided by OCT. The role of amblyopia therapy requires a prospective study.

Optical Coherence Tomography-Guided Decisions in Retinoblastoma Management.

PURPOSE: Assess the role of handheld optical coherence tomography (OCT) in guiding management decisions during diagnosis, treatment, and follow-up of eyes affected by retinoblastoma. DESIGN: Retrospective, noncomparative, single-institution case series. PARTICIPANTS: All children newly diagnosed with retinoblastoma from January 2011 to December 2015 who had an OCT session during their active treatment at The Hospital for Sick Children (SickKids) in Toronto, Canada. The OCT sessions for fellow eyes of unilateral retinoblastoma without any suspicious lesion and those performed more than 6 months after the last treatment were excluded. METHODS: Data collected included age at presentation, sex, family history, RB1 mutation status, 8th edition TNMH cancer staging and International Intraocular Retinoblastoma Classification (IIRC), and number of OCT sessions per eye. Details of each session were scored for indication-related details (informative or not) and assessed for guidance (directive or not), diagnosis (staging changed, new tumors found or excluded), treatment (modified, stopped, or modality shifted), or follow-up modified. MAIN OUTCOME MEASURES: Frequency of OCT-guided management decisions, stratified by indication and type of guidance (confirmatory vs. influential). RESULTS: Sixty-three eyes of 44 children had 339 OCT sessions over the course of clinical management (median number of OCT scans per eye, 5; range, 1-15). The age at presentation and presence of a heritable RB1 mutation significantly correlated with an increased number of OCT sessions. Indications included evaluation of post-treatment scar (55%) or fovea (16%), and posterior pole scanning for new tumors (11%). Of all sessions, 92% (312/339) were informative; 19 of 27 noninformative sessions had large, elevated lesions; of these, 14 of 19 were T2a or T2b (IIRC group C or D) eyes. In 94% (293/312) of the informative sessions, OCT directed treatment decisions (58%), diagnosis (16%), and follow-up (26%). Optical coherence tomography influenced and changed management from pre-OCT clinical plans in 15% of all OCT sessions and 17% of directive sessions. CONCLUSIONS: Optical coherence tomography improves the accuracy of clinical evaluation in retinoblastoma management.

Assessment of change of optic nerve head cupping in pediatric glaucoma using the RetCam 120.

PURPOSE: To ascertain the interobserver agreement among various specialists when analyzing images of the optic nerve head taken with the RetCam 120 (Clarity Medical Systems, Inc., Pleasanton, CA) for assessing the progression of optic disk cupping in pediatric glaucoma. METHODS: Five observers compared pairs of RetCam 120 two-dimensional photographs from 64 eyes to identify whether or not changes in the optic disk had occurred. Observers included a pediatric glaucoma specialist, pediatric neurophthalmologist, strabismologist, pediatric ophthalmic imaging specialist, and pediatric ophthalmology fellow. Each was masked to the patient identity and clinical course. For each patient, the observers reviewed the photographs taken prior to therapy and at a minimum of 6 months following initiation of therapy. Interobserver variability was calculated and analysis conducted to identify influential variables. RESULTS: Poor agreement was noted between the observers. No significant association was obtained between agreement level and variables affecting photographic quality: variability of contrast and illumination, exposure and magnification, image clarity, presence of fluorescein on the cornea, and pixelation of the image with poor resolution. Raters noted problems with these image variables in 75.6% of the observations. The Kappa statistic obtained was kappa = 0.29 with overall proportion of agreement of p = 0.69, suggesting a fair agreement but not a moderate agreement. CONCLUSIONS: Comparison of RetCam 120 pediatric optic nerve head photographs, when performed subjectively by observers with different perspectives, and in isolation from clinical information, may not be a reliable indicator of cupping change. In addition, variations in image contrast, luminance, color, and pixelation pose significant challenges to agreement between observers. Comparison of RetCam 120 images should not be considered the sole criterion for monitoring pediatric glaucoma.

Characteristic retinal atrophy with secondary "inverse" optic atrophy identifies vigabatrin toxicity in children.

OBJECTIVE: To describe the clinical pattern of retinal atrophy in children caused by the anticonvulsant vigabatrin. DESIGN: An interventional case series report. PARTICIPANTS: One hundred thirty-eight patients, mainly infants, were evaluated regularly for evidence of possible vigabatrin toxicity in the Eye and Neurology clinics at the Hospital for Sick Children, Toronto. METHOD: Sequential clinical and electroretinographic (International Society for Clinical Electrophysiology of Vision standards) evaluations every 6 months. MAIN OUTCOME MEASURES: Presence of recognizable retinal and optic atrophy in the presence of abnormal electroretinogram (ERG) and other clinical findings. RESULTS: Three children being treated for seizures with vigabatrin showed definite clinical findings of peripheral retinal nerve fiber layer atrophy, with relative sparing of the central or macular portion of the retina and relative nasal optic nerve atrophic changes. Some macular wrinkling was evident in 1 case. Progressive ERG changes showing decreased responses, especially the 30-Hz flicker response, supported the presence of decreased retinal function. CONCLUSIONS: A recognizable and characteristic form of peripheral retinal atrophy and nasal or "inverse" optic disc atrophy can occur in a small number of children being treated with vigabatrin. The changes in superficial light reflexes of the retina in children facilitate the clinical recognition of nerve fiber layer atrophy. The macula is relatively spared, although superficial retinal light reflexes indicating wrinkling of the innermost retina suggest early macular toxicity as well. Because these changes are accompanied by electrophysiologic evidence of retinal dysfunction, discontinuation of vigabatrin should be strongly considered.

Assessment of extraocular muscles position and anatomy by 3-dimensional ultrasonography: a trial in craniosynostosis patients.

PURPOSE: To determine whether 3-dimensional ultrasonography (3D US) provides information about anatomy and position of extraocular muscles to better guide surgeons approaching strabismus in patients with craniosynostosis who often have anomalous or absent eye muscles. METHODS: The 4 rectus eye muscles were imaged using 3D US for 7 children with craniosynostosis before or after strabismus surgery. Reconstructed 3D images were interpreted as having normal or abnormal anatomy and position, based on comparison with images acquired from 6 normal eyes. Interpretation was validated against the intraoperative findings from strabismus surgery. RESULTS: A total of 34 scans from the study group were used for comparison and validation purposes. Accuracy of anatomical assessments was 85% +/- 12% (percentage +/- confidence interval) and of positional assessments was 62% +/- 16%. Sensitivity and specificity of anatomical assessments was 80% +/- 14% and 88% +/- 10%, respectively. Anatomic anomalies detected by 3D US included excessively thick, thin, scarred and fibrotic, and absent muscles. Sensitivity and specificity of positional assessments was 48% +/- 17% and 85% +/- 12%, respectively. Positional anomalies such as muscle displacement off the normal clock hour axis or posteriorly displaced insertion were also detected. CONCLUSIONS: The 3D US may have an adjunctive role in determining anatomy and position of rectus muscles in patients with craniosynostosis, although it was more accurate in assessing anatomic features rather than positional features of rectus muscles.