OPD Scan QC
Quality control for scanned hospital records: an OpenCV engine scores every page, flags rescans and hands AI transcriptions to a human reviewer.
- Category: Healthcare AI
- Client: Subharti University Hospital
- Role: Full Stack Developer
- Status: live
- Built with: FastAPI, React, OpenCV, Celery, PostgreSQL, Expo
A records-digitisation team scans thousands of paper case files. OPD Scan QC measures how readable each page actually is, explains every verdict, flags pages that need rescanning and attaches the rescan to the page it replaces. Optional AI providers locate handwriting and transcribe any diagnosis already written on the page, and a reviewer confirms or corrects every result. It ships with a web app, an Expo mobile app and exports for the records office.
What it does
- Offline OpenCV quality engine with 14 defect types and a plain-English reason for each
- Versioned pages: a rescan replaces its page instead of creating a duplicate
- Handwriting and diagnosis transcription through Google Document AI, Azure or Tesseract
- Append-only human review that keeps the original machine output
- CSV, Excel, PDF and ZIP exports, plus an Android and iOS app
- 231 automated backend tests