Skip to content

HANDOFF.md

Latest commit

 

History

History
100 lines (72 loc) · 2.5 KB

HANDOFF.md

File metadata and controls

100 lines (72 loc) · 2.5 KB

Frontend Handoff — NeuraScan

Everything you need to connect the React frontend to the backend API.

1. Running the backend

cd backend
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt
uvicorn main:app --reload

The API will be available at http://localhost:8000. Interactive docs (Swagger) at http://localhost:8000/docs.

CORS is open (*), so requests from http://localhost:5173 (Vite dev server) will work out of the box.

2. Endpoint

POST /upload-mri

Accepts a brain MRI image and returns the model's classification.

Request:

  • Method: POST
  • Content-Type: multipart/form-data
  • Field name: file
  • Value: an image file (JPEG, PNG, etc.)

Example fetch call:

const formData = new FormData();
formData.append("file", selectedFile); // File from <input type="file">

const res = await fetch("http://localhost:8000/upload-mri", {
  method: "POST",
  body: formData,
});
const data = await res.json();

Response (200 OK):

{
  "predicted_class": "meningioma",
  "confidence": 0.9231,
  "probabilities": {
    "glioma": 0.0312,
    "meningioma": 0.9231,
    "notumor": 0.0198,
    "pituitary": 0.0259
  }
}

Error (400): returned when the uploaded file is not an image.

{ "detail": "File provided is not an image." }

3. Response fields

Field Type Description
predicted_class string One of: glioma, meningioma, notumor, pituitary
confidence float Probability of the predicted class (0–1)
probabilities object Per-class probabilities, keys are class names, values sum to ~1.0

4. Using probabilities for the 3D visualization

The probabilities object gives you a score for each of the four classes. You can use these to drive the 3D brain visualization — for example:

  • Highlight the predicted region on the brain model.
  • Use confidence to control opacity or color intensity.
  • Show a breakdown bar/chart alongside the 3D view.

5. Class descriptions

Class Meaning
glioma Tumor originating in glial cells
meningioma Tumor on the membranes covering the brain
notumor Healthy brain scan, no abnormalities
pituitary Tumor in the pituitary gland at the base of the brain

6. Health check

GET / returns:

{ "message": "NeuraScan API is running! Use /upload-mri to classify an image." }

Use this to verify the backend is up before enabling the upload button.