Questions clinical teams ask

No identifiable data does. A lightweight on-premises DICOM gateway sits inside your hospital network and anonymises every series before any transfer. Only anonymised DICOM crosses the perimeter, over a TLS 1.3 encrypted tunnel, and nothing is retained in the cloud after inference completes.

Your PACS, TPS and imaging modalities stay entirely on-premises. Nothing about your existing clinical workflow changes.

ContourOAI deploys on region-specific cloud infrastructure so you can meet data-residency requirements — EU, Mumbai or Hyderabad regions, depending on your obligations. Combined with full anonymisation before transfer and zero retention post-processing, the pipeline is GDPR-compliant.

Yes. For air-gapped environments ContourOAI deploys entirely on your own hospital GPU servers — the same workflow and the same interface, with zero cloud dependency.

A mandatory clinician review gate. No RT Struct leaves the platform without explicit, structure-by-structure sign-off, and the gate is a hard architectural constraint — it cannot be configured away or bypassed. ContourOAI operates strictly as a decision-support tool.

Every approval is written to a timestamped, identity-linked audit log, and AI-generated versions are tracked separately from clinician-edited ones.

ContourOAI is pure DICOMweb (STOW-RS / WADO-RS), so it works with any DICOM-compatible PACS or TPS — including Eclipse, RayStation and Monaco, and PACS such as Orthanc, Sectra and Synapse. There is no plugin to install, no vendor agreement, and no version coupling.

DSC and volumetric metrics are computed per structure and built into the platform, so quality can be benchmarked on your own cases rather than taken on trust. Batch processing across large cohorts and an integrated anonymisation pipeline mean the same tooling supports prospective and retrospective clinical studies with full audit trails.

Never. Custom models are institution-private: your training data stays in your environment and is never used to train another customer's model. A model fine-tuned on your cases is available only to you, alongside the base OAR models.

Yes. ContourOAI is not a closed black box. Annotate your own cases in the built-in Contour Station — any structure, from unusual OARs to GTV and CTV tumour volumes — and submit the dataset to the training pipeline. The resulting model captures your patient population, scanner protocols and delineation style, and is selectable at contouring time.

The annotation interface used clinically is the same one that produces training data, so there is no separate labelling workflow to maintain.

Only the on-premises DICOM gateway, which is a lightweight agent in your network. Clinicians themselves need nothing — ContourOAI is fully browser-based and works from any computer or tablet.

Head & neck OARs complete in under 3 minutes, with other treatment sites at matching GPU-accelerated speed. In batch mode, roughly 20 patients queue and process in about an hour — upload at end of day, review a full set in the morning.

These timings assume a stable, low-latency network connection; actual processing time varies with network latency, image size and server load.

Yes, both. For education it provides AI-generated reference contours for residents and fellows to compare against, critique and correct, with side-by-side trainee, AI and consultant comparison and tracked edits for structured feedback.

For research it works as a complete instrument — batch processing across cohorts, per-structure DSC and volumetric metrics, an end-to-end anonymisation pipeline, and annotated dataset export for publication or peer review.