Tumor Slice Classifier (Browser Demo)

Runs fully in your browser (static Netlify site). Educational demo only — not medical advice.

1) Load model

Place TFJS files at /model/model.json.

Model not loaded.


2) Choose a slice image

Use the same kind of slice images your model was trained on (single slice, grayscale).


2b) Or pick a built-in sample

These samples are bundled with the site from your BraTS raw volumes.


3) Predict

Waiting…

Threshold: 0.50

Preview

preview

Output

Tumor probability
If prediction seems wrong, it’s usually one of: wrong input size, wrong normalization, or model expects 3 channels instead of 1.

Config (edit these to match training)