Artificial Intelligence System for Assessing Image Quality of Fundus Images and Its Effects on Diagnosis

• Conditions: Artificial Intelligence; Retinal Diseases
• Interventions: Device: Taking a fundus image

• Registration Number: NCT04289064
• Lead Sponsor: Sun Yat-sen University

Brief Summary

Fundus images are widely used in ophthalmology for the detection of diabetic retinopathy, glaucoma and other diseases. In real-world practice, the quality of fundus images can be unacceptable, which can undermine diagnostic accuracy and efficiency. Here, the researchers established and validated an artificial intelligence system to achieve automatic quality assessment of fundus images upon capture. This system can also provide guidance to photographers according to the reasons for low quality.

Detailed Description

Not available

Study Timeline

• Status Verified: 2020-02
• Study Start: 2020-02-01
• Study First Submitted: 2020-02-26
• Study First Submitted QC: 2020-02-26
• Last Update Submitted: 2020-02-26
• Study First Posted: 2020-02-28
• Last Update Posted: 2020-02-28
• Primary Completion: 2020-07-01
• Study Completion: 2020-07-01

Recruitment & Eligibility

• Status: UNKNOWN
• Sex: All
• Target Recruitment: 300

Inclusion Criteria

• Patients should be aware of the contents and signed for the informed consent.

Exclusion Criteria

• 1. Patients who cannot cooperate with a photographer such as some paralytics, the patients with dementia and severe psychopaths.
• 2. Patients who do not agree to sign informed consent.

Study & Design

• Study Type: OBSERVATIONAL
• Study Design: Not specified

Arm && Interventions

Group	Intervention	Description
Fundus image quality assessment	Taking a fundus image	Device: an artificial intelligence system for quality assessment of fundus images. These patients are enrolled in primary healthcare units or the AI clinic at Zhongshan Ophthalmic Center.

Primary Outcome Measures

Name	Time	Method
Performance of artificial intelligence system for distinguish between good image quality and poor image quality	3 months	Area under the receiver operating characteristic curves, sensitivity, specificity, positive and negative predictive values, accuracy

Secondary Outcome Measures

Name	Time	Method
The comparison of the performance for previous artificial intelligence diagnostic system with fundus images of different image quality	3 months	Cohen's kappa coefficient, P value and other related statistic results

Trial Locations

Locations (1)

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