Android

Your AI-Powered Companion in Retinal Care

Designed & Programmed by Dr. Ameen Marashi

Empowering Ophthalmologists

mRetinaAI seamlessly integrates cutting-edge artificial intelligence into the retina clinic, providing smarter diagnostics and management tools for a range of retinal diseases.

Key Features

Case Generator

• Utilize intuitive drop-down menus and toggle controls for more than 20 of the most common retina diseases.

• Receive evaluations and guided management plans for retinal conditions.

OCT Classifier

• OCT classifier helps in detecting and diagnosing macular diseases with up to 81.6 % accuracy.

• Choose a disease entity, view treatment recommendations, and streamline your plans.

• Focuses on predictive outcomes rather than extracting features or findings.

Overall Accuracy by Condition

• DME (Diabetic Macular Edema): 70.0 %

• RVO (Retinal Vein Occlusion): 82.4 %

• AMD (Age-Related Macular Degeneration): 96.8 %

• VMA (Vitreo-Macular Adhesion): 78.0 %

• Pachychoroid (CSCR & PCV): 82.1 %

• HRD (Hereditary Retinal Diseases): 80.6 %

Highlights

• AMD (96.8 %): Exceptional accuracy; only rare misses when clinical signs are very subtle.

• RVO (82.4 %): Strong detection of RVO- and RAO-related edema, with minor errors in atypical presentations.

• Pachychoroid (82.1 %): Reliable overall; challenges remain distinguishing PCV from MNV-complicated CSCR.

• HRD (80.6 %): Good performance across hereditary conditions; ongoing refinements aim to reduce subtype confusion.

• DME (70.0 %): Solid baseline accuracy; future updates will target complex DME and tractional patterns.

• VMA (78.0 %): Moderate accuracy; most errors involve subtle lamellar holes or overlap with other pathologies.

Guidance

• If a result seems uncertain, please consult a retinal specialist for confirmation.

• Adhere to recommended OCT imaging protocols to minimize artifacts and ensure best performance.

• Always corroborate classifier outputs with a full clinical examination.

FFA Classifier

• FFA classifier helps detect and diagnose macular diseases with up to 85.5 % accuracy.

• Quickly access treatment recommendations and streamline plans.

• Predicts diagnoses and recommends treatments without extracting features or findings.

Overall Accuracy by Condition

• Diabetic Retinopathy (FFA): 76.2 %

• Retinal Vein Occlusion (RVO): 98.1 %

• Age-Related Macular Degeneration (AMD): 80.0 %

• Pachychoroid (CSCR): 90.9 %

Highlights

• RVO (98.1 %): Near-perfect detection of neovascular signs; very few misclassifications.

• CSCR (90.9 %): Strong performance; occasional misses in subtle leakage patterns.

• AMD (80.0 %): Good overall accuracy; most errors involve subtle macular neovascularization.

• Diabetic Retinopathy (76.2 %): Solid baseline accuracy; ongoing refinements aim to improve early neovascular detection.

Guidance

• If a result seems uncertain, consult a retinal specialist for confirmation.

• Follow best-practice angiographic imaging protocols to minimize artifacts and overexposure.

• Always corroborate classifier outputs with a comprehensive clinical examination.

Privacy & Security

• No data is uploaded or shared; everything is processed locally.

• Patient and user data securely protected.

• No personal information or registration required.

Offline Functionality

• All functionalities available offline, no internet connection required.

• Fully operational without reliance on external servers or cloud-based services.

• Secure local data processing with instant access.

Important Notes

• Scope: Not intended for uveitis, tumors, or surgical retina.

• Always use clinical judgment; mRetinaAI does not replace professional expertise.