Fairness and generalisability in deep learning of retinopathy of prematurity screening algorithms: a literature review.

BACKGROUND: Retinopathy of prematurity (ROP) is a vasoproliferative disease responsible for more than 30 000 blind children worldwide. Its diagnosis and treatment are challenging due to the lack of specialists, divergent diagnostic concordance and variation in classification standards. While artificial intelligence (AI) can address the shortage of professionals and provide more cost-effective management, its development needs fairness, generalisability and bias controls prior to deployment to avoid producing harmful unpredictable results. This review aims to compare AI and ROP study's characteristics, fairness and generalisability efforts. METHODS: Our review yielded 220 articles, of which 18 were included after full-text assessment. The articles were classified into ROP severity grading, plus detection, detecting treatment requiring, ROP prediction and detection of retinal zones. RESULTS: All the article's authors and included patients are from middle-income and high-income countries, with no low-income countries, South America, Australia and Africa Continents representation.Code is available in two articles and in one on request, while data are not available in any article. 88.9% of the studies use the same retinal camera. In two articles, patients' sex was described, but none applied a bias control in their models. CONCLUSION: The reviewed articles included 180 228 images and reported good metrics, but fairness, generalisability and bias control remained limited. Reproducibility is also a critical limitation, with few articles sharing codes and none sharing data. Fair and generalisable ROP and AI studies are needed that include diverse datasets, data and code sharing, collaborative research, and bias control to avoid unpredictable and harmful deployments.

Sociodemographic disparities in ophthalmological clinical trials.

INTRODUCTION: In ophthalmology, clinical trials (CTs) guide the treatment of diseases such as diabetic retinopathy, myopia, age-related macular degeneration, glaucoma and keratoconus with distinct presentations, pathological characteristics and responses to treatment in minority populations.Reporting gender and race and ethnicity in healthcare studies is currently recommended by National Institutes of Health (NIH) and Food and Drug Administration (FDA) guidelines to ensure representativeness and generalisability; however, CT results that include this information have been limited in the past 30 years.The objective of this review is to analyse the sociodemographic disparities in ophthalmological phases III and IV CT based on publicly available data. METHODS: This study included phases III and IV complete ophthalmological CT available from clinicaltrials.org, and describes the country distribution, race and ethnicity description and gender, and funding characteristics. RESULTS: After a screening process, we included 654 CTs, with findings that corroborate the previous CT reviews' findings that most ophthalmological participants are white and from high-income countries. A description of race and ethnicity is reported in 37.1% of studies but less frequently included within the most studied ophthalmological specialty area (cornea, retina, glaucoma and cataracts). The incidence of race and ethnicity reporting has improved during the past 7 years. DISCUSSION: Although NIH and FDA promote guidelines to improve generalisability in healthcare studies, the inclusion of race and ethnicity in publications and diverse participants in ophthalmological CT is still limited. Actions from the research community and related stakeholders are necessary to increase representativeness and guarantee generalisability in ophthalmological research results to optimise care and reduce related healthcare disparities.

Strabismus Surgery in Orthophoric Patients With Symptomatic, Asymmetric Vertical or Horizontal Incomitance.

PURPOSE: To report the indications, operative strategies, and surgical outcomes of patients who undergo vertical and horizontal rectus muscle surgery for incomitant strabismus despite being orthophoric in primary gaze. DESIGN: Retrospective, interventional case series. METHODS: The setting for this study was an academic practice at Boston Children's Hospital. The patient population comprised 8 orthophoric patients who underwent strabismus surgery to treat vertical/horizontal incomitance. Observation procedures included review of surgical strategies, strabismus measurements in diagnostic gaze positions, and development of postoperative diplopia. The main outcome measures were preserved single vision in primary gaze, comitance, reoperation rate, and patient/surgeon satisfaction. RESULTS: Surgical strategies included the following: (1) simultaneous recession of ipsilateral antagonist rectus muscles; (2) recession or resection of 1 rectus muscle with balancing surgery on the fellow eye; (3) restricting the range of 1 muscle (combined resection and recession or posterior fixation suture); and (4) creating an acceptable deviation in primary gaze. Mean follow-up was 5.4 months (median, 2 months; range, 2-25 months). No patient had new-onset primary gaze diplopia. The median incomitance improved by 9.5 prism diopters. No patient required additional surgery. Patient satisfaction and surgeon assessment of outcomes were high. CONCLUSIONS: Although the risk of operating on orthophoric patients with incomitant strabismus may discourage surgeons from offering treatment, the use of specific strategies to address incomitance can preserve alignment in primary gaze while improving patient satisfaction. These strategies may also benefit patients with incomitant strabismus that is symptomatic in primary gaze.

Validation of StrabisPIX, a Mobile Application for Home Measurement of Ocular Alignment.

PURPOSE: StrabisPIX is a smartphone application that allows clinicians to prescribe a series of self-obtained images of head position and eye alignment in nine positions of gaze that are uploaded onto a secure platform for clinician review. This study evaluates the clinical utility of this application. METHODS: In this prospective, nonmasked, cross-sectional study, 30 strabismus patients aged ≥2 years were evaluated. Participants received standardized instructions, used StrabisPIX to obtain images as prompted, and completed a satisfaction survey. During the same visit, an orthoptist obtained standard clinical images with a professional camera. All 60 image sets were evaluated by three observers. RESULTS: StrabisPIX image quality was similar to that of clinic photographs. Clinic photographs had significantly higher acceptability for horizontal versions (81% vs. 67%), vertical versions (76% vs. 60%), and head posture (93% vs. 81%). Abnormalities were detected at a similar rate for versions, head posture, eyelid position, and pupil size. StrabisPIX had significantly higher detection of alignment abnormalities (89% vs. 77% for clinical photos). Interrater/intrarater agreements were moderate to high (κ = 0.44-1.00) for all parameters except pupil abnormality, which had poor to fair agreement. Most patients reported that StrabisPIX was easy to learn and use. CONCLUSIONS: Overall, StrabisPIX images had similar quality and were as useful as images obtained in the clinic in detecting abnormalities. TRANSLATIONAL RELEVANCE: The StrabisPIX application will enhance the use of telemedicine by allowing physicians to prescribe self-obtained images documenting strabismus.

Anomalous Vertical Deviations in Attempted Abduction Occur in the Majority of Patients With Esotropic Duane Syndrome.

PURPOSE: To describe a phenomenon, depression in attempted abduction, not previously recognized as a feature of Duane syndrome (DS). DESIGN: Retrospective, observational case series. METHODS: Setting: Institutional practice. PATIENT POPULATION: Patients diagnosed with esotropic DS at Boston Children's Hospital from 2002 to 2015. Patients with clinical photographs documenting horizontal gaze were included. Patients with prior strabismus surgery were excluded. OBSERVATION PROCEDURES: Patients were classified into 3 groups according to their vertical eye position in attempted abduction: midline group, depression group, and elevation group. Group assignment was performed by 3 independent ophthalmologists. Baseline characteristics, eye movement, and ocular deviation were compared among the 3 groups. MAIN OUTCOME MEASURES: Horizontal and vertical deviation on attempted abduction in the DS eye. RESULTS: Depression in attempted abduction was present in 74 of 113 unilateral patients (66%) and 18 of 42 gradable eyes (43%) of bilateral patients. Abduction limitation was significantly less severe in the midline group (median: -3.0) than in the depression group (median: -4.0) (P = .01). Vertical deviation in attempted abduction was more severe in the elevation group than in the depression group (P = .003). CONCLUSIONS: Depression of the eye in attempted abduction has not been widely described, yet it is present in the majority of DS patients. It is more likely to occur with more severe abduction limitation. This phenomenon is likely another form of dysinnervation in DS, the result either of anomalous vertical rectus muscle activation or asymmetric lateral rectus muscle innervation during attempted abduction. Awareness of vertical deviation in attempted abduction may facilitate surgical planning in affected patients.

Sedated suture adjustment in children undergoing adjustable suture strabismus surgery.

PURPOSE: To study methods and adverse events of postoperative, sedated suture adjustment after strabismus surgery in the post-anesthesia care unit (PACU). METHODS: We reviewed the postoperative experience of all children ≤18 years of age undergoing adjustable suture strabismus surgery at Boston Children's Hospital over a 3-year period. Time in the hospital, adverse events, and surgical outcomes were reviewed to evaluate safety and healthcare resource utilization. RESULTS: Of 356 patients, 113 required suture adjustment in the PACU, including 24 adjusted while awake and 89 adjusted under sedation. For sedation, sequential boluses of propofol were administered until adjustment was complete. Complete data from the sedated adjustment was available in 76 patients. The median initial bolus was 30 mg; the median total propofol rate was 273 mcg/kg/min. Twelve patients (16%) required only a single bolus of propofol. Of remaining 64 patients, median time from initial to final propofol dose was 7 minutes. Median anesthesiologist time in the PACU was 13 minutes. In the sedated adjustment group, there were no clinically significant adverse events, and the pain score never exceeded 6 (of a possible 10). Median duration of PACU stay was shortest in the group not requiring adjustment. CONCLUSIONS: Sedated suture adjustment allows for fine-tuning of postoperative binocular alignment in children and uncooperative adults. No adverse events were observed in our study group, but the procedure does increase the time patients spend in the hospital. This work will inform disclosure of risks and benefits of sedated adjustment while allowing for more accurate assessment of the cost and quality of adjustable sutures in children.

Accuracy of Intraocular Lens Calculation by SRK/T Formula in Pediatric Cataracts.

OBJECTIVE: To examine the accuracy of the intraocular lens calculation by SRK/T formula in pediatric cataract patients, by means of the prediction error after the cataract surgery. MATERIAL AND METHOD: A retrospective review of pediatric patients requiring cataract extraction with intraocular lens implantation, between January 2006 and December 2013 was performed. Post-operative spherical equivalent was compared with the predicted calculation value from pre-operative SRK/T measurement, defined as prediction error (PE). RESULTS: 139 eyes from 106 patients were studied. The median age of patients at surgery was 4.56 years (IQR, 14). The two most common causes of cataract were idiopathic and traumatic; 85 eyes (61.2%) and 36 eyes (25.9%), respectively. Primary intraocular lens (IOL) implantation was performed in 87 (62.6%) eyes. The mean PE was -1.32±1.89 D (-7.78, 2.24). At one to three months after surgery, the mean PE of IOL in bag and sulcus were -0.87 D and -2.02 D, respectively. In non-traumatic group, the mean PE of IOL in bag and sulcus were -1.1 D and 2.28 D, respectively, whilst the traumatic group was -0.33 D and -1.16 D, respectively. Post-operative PE within ±1 diopter was found in 30 eyes (37.97%) at one to three months of follow-ups. CONCLUSION: The majority of post-operative refraction was more myopic than the taigeted refractive error, which was calculated by SRK/T formula. Aiming for a more hyperopic target and IOL implantation in the bag would decrease the post-operative prediction error in the use of the SRK/T formnula in pediatric cataract patients.

Contamination risk of 100% autologous serum eye drops in management of ocular surface diseases.

PURPOSE: To evaluate the sterility and safety of 100% nonpreserved, autologous, serum eye drop treatment in patients with ocular surface diseases. METHODS: A total of 147 autologous serum bottles (294 samples) from 21 patients with ocular surface diseases were included. Seven bottles with autologous serum were prepared for each patient, and each bottle was used for only 1 day, then discarded. Two samples from each bottle were taken (before use and after 24 hours of use) and then cultured in standard media. The culture plates were held for at least 6 weeks or until no growth could be confirmed. To monitor safety, all patients were admitted and evaluated for the occurrence of infection. RESULTS: In the pretreatment group, 4 samples from 4 patients (1.36%) were positive for bacteria and 7 samples from 7 patients (2.38%) were positive for fungi. In the 24-hour-after-treatment group, 1 culture (0.34%) was positive for bacteria, and 6 samples (2.04%) from 6 patients were positive for fungi. Aspergillus spp and Fonsecaea spp were the most common organisms identified in any of the cultures. Neither clinical nor microbiological evidence of infection was demonstrated in any patient during the treatment or follow-up periods. CONCLUSIONS: Although no ocular surface infection was observed, the patients under 100% autologous serum eye drops therapy should be closely monitored for clinical evidence of fungal and bacterial infections.