Plugins

Kantrad’s core viewer + reporting workflow is built-in. Heavier features that not every user needs - like AI organ segmentation - ship as optional plugins so you only download them if you’ll use them.

Where to manage plugins

Open Settings → Plugins.

You’ll see one card per available plugin with:

• A short description of what it does.
• The current state: Not installed, Installing…, Installed, or Disabled.
• An Install / Uninstall button.
• A toggle to enable / disable the plugin without uninstalling.

Available plugins

MONAI Segmentation

AI-powered organ and lesion segmentation, powered by NVIDIA’s MONAI Label framework. Adds:

• A Segmentation tab in the right-side panel for the active study.
• One-click presets for common targets (whole-body organs, lung lobes, spine vertebrae, brain structures, etc.).
• Editable masks that can be re-run, refined, or exported as DICOM SEG for downstream tools.

Install size: ~500 MB for the MONAI server + per-bundle model weights downloaded on demand (typically 100–800 MB each).

Hardware: Runs on CPU, but a CUDA-capable NVIDIA GPU or Apple Silicon is strongly recommended for usable inference times.

For step-by-step usage instructions once installed, see the Segmentation guide.

How plugins are isolated

Plugins run as sidecar processes managed by Kantrad - not inside the main app. This means:

• A plugin crash doesn’t take down the viewer or your unsaved report.
• Plugins can use their own runtime (e.g. MONAI uses Python + PyTorch under the hood) without polluting Kantrad’s environment.
• Uninstalling a plugin removes its files cleanly - there’s no “leftover state” in the main database.

Privacy

Plugins follow the same offline-first rule as the rest of Kantrad: images and patient data stay on your machine. Plugin model weights are downloaded once from a Kantrad-controlled mirror, then run locally.