Diabete Mellitus, Diabetic Retinopathy
Conditions
Keywords
Diabete Mellitus, Diabetic Retinopathy, Artificial Intelligence
Brief summary
This study is to build an multi-modal artificial intelligence ophthalmological imaging diagnostic system covering multi-level medical institutions. We are going to evaluate this system in an evidence-based medicine view, taking diabetic retinopathy as an example. And clinical diagnostic criteria will be made based on this multi-modal artificial intelligence imaging diagnostic system. The study is designed as a cross-sectional study involving 1,000 normal individuals, 1,000 diabetes patients without ocular complications, and 1,000 with diabetic ocular complications. Statistical analysis of the diagnostic sensitivity and specificity of the artificial intelligence system will be made, and ROC curve wil be draw.
Interventions
None listed
Sponsors
Zhongshan Ophthalmic Center, Sun Yat-sen University
Study design
Observational model
COHORT
Time perspective
CROSS_SECTIONAL
Eligibility
Sex/Gender
ALL
Healthy volunteers
Yes
Inclusion criteria
Group A(normal individuals): Meet all the items 1 \ 5 below 1. No familial or hereditary retinopathy; 2. No history of ocular trauma and / or history of ocular disease (except for mild to moderate refractive errors and / or age-related cataracts); 3. No drug use history that may cause side effects of retina (such as chloroquine, hydroxychloroquine, chlorpromazine, rifampin, etc.); 4 binocular diopter ≤ ± 6.00D; No systemic diseases that may affect the retina. Group B(diabetes patients without ocular complications): meet any of 1 to 3, and both 4 to 5 items 1. Diagnosed with type 1 diabetes for more than 5 years; 2. Diagnosed with type 2 diabetes patients; 3. Diagnosis of gestational diabetes patients; 4. Without systemic disease that may affect the retina except for diabetes; 5. Meet the standards of 1 to 4 items. Group C(patients with diabetic ocular complications): meet with any of 1 to 3, and all 4 to 6 items 1. Diagnosed with type 1 diabetes for more than 5 years; 2. Diagnosed with type 2 diabetes patients; 3. Diagnosis of gestational diabetes patients; 4. Meet the guidelines for diagnosis and treatment of diabetic retinopathy in any stage of diabetic retinopathy and / or diabetic macular edema. 5. Without systemic disease that may affect the retina except for diabetes; 6. A group to meet the criteria for 1 to 4 items.
Exclusion criteria
1. Refractive pathway is turbid, so that fundus image can not be clearly photographed; 2. With other fundus diseases and / or other diseases that seriously affect the visual function; 3. Any eye infection, such as conjunctivitis, keratitis, scleritis, endophthalmitis, acute and chronic dacryocystitis, or eye injury; 4. Serious systemic diseases such as diabetes, hypertension, heart failur,renal failure; 5. Can not cooperate with the examiner due to mental or other reasons.
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| The diagnostic sensitivity, specificity and ROC curve of the artificial intelligence（AI） system compared with reference standard (ophthalmologist). | It will take 15~20min for each subject to take the exams. | Sensitivity: the percentage of diabetic retinopathy（DR） patients who are correctly identified as having the condition by AI. Sensitivity=True positive/(True positive+False negative). Specificity: the percentage of healthy people who are correctly identified as not having the condition by AI. Specificity=True negative /(True negative +False positive). The ROC curve was plotted with true positive rate (sensitivity) as the ordinate and false positive rate (1-specificity) as the abscissa. |
Countries
China
Contacts
Primary ContactYuan Jin, Professor
Outcome results
None listed