ANDROID

mRetinaAI — Your AI‑Powered Companion in Retinal Care

Designed & programmed by Dr. Ameen Marashi

To support faster, safer, and more consistent decision‑making in the retina clinic.

🤖 Download Instructions for Android 🛡️ Privacy Policy 🔑 Buy License

Introduction

mRetinaAI is built for fast, offline decision support in the retinal clinic. It combines a structured Case Generator to document findings and craft plans; an on‑device OCT Classifier targeting common macular entities; and an on‑device FFA Classifier tuned for angiographic patterns. All processing happens locally on your device—no cloud, no patient data upload.

Use the Case Generator to capture key elements (VA, stage, imaging, prior treatments), then review tailored recommendations. The OCT and FFA modules provide outcome‑focused predictions to help triage and streamline management. Results are intended to complement—never replace—your clinical judgment.

  • Offline by design: privacy‑first, no accounts, no syncing.
  • Clinic‑ready: quick inputs, readable outputs, and practical treatment guidance.
  • Transparent: targeted accuracies and known limitations by condition.
On‑device No cloud Clinic‑ready
mRetinaAI screenshot 2
mRetinaAI screenshot 1
mRetinaAI screenshot 3
mRetinaAI screenshot 4
mRetinaAI screenshot 5
mRetinaAI screenshot 6
mRetinaAI screenshot 7

📋 Case Generator

Build complete macula cases in minutes. The Case Generator structures history, exam, and imaging into a clean, clinic‑ready summary. It's designed for rapid triage and consistent documentation across common macular diseases.

  • Inputs: Visual acuity, symptoms & duration, laterality, key comorbidities, prior treatments (anti‑VEGF, laser, steroids), OCT/FFA findings, and risk factors (DM, HTN, anticoagulation).
  • Smart defaults: Context‑aware pickers and toggles minimize typing and keep notes standardized.
  • Outputs: Structured assessment, differential when relevant, and actionable plan suggestions (e.g., treat/observe, follow‑up interval, adjunct imaging).
  • Consistency: Uses the same terminology and layout every time, improving communication between providers and across visits.
  • Offline: Everything runs on‑device—no network, no data leaves your phone.

Tip: after generating the case, you can quickly cross‑check with the OCT/FFA modules to align imaging‑based predictions with your clinical impression.

On‑device Standardized Time‑saving
Case Generator preview

🔬 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 Vascular Occlusion): 82.4%
  • AMD (Age-Related Macular Degeneration): 96.8%
  • VMA (Vitreo-Macular Abnormalities): 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.
OCT Classifier accuracy chart
OCT Classifier statistics

📸 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 (DR): 76.2%
  • Retinal Vascular Occlusion (RVO): 98.1%
  • Age-Related Macular Degeneration (AMD): 80.0%
  • Pachychoroid (CSCR): 90.91%

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.
FFA Classifier accuracy chart
FFA Classifier statistics

⚠️ Clinical Use Disclaimer

mRetinaAI supports, but does not replace, expert judgment. All outputs must be reviewed by a qualified retinal specialist, and treatment should be individualized. To achieve optimal results, it is imperative to adhere to the recommended image-quality guidelines provided within the application to enhance the performance of the classifier.

Ready to get started?

Download, review the privacy policy, or purchase a license to activate on your device.

🤖 Download for Android 🛡️ Privacy Policy 🔑 Buy License
mRetina Ecosystem

Elevate Your Practice

Beyond clinical tools, master retina with video courses, digital textbooks, and clinical manuscripts.

mRetina School

mRetina School

Professional development through curated video courses with self-assessment tests and cases.

Start Learning →
mRetina Books

mRetina Books

A digital library of 10+ clinical textbooks and atlases authored for retina specialists.

View Collection →
📖

Publications

An extensive database of peer-reviewed articles, case reports, and clinical manuscripts.

Read Articles →