Image analysis-based machine learning for the diagnosis of retinopathy of prematurity: A meta-analysis and systematic review.

Researchers

Journal

Modalities

Models

Abstract

To evaluate the performance of machine learning (ML) in the diagnosis of retinopathy of prematurity (ROP) and to assess whether it can be an effective automated diagnostic tool for clinical applications.Early detection of retinopathy of prematurity (ROP) is crucial for preventing tractional retinal detachment and blindness in preterm infants, which has significant clinical relevance.Web of Science, PubMed, Embase, IEEE, and Cochrane Library were searched for published studies on image-based ML for diagnosis of ROP or classification of clinical subtypes from inception to October 1, 2022. QUADAS-AI was used to conduct the risk of bias (Rob) research on the included original studies. The bivariate mixed effects model was used for quantitative analysis of the data, and the Deek’s test was used for calculating publication bias. Quality of evidence was assessed using Grading of Recommendations Assessment, Development and Evaluation (GRADE).Twenty-two studies were included in the systematic review; four studies were at high or unclear Rob. In the area of indicator test items, only two studies were at high or unclear Rob because they did not establish predefined thresholds. In the area of reference standards, three studies had a high or unclear risk of bias. Regarding applicability, only one study was considered to have high or unclear applicability in terms of patient selection. The machine learning methods involved used deep learning in 86% of the studies. The sensitivity and specificity of image-based ML for the diagnosis of ROP 93% (95% CI:0.90-0.94), 95% (95% CI:0.94-0.97), AUC was 0.98(95% CI:0.97-0.99) and the sensitivity, specificity was 93% (95% CI:0.89-0.96), 93% (95% CI:0.89-0.95), AUC was 0.97(95% CI:0.96-0.98) for the classification of clinical subtypes of ROP. The classification results were highly similar to those of clinical experts (Spearman’s R=0.879).Machine learning algorithms is no less accurate than the human expert and hold considerable potential as automated diagnostic tools for ROP. However, given the quality and high heterogeneity of the available evidence, these algorithms should be considered as supplementary tools to assist clinicians in diagnosing ROP.Copyright © 2024. Published by Elsevier Inc.