Real world variability in plus disease identified using a deep learning-based retinopathy of prematurity severity scale.

Researchers

Journal

Modalities

Models

Abstract

Retinopathy of prematurity is a leading cause of childhood blindness worldwide, but clinical diagnosis is subjective, which leads to real world treatment differences. Our goal is to determine objective differences in the diagnosis of plus disease between clinicians using an automated retinopathy of prematurity vascular severity score.
This retrospective cohort study used data from the Imaging & Informatics in ROP consortium, which comprises 8 tertiary care centers in North America. Fundus photographs of all babies undergoing retinopathy of prematurity screening exams between July 1, 2011, to December 31, 2016 were obtained.
Infants meeting retinopathy of prematurity screening criteria who were diagnosed with plus disease and treatment initiated by an examining physician based on ophthalmoscopic exam.
A retinopathy of prematurity severity score (1-9) was generated for each image using a deep learning algorithm.
The mean, median, and range of retinopathy of prematurity vascular severity scores overall and for each examiner when the diagnosis of plus disease was made.
A total of 5255 clinical examinations in 871 babies were analyzed. Of these, 168 eyes were diagnosed with plus disease by 11 different examiners and were included in the study. The mean (±standard deviation) vascular severity score for patients diagnosed with plus disease was 7.4 (±1.9), median was 8.5 (interquartile range [IQR] 5.8 – 8.9), and range went from 1.1 – 9.0. Within some examiners, there was variability in the level of vascular severity diagnosed as plus disease, and one examiner routinely diagnosed plus disease in patients with less severe disease than the others (P<.01).
We observed variability both between and within examiners in the real-world diagnosis of plus disease using deep learning. Prospective evaluation of clinical trial data using objective measurement of vascular severity may help better define the minimum necessary level of vascular severity for the diagnosis of plus disease, or how other clinical features such as zone, stage, and extent of peripheral disease ought to be incorporated in treatment decisions.
Copyright © 2020. Published by Elsevier Inc.