RetinAI web application - MedTech by Audrey Ntep

Retinal Diagnosis by Image-Based Deep Learning

Windows configuration

1. Install dependencies

pip install requirements.txt

2. Launch flask local server with Powershell

$env:FLASK_APP = "main"

flask run

3. Access web app with navigator

GET http://127.0.0.1:5000/

1. First browse local directory to select one or many OCT images (PNG or JPEG extension only), open

2. Then click "Upload" button, to load images on web server, images are loaded on the left side of the page,

3. Finally click "Diagnose" button to get results :

   • CNV and DME retinas are red

   • Drusen retina is orange

   • Normal retina is green

4. Consumming API

Endpoint	Method	Result
/diagnosis	POST	JSON prediction
/model	GET	JSON model
/metrics	GET	JSON metrics
/classes	GET	JSON classses

5. Data storage

OCT image files are temporary stored in web app server directory : '/static/oct_image'

OCT image files are temporary stored in api server directory : '/data'

Directories are cleared when server is shut down.