• Center on Health Equity & Access
  • Clinical
  • Health Care Cost
  • Health Care Delivery
  • Insurance
  • Policy
  • Technology
  • Value-Based Care

Deep Learning Model Detects Sea Fan Neovascularization in Patients With Sickle Cell Hemoglobinopathy

Article

In an effort to detect sea fan neovascularization from ultra-widefield color fundus photographs from patients with sickle cell hemoglobinopathy, researchers developed an automated system with high sensitivity and specificity, which holds potential to improve screening for vision-threatening proliferative sickle cell retinopathy.

In an effort to detect sea fan neovascularization from ultra-widefield color fundus photographs (UWF-FPs) from patients with sickle cell hemoglobinopathy (SCH), researchers developed an automated system with high sensitivity and specificity, which holds potential to improve screening for vision-threatening proliferative sickle cell retinopathy. Results of the cross-sectional study were published in JAMA Ophthalmology.

Around the world, at least 1 in 1000 infants is born with SCH, and the birth prevalence rate is at least 10 times higher in Africa compared with other countries, researchers explained. “In addition to the morbidity and mortality caused by systemic vaso-occlusive disease, patients with SCH are at risk for vision loss from complications of proliferative sickle cell retinopathy (PSR).”

The hallmark of PSR is pathologic sea fan neovascularization in the peripheral retina which develops as a result of chronic retinal vaso-occlusion and retinal ischemia. Sea fans can be associated with severe vision loss and recurrent vitreous hemorrhages, and the most advanced stage of PSR can lead to tractional retinal detachments.

According to the authors, “early detection of potentially asymptomatic sea fan neovascularization affords the opportunity to offer prophylactic scatter laser photocoagulation, which a randomized clinical trial reported to approximately halve the rates of PSR-associated vision loss and vitreous hemorrhage compared with observation alone.”

Current guidelines recommend screening dilated fundus examinations in patients with SCH starting by age 10 years. However, adherence to these guidelines is low, and, as many patients with PSR present with no visual symptoms, patients with SCH with good vision may not prioritize screening.

“Color UWF-FPs have been shown to have excellent sensitivity for capturing PSR when reviewed by retinal specialists or ophthalmology residents,” the authors wrote. To test a deep learning system’s capability to classify color UWF-FPs, researchers retrospectively identified adults who underwent the imaging procedure at a single eye institute between January 1, 2012, and January 30, 2019.

All participants had a diagnosis of sickle cell retinopathy or SCH, or had the sickle cell trait. Any patients with diabetes or a retinal vascular disease other than sickle cell retinopathy were excluded. In addition, “images from eyes with previous laser or surgery for PSR, macula-involving retinal detachment, severe media opacity, or vitreous hemorrhage obscuring more than 10% of fundus detail were excluded.”

Investigators used an updated version of the Inception V4 convolutional neural network (CNN) architecture to train a deep learning model to distinguish which images had sea fan neovascularization.

Retinal specialist graders also agreed on reference standard grades for each image to serve as a baseline.

A total of 1182 color UWF-FPs from 190 patients were included in training, validation, and testing, while corresponding fluorescein angiography (FA) images were available for 898 of included color UWF-FPs (76%). The majority of included patients (62.6%) had hemoglobin SS disease and 24.2% had hemoglobin SC disease. Around 94% of included patients were of Black or African descent.

Researchers found:

  • Images with sea fan neovascularization were obtained in 57 patients (30%).
  • The CNN had an area under the curve of 0.988 (95% CI, 0.969-0.999).
  • Sensitivity and specificity for detecting sea fan neovascularization from UWF-FPs were measured at 97.4% (95% CI, 86.5%-99.9%) and 97.0% (95% CI, 93.5%-98.9%), respectively.
  • Only 1 image received a false-negative classification from the CNN in the setting of a severe lid artifact obscuring most of the retinal vasculature.

“Because patients with SCH are predominantly of Black or African and Hispanic descent, PSR disproportionately affects medically underserved patients who may have less access to retinal specialist care,” the authors wrote. An automated system, such as the one developed in this study, could increase accessibility to regular fundus screenings and help identify patients most in need of further evaluation, they argued.

Ingratiating the system into nonophthalmologic medical practices may also help reduce the burden of in-person ophthalmology visits for asymptomatic patients with SCH.

The relatively small number of images and the single-institution design of the study mark limitations. External validation of the CNN using data sets from adult and pediatric patients is also warranted.

Reference

Cai S, Parker F, Urias MG, Goldberg MF, Hager GD, Scott AW. Deep learning detection of sea fan neovascularization from ultra-widefield color fundus photographs with sickle sell hemoglobinopathy. JAMA Ophthalmol. Published online December 30, 2020. doi:10.1001/jamaophthalmol.2020.5900

Related Videos
Masanori Aikawa, MD
Glenn Balasky, executive director of the Rocky Mountain Cancer Center.
Benjamin Scirica, MD, MPH, associate professor of medicine at Harvard Medical School and director of quality initiatives at Brigham and Women’s Hospital’s Cardiovascular Division
Glenn Balasky during a video interview
dr joseph alvarnas
Michael Lynch, MD, UPMC
dr alex jahangir
Fahad Tahir, MAS, MBA, FACHE, Ascension St Thomas
Leland Metheny, MD, University Hospitals Seidman Cancer Center
Quint Petris
Related Content
© 2024 MJH Life Sciences
AJMC®
All rights reserved.