Smartphone-based retinal camera modified for ROP screening

TL;DR

A modified smartphone retinal camera was tested for ROP screening and showed high accuracy, especially when used with a technique to better view the retina's periphery.

Contribution

The study introduces a modified smartphone-based retinal camera with selective scleral indentation for improved ROP screening in resource-limited areas.

Findings

The device achieved 82% sensitivity and 100% specificity without scleral indentation.

Using scleral indentation increased sensitivity to 91.3% and negative predictive value to 98.3%.

Agreement between the camera and standard IBO was almost perfect (κ = 0.94).

Abstract

Retinopathy of prematurity (ROP) is a leading cause of preventable childhood blindness worldwide, particularly affecting infants in resource-limited settings. Although indirect binocular ophthalmoscopy (IBO) is considered the definitive method for ROP screening, its broad adoption is restricted by the scarcity of skilled professionals and the practical difficulties involved in performing exams at the patient’s bedside. Smartphone-based retinal imaging is emerging as a viable, low-cost alternative for telemedicine-based ROP screening in underserved regions.

This cross-sectional study was conducted from June 2022 to December 2023 in a Brazilian tertiary hospital. Preterm infants (GA ≤ 32 weeks and/or BW ≤ 1500 g) underwent standard ROP screening using IBO and a modified smartphone-based retinal camera (Phelcom Eyer®). Images of five retinal fields per eye were acquired by a single examiner, both with and without scleral indentation, which was performed when necessary to visualize the peripheral retina. A masked retina specialist graded image quality and classified ROP stages. Sensitivity, specificity, predictive values, and agreement with IBO (Cohen’s kappa) were calculated.

Seventy-one preterm infants (142 eyes) were included. ROP was diagnosed in 14 infants (19.7%), with six requiring treatment. Among the 337 images obtained, 69% were rated as excellent and 31% as acceptable; all were gradable. Compared to IBO, the device demonstrated a sensitivity of 82.0% and specificity of 100.0% without scleral indentation. When indentation was used—selectively, to improve visualization of Zone III—sensitivity increased to 91.3% and the negative predictive value to 98.3%. Agreement between methods was almost perfect (κ = 0.94). Only one transient episode of bradycardia occurred during indentation.

The modified Phelcom Eyer® camera demonstrated high diagnostic performance for ROP screening, particularly when complemented by selective scleral indentation to enhance peripheral retinal assessment. Its portability, affordability, and integration with telemedicine platforms make it a valuable tool for expanding ROP screening in resource-limited settings with restricted access to specialists.

Figures