< 3 min

Head & neck OARs

225+

Structures across all model groups

20

Head & neck nodal levels

0

Seed points required

All other treatment sites run at matching GPU-accelerated speed. Timings measured under a stable, low-latency network connection; actual processing time varies with network latency, image size and server load.

What it does

One unified platform.

From automated contouring to fine-tuning, training and deployment — every step integrated into a single workflow, not stitched together from separate tools.

OAR auto-contouring

Deep learning segmentation across 225+ structures — from cranial and head & neck through thorax, abdomen and pelvis. Every major treatment site, one platform.

  • Cranial & brain structures
  • Head & neck OARs — glands, vessels, muscles, bones
  • Thorax — lungs, heart, great vessels
  • Abdomen & pelvis — organs, bowel, reproductive structures
  • Skeleton — vertebrae, ribs, sternum, pelvis, long bones

Lymph node region contouring

Atlas-consistent nodal volumes delineated automatically — eliminating the most variable and time-intensive step in head & neck radiotherapy planning.

  • Bilateral levels IA, IB, IIA, IIB, III, IV, VA, VB
  • Retropharyngeal and level VI nodes
  • Consistent between clinicians and between sites

Contour Station

A cloud-based annotation workspace serving three purposes in one tool: correcting AI output, contouring manually from scratch, and creating training data.

  • Slice-by-slice brush and eraser correction
  • Full manual contouring of any structure
  • Interpolation between manually edited slices
  • Structure addition, splitting and merging

Mandatory review gate

No RT Struct leaves the platform without explicit clinician sign-off. The gate is a hard architectural constraint — it cannot be bypassed.

  • Per-structure approval workflow
  • Timestamped, identity-linked audit log
  • AI-generated vs. clinician-edited versions tracked

Education & training

Accelerates radiotherapy education for residents, fellows and practising clinicians by providing AI reference contours to compare against, critique and correct.

  • AI contour as a teaching baseline
  • Side-by-side trainee, AI and consultant comparison
  • Annotation tools for supervised correction exercises
  • Tracked edits support structured feedback

Research & clinical studies

A complete research instrument, from DICOM ingestion through to annotated dataset export, with full audit trails and DSC benchmarking built in.

  • Batch processing across large patient cohorts
  • DSC and volumetric metrics computed per structure
  • Export annotated datasets for publication or peer review
  • Anonymisation pipeline integrated end to end

One tool, two purposes. The annotation interface used clinically is also the training data creation tool — annotated cases flow directly into the custom ML pipeline, eliminating any separate labelling workflow.

Custom ML training

Train your own structures. Even tumours.

ContourOAI is not a closed black box. Institutions can upload their own annotated datasets and fine-tune the model — adding custom structures, institutional protocols, or tumour contours specific to their patient population.

  1. 1

    Annotate your data

    Use Contour Station to label your own CT cases. Any structure — from unusual OARs to GTV and CTV tumour volumes — can be annotated and curated as training data.

  2. 2

    Train a custom model

    Submit the annotated dataset to the training pipeline. Models are fine-tuned on your institution's own cases, capturing your patient population, scanner protocols and delineation style.

  3. 3

    Deploy your model

    Your model becomes available alongside the base OAR models. Select it at contouring time; it runs with the same speed and DICOMweb integration as the standard workflow.

Custom models are institution-private. Your training data never leaves your environment and is never used to train another customer's model.

Infrastructure & security

Cloud-powered. On-prem secure.

ContourOAI runs as a cloud AI service connected to your hospital via a secure on-premises DICOM gateway. Anonymisation happens before any data leaves your network — zero patient-identifiable data is retained in the cloud.

Patient data privacy

Full anonymisation before transfer. GDPR-compliant. Zero identifiable data retained post-processing.

Cloud data residency

Deploy on region-specific infrastructure for data residency — EU, Mumbai or Hyderabad regions, based on your requirements.

No TPS plugin required

Pure DICOMweb. Works with any DICOM-compatible PACS or TPS — no vendor agreements, no plugin installs.

Full on-prem option

For air-gapped environments, ContourOAI deploys fully on-premises on hospital GPU servers. Same workflow, zero cloud dependency.