Adapt an image model to your CT or MRI scans.
What it does
Fine-tunes the NV-Generate-CTMR MAISI autoencoder on CT or MRI volumes to produce model checkpoints and a JSON run summary.
When to use it
Use to fine-tune the MAISI VAE on CT or MRI NIfTI training volumes, not for clinical interpretation or production-data approval.
How to use it
Give it a datalist and CT or MRI volumes; it validates and stages the inputs, then produces checkpoints and a run summary when GPU training is explicitly requested.
What you provide
- the datalist path
- the image data directory
- the output directory
- Your files
Uses
What you need · 3
Requires Python 3.10 or later.
Real training requires a local upstream checkout containing the VAE configs and helper APIs; the instructions provide a pinned checkout setup when none is supplied.
Access · 8
This skill
GitHub
Fetches the upstream training repository
https://huggingface.co
Contacts Hugging Face