Development and Validation of a Deep Learning System for Multiple Ocular Fundus Diseases Using Retinal Images

• Conditions: Ophthalmological Disorder
• Interventions: Other: diagnostic

• Registration Number: NCT04213430
• Lead Sponsor: Sun Yat-sen University

Brief Summary

Retinal images can reflect both fundus and systemic conditions (diabetes and cardiovascular disease) and firstly to be used for medical artificial intelligence (AI) algorithm training due to its advantages of clinical significance and easy to obtain. Here, the investigators developed a single network model that can mine the characteristics among multiple fundus diseases, which was trained by plenty of fundus images with one or several disease labels (if they have) in each of them. The model performance was compared with those of both native and international ophthalmologists. The model was further tested by datasets with different camera types and validated by three external datasets prospectively collected from the clinical sites where the model would be applied.

Detailed Description

Not available

Study Timeline

• Study Start: 2014-01
• Status Verified: 2019-12
• Study First Submitted: 2019-12-23
• Study First Submitted QC: 2019-12-24
• Last Update Submitted: 2019-12-24
• Study First Posted: 2019-12-30
• Last Update Posted: 2019-12-30
• Primary Completion: 2020-02
• Study Completion: 2020-05

Recruitment & Eligibility

• Status: UNKNOWN
• Sex: All
• Target Recruitment: 300000

Inclusion Criteria

• The quality of fundus images should clinical acceptable. More than 80% of the fundus image area including four main regions (optic disk, macular, upper and lower retinal vessel archs) are easy to read and discriminate.

Exclusion Criteria

• Images with light leakage (>30% of area), spots from lens flares or stains, and overexposure were excluded from further analysis.

Study & Design

• Study Type: OBSERVATIONAL
• Study Design: Not specified

Arm && Interventions

Group	Intervention	Description
Testing dataset	diagnostic	Retinal images prospectively collected from the hospitals and ocular disease screening sites totally different from training dataset
Validation dataset	diagnostic	Retinal images separated from training dataset

Primary Outcome Measures

Name	Time	Method
Area under the receiver operating characteristic curve of the deep learning system	baseline	The investigators will calculate the area under the receiver operating characteristic curve of deep learning system and compare this index between deep learning system and human doctors.

Secondary Outcome Measures

Name	Time	Method
Sensitivity of the deep learning system	baseline	The investigators will calculate the sensitivity of deep learning system and compare this index between deep learning system and human doctors.
Specificity of the deep learning system	baseline	The investigators will calculate the specificity of deep learning system and compare this index between deep learning system and human doctors.

Trial Locations

Locations (1)

Zhongshan Ophthalmic Center, Sun Yat-sen Univerisity; 🇨🇳 Guangzhou, Guangdong, China