A lightweight DeepME model based on improved YOLOv11 architecture for macular edema detection and treatment monitoring
-
By
-
Xue Bai
-
Ming Yi
-
Tianye Chen
-
Naijing Feng
-
Quan Shi
-
Tingxue Li
-
Rui Hua
-
July 3, 2026
-
Clinical Scorecard: A streamlined DeepME framework utilizing an enhanced YOLOv11 architecture for the identification and management of macular edema
At a Glance
| Category | Detail |
|---|
| Condition | Macular Edema |
| Key Mechanisms | Deep learning algorithm using YOLOv11 architecture for OCT image analysis. |
| Target Population | Patients with diabetic retinopathy or retinal vein occlusion. |
| Care Setting | Clinical decision support system for ophthalmology. |
Key Highlights
- DeepME achieved accuracy of 0.980 and specificity of 0.990.
- Significant decrease in central foveal thickness at one-month follow-up (p < 0.001).
- DeepME shows substantial agreement with manual grading (p < 0.001).
- Integrates clinical guidelines and expert knowledge for treatment recommendations.
- Utilizes a comprehensive dataset from hospital and public OCT resources.
Guideline-Based Recommendations
Diagnosis
- Use optical coherence tomography (OCT) for diagnosing macular edema.
Management
- First-line therapy for DME and RVO-ME is intravitreal anti-VEGF injections.
Monitoring & Follow-up
- Regular follow-ups are necessary to assess treatment response.
Risks
- Over 30% of patients may exhibit persistent DME despite treatment.
Patient & Prescribing Data
Patients with diabetic retinopathy or retinal vein occlusion.
DeepME provides rapid, accurate diagnosis and treatment guidance.
Clinical Best Practices
- Implement AI-assisted tools for enhanced diagnostic accuracy.
- Utilize comprehensive datasets for model training to improve adaptability.
- Ensure compliance with GDPR and HIPAA regulations in data handling.
Related Resources & Content
