iROP DL for Retinopathy of Prematurity

Overseen By Susan Ostmo, MS

Age: 18+

Sex: Any

Trial Phase: Academic

Sponsor: Siloam Vision

Disqualifiers: Non-U.S. board-certified, others

No Placebo Group All trial participants will receive the active study treatment (no placebo)

Approved in 1 Jurisdiction This treatment is already approved in other countries

Compensation: Varies

Number of Visits: 2

16 Participants Needed

What You Need to Know Before You Apply

What is the purpose of this trial?

This trial tests a tool called iROP DL (Imaging and Informatics in Retinopathy of Prematurity Deep Learning) to determine if it helps doctors diagnose retinopathy of prematurity (ROP) more accurately in premature babies. ROP can affect vision, and the trial compares doctors' ability to identify disease stages with and without the iROP DL tool. Licensed ophthalmologists interested in improving their diagnostic skills with this new technology can participate. Participants must be board-certified ophthalmologists in the U.S., agree to the study terms, and complete specific training. As an unphased trial, this study offers a unique opportunity for ophthalmologists to enhance their diagnostic skills with cutting-edge technology.

Do I have to stop taking my current medications for the trial?

The trial protocol does not specify whether you need to stop taking your current medications.

What prior data suggests that the i-ROP DL is safe for diagnosing retinopathy of prematurity?

Research has shown that the iROP DL system, a tool using advanced computer technology, is generally safe for diagnosing retinopathy of prematurity (ROP). Studies have found that this technology accurately detects different stages of ROP without causing harm. For instance, one study noted that the iROP DL system effectively identified both those with and without the disease.

These findings suggest that the system is safe for patients. Since the iROP DL is used for diagnosis and is not a drug or surgery, the risks remain very low. The main goal is to improve the accuracy of diagnosing ROP, which is crucial for timely treatment and preventing vision loss in babies. ^{1 2 3 4 5}

Why are researchers excited about this trial?

Most treatments for Retinopathy of Prematurity (ROP) involve laser therapy or injections of medications like anti-VEGF agents. However, the iROP DL approach is different because it uses artificial intelligence to provide a Vascular Severity Score based on retinal images. This offers a non-invasive, rapid, and potentially more precise way to assess the severity of ROP. Researchers are excited because this could lead to earlier and more accurate interventions, reducing the risk of vision loss in premature infants.

What evidence suggests that the i-ROP DL is effective for diagnosing retinopathy of prematurity?

Research has shown that the iROP DL system, a tool using advanced computer technology, can effectively diagnose retinopathy of prematurity (ROP). In past studies, this system accurately identified which cases require treatment and which do not. By analyzing eye images, it predicts ROP severity, aiding doctors in decision-making before the condition worsens. Studies have also found that systems like iROP DL learn quickly and accurately identify ROP cases. Overall, the iROP DL system shows promise in improving ROP diagnosis and treatment. ^{2 3 5 6 7}

Who Is on the Research Team?

John P Campbell, MD/MPH

Principal Investigator

Oregon Health and Science University

Are You a Good Fit for This Trial?

This trial is for U.S. board-eligible or certified ophthalmologists who have signed consent forms, completed training on the study protocol and software, and agreed to participate in a reader study. It's focused on improving diagnosis of Retinopathy of Prematurity (ROP) using an AI system.

Inclusion Criteria

You have completed the required training and are proficient in using the associated software.

Signed informed consent.

You have accepted the reader study agreement.

Timeline for a Trial Participant

Screening

Participants are screened for eligibility to participate in the trial

2-4 weeks

Reading Process

Ophthalmologists perform two reads on all cases, first without and then with the aid of the i-ROP DL, separated by a memory washout period

4 weeks

2 reading sessions

Follow-up

Participants are monitored for safety and effectiveness after the reading process

4 weeks

What Are the Treatments Tested in This Trial?

Interventions

• iROP DL

Trial Overview

The i-ROP DL system is being tested for its ability to accurately diagnose 'plus disease' in ROP through image analysis. The study will assess how image quality and camera operator influence results, aiming to validate the system as an autonomous AI screening device.

How Is the Trial Designed?

Group I: Reader Study Experimental Treatment 1 Intervention

Vascular Severity Score Provided on second reading

iROP DL is already approved in United States for the following indications:

Approved in United States as i-ROP DL for:

• Detection of more-than-mild retinopathy of prematurity (ROP)

Who Is Running the Clinical Trial?

Siloam Vision

Lead Sponsor

Trials

2

Recruited

1,300+

National Eye Institute (NEI)

Collaborator

Trials

572

Recruited

1,320,000+

Published Research Related to This Trial

A deep convolutional neural network (CNN) was developed to automatically detect and classify early stages of retinopathy of prematurity (ROP) in a study involving 11,372 retinal fundus images from premature infants, achieving an impressive accuracy of 99.93% during training and 92.23% during testing.

The model demonstrated high sensitivity (up to 96.14%) and specificity (up to 98.99%) in distinguishing between different stages of ROP, indicating its potential to assist ophthalmologists in early diagnosis and classification of this condition.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Automated detection of early-stage ROP using a deep convolutional neural network. Huang, YP., Basanta, H., Kang, EY., et al.[2021]

A deep learning-derived vascular severity score for retinopathy of prematurity (ROP) effectively correlates with clinical ROP diagnoses, indicating that higher scores are associated with more severe disease stages and greater extent of stage 3 disease.

The study demonstrated good interobserver agreement among clinicians using the 1-to-9 vascular severity scale, with a weighted κ value of 0.67 and a Pearson correlation coefficient of 0.88, suggesting that this quantitative scale could enhance the consistency and reliability of ROP assessments in clinical settings.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Evaluation of a Deep Learning-Derived Quantitative Retinopathy of Prematurity Severity Scale. Campbell, JP., Kim, SJ., Brown, JM., et al.[2022]

The developed deep learning system for diagnosing retinopathy of prematurity (ROP) achieved high accuracy, with sensitivities ranging from 90.21% for stage I to 91.84% for stage III, and specificities above 97%, indicating it can effectively identify early stages of ROP in preterm infants.

By integrating quantitative features from retinal images, the system significantly improved diagnostic accuracy when used alongside physician assessments, suggesting it could enhance clinical decision-making in preventing childhood blindness.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Early Diagnosis and Quantitative Analysis of Stages in Retinopathy of Prematurity Based on Deep Convolutional Neural Networks. Li, P., Liu, J.[2022]

A deep learning algorithm developed for screening retinopathy of prematurity (ROP) demonstrated high accuracy, achieving sensitivity rates between 91.8% and 98.2% and specificity rates between 94.9% and 99.2% across 14,108 eyes of 8,652 preterm infants.

The automated platform's performance was comparable to experienced ophthalmologists, with a Cohen unweighted κ of 0.86 to 0.98, suggesting it could effectively assist in routine ROP screenings to improve early detection and treatment outcomes.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Automated Explainable Multidimensional Deep Learning Platform of Retinal Images for Retinopathy of Prematurity Screening. Wang, J., Ji, J., Zhang, M., et al.[2021]

The i-ROP deep learning system demonstrated a high accuracy for detecting type 1 retinopathy of prematurity (ROP), with an area under the receiver operating curve of 0.960 based on analysis of 4861 examinations from 870 infants.

With a threshold score of 3, the system achieved 94% sensitivity and 99.7% negative predictive value, indicating it is highly effective in identifying infants at risk for severe ROP, which could enhance the objectivity and accessibility of screening for this condition.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Evaluation of a deep learning image assessment system for detecting severe retinopathy of prematurity. Redd, TK., Campbell, JP., Brown, JM., et al.[2022]

The deep learning algorithm (i-ROP) demonstrated high efficacy in detecting plus disease in retinopathy of prematurity (ROP), achieving an area under the receiver operating characteristic curve of 0.999 in Nepal and 0.968 in Mongolia, indicating its strong performance in screening.

The prevalence of type 1 ROP was significantly higher in Mongolia (14.0%) compared to Nepal (2.2%), suggesting a need for targeted screening and intervention strategies in regions with higher incidence rates.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Evaluation of an Artificial Intelligence System for Retinopathy of Prematurity Screening in Nepal and Mongolia. Cole, E., Valikodath, NG., Al-Khaled, T., et al.[2022]

The DeepROP score, derived from a deep learning algorithm, demonstrated high accuracy in detecting severe retinopathy of prematurity (ROP), with an area under the receiver operating curve of 0.981 for Type 1 ROP and 0.986 for Type 2 ROP, based on analysis of 9,882 examinations from 2,801 infants.

The algorithm effectively identified all cases of severe ROP while excluding 87.6% of eyes with no or mild ROP when a cutoff score of 35 was used, indicating its potential for automated screening in clinical settings.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

EVALUATION OF ARTIFICIAL INTELLIGENCE-BASED QUANTITATIVE ANALYSIS TO IDENTIFY CLINICALLY SIGNIFICANT SEVERE RETINOPATHY OF PREMATURITY. Li, J., Huang, K., Ju, R., et al.[2023]

Deep learning algorithms demonstrated high diagnostic accuracy for detecting and grading retinopathy of prematurity (ROP) in preterm infants, with pooled sensitivity of 95.3% and specificity of 97.5% across nine studies involving over 521,000 images.

The area under the curve (AUC) for the overall performance of these models was 0.984, indicating excellent ability to distinguish between different grades of ROP, suggesting that DL models could significantly enhance ROP screening and diagnosis in clinical settings.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Accuracy of Deep Learning Algorithms for the Diagnosis of Retinopathy of Prematurity by Fundus Images: A Systematic Review and Meta-Analysis. Zhang, J., Liu, Y., Mitsuhashi, T., et al.[2022]

Citations

1.

pmc.ncbi.nlm.nih.gov

pmc.ncbi.nlm.nih.gov/articles/PMC9560533/

Improved Training Efficiency for Retinopathy of Prematurity ...

Neural networks learned more efficiently by comparison, generating significantly higher AUCs in both classification tasks across both datasets.

2.

jamanetwork.com

jamanetwork.com/journals/jamanetworkopen/fullarticle/2793479

A Deep Learning Model to Predict the Occurrence and ...

To develop and validate a deep learning (DL) system to predict the occurrence and severity of ROP before 45 weeks' postmenstrual age.

3.

nature.com

nature.com/articles/s41433-025-04096-3

Deep learning algorithms for timely diagnosis of ...

This study aimed to evaluate the effectiveness of DL algorithms in diagnosing ROP cases that requires treatment using retinal images submitted ...

4.

aao.org

aao.org/education/editors-choice/autonomous-deep-learning-system-shows-potential-de

An Autonomous Deep-Learning System Shows Potential ...

Outcomes. The iROP DL system showed good sensitivity and specificity in detecting mtm ROP and type 1 ROP in both the SUNDROP and ACES cohorts.

5.

dl.acm.org

dl.acm.org/doi/10.1145/3596223

Deep Learning-assisted Retinopathy of Prematurity (ROP) ...

The ROP classification algorithm uses OD as a reference point to determine the degree and progression of a disease based on the extent of blood vessels. In ...

6.

pmc.ncbi.nlm.nih.gov

pmc.ncbi.nlm.nih.gov/articles/PMC8364291/

Deep Learning for the Diagnosis of Stage in Retinopathy ...

The purpose of this study was to implement a convolutional neural network (CNN) for binary detection of stage 1–3 in ROP and evaluate its generalizability ...

7.

ophthalmologyscience.org

ophthalmologyscience.org/article/S2666-9145(22)00011-2/fulltext

Improved Training Efficiency for Retinopathy of Prematurity ...

Improved training efficiency for retinopathy of prematurity deep learning models using comparison versus class labels.