Admin token
Needed to view or change settings. Use the ADMIN_TOKEN you set, or the one generated on first start
(see docker compose logs app, or data/admin_token).
Which model writes the draft?
Pick one of your servers, then one of the models that server offers (the list comes from the server itself). Servers are managed on the Servers tab.
Main model
Backup model (optional: used only if the main one fails)
Advanced: use a different model for certain kinds of study
For example, send CT studies to a different model. A study that matches a rule uses that model instead of the main one.
Servers
A server is a place where models run: Ollama (in Docker or on another computer), a local server like GenieX, or a cloud service. You choose which model to use on the Models tab.
Add a server
PACS connections (DICOMweb)
Each connection has its own address and its own sign-in. An access key can use only the PACS connections it is mapped to (API access tab). Tokens and secrets are saved on the server and are never shown again. Changes apply immediately (no Save bar).
Audit trail
Who did what, when and with what result. It holds no patient data, drafts, history, file names or search terms. A study appears only as a short one-way reference: type a study ID in the filter to find every entry for it.
| Time | Who | Event | Result | Source | Address | Study ref | Details |
|---|
API access (PACS / RIS systems)
Each system that calls this service gets its own key (developer guide: docs/API.md). A key is shown once, when created.
Local only means that system can never cause images to be sent to a cloud service. Changes here apply immediately.