Project at a glance
The Smart Lead Tech team built the Diabetes Vision Detector to support clinics and hospitals in the early screening of diabetic retinopathy — one of the most common diabetes complications. The system analyses fundus images using advanced computer-vision models and produces a clinician-readable medical report.
Highlight features
- Smart image analysis: detection of early signs (haemorrhages, exudates, abnormal vessels).
- Stage classification: NPDR / PDR per international standards.
- Clinical report: a structured PDF summarising risks and clinician recommendations.
- DICOM integration: direct import from retinal-imaging devices.
- Patient record: link screening results to the patient’s medical history.
- Bilingual: Arabic and English UI and reports.
- Medical-grade security: HIPAA compliance with AES-256 encryption for images and data.
Technology
- Deep learning: CNNs trained on thousands of labelled images (EyePACS / APTOS).
- Computer vision: region segmentation and pathological-sign detection.
- Backend: FastAPI + PyTorch, Laravel for the admin UI, PostgreSQL for records.
Outcome: detection accuracy above 92% in validation, with initial-screening time under 30 seconds.
Engineered by the Smart Lead Tech team — clinical-research-grade medical AI.