INTContour · AI-Native Auto-Contouring Platform

FDA-cleared AI auto-contouring — plus the tools to train it on your own data

INTContour ships with FDA 510(k)-cleared models covering 120+ OARs and lymph node groups across CT, MRI, and PET/CT. On top of that, your team can train custom segmentation models on your own institutional data — the capability the Mayo Clinic study used to outperform vendor baselines on its own protocols.

Request a demo Download the one-pager →

Why teams pick INTContour

Cleared base models, customizable for your clinic

FDA 510(k)-cleared base

Ready to deploy on day one. 120+ OARs across CT, MRI, and PET/CT. Eclipse TPS integration, stand-alone web, automated DICOM workflows.

Train on your own data research-use

When a vendor model doesn't fit your protocol — iodinated spacers, rectal balloons, custom labels — your team retrains on local data without writing code. Mayo Clinic proof point ↓

QA on every contour, one click

Dice, Mean Surface Distance, and Hausdorff Distance against ground truth — built into the same tool that produced the contour.

Available segmentations

Four anatomical regions. CT, MRI, PET/CT

Approved structures plus investigational-use structures across the four major regions covered today.

Head and neck OAR contouring example — Head & Neck

Mandible, Brachial Plexus, Brain, Brainstem, Cochlea, Eye, Lacrimal Gland, Submandibular Gland, Larynx, Optic Chiasm, Optic Nerve, Parotid, Pharynx, Pituitary, and more. Investigational: GTV Lymph Node, GTV Primary Tumor, Cerebellum, Neck Levels Ib/II/III/IVa, NRG-standard variants.

Thoracic CT contour example with lung, heart and substructures, esophagus, and spinal cord — Thorax

Lung, heart and cardiac substructures, esophagus, spinal cord, and chest-wall structures.

Abdomen-pelvis CT contour example with liver, kidneys, and bowel structures — Abdomen

Liver, kidneys, stomach, duodenum, pancreas, bowel, spleen, and adrenal glands.

Axial CT slice of the pelvis with prostate, bladder, and rectum contoured — Pelvis

Male: prostate, seminal vesicles, bladder, rectum, femur heads, penile bulb, urethra. Female: uterus, cervix, ovaries, bladder, rectum, vagina, sigmoid, bowel bag.

Robust quality assurance

One click. Three metrics. Every contour

INTContour automatically generates quality assurance reports based on user selections to validate segmentation. Calculate Dice Score, Mean Surface Distance, and Hausdorff Distance between AI outputs and ground truth values with a single click.

INTContour quality-assurance panel showing per-structure Dice score, Mean Surface Distance, and Hausdorff Distance bars

Powerful workflow connectivity

Stand-alone web. Eclipse TPS. Automated DICOM

INTContour operates as a stand-alone web-based system, or integrates with applications via Automatic Workflow or TPS Connectivity. Using Automatic Workflow, INTContour efficiently segments the desired studies and transfers them back to the main application in use.

INTContour native web interface for browser-based contouring — Native web

INTContour Eclipse TPS integration via ESAPI — Eclipse TPS scripting

INTContour automated DICOM listener monitoring incoming studies — Automated DICOM

Research tools

Made for generating new insights

Streamline your research with intuitive features to train models without coding and process large volumes of studies with a single command using award-winning AI algorithms.

Interactive Study Viewer

Effortlessly refine segmentations directly in your browser before sending them to your Treatment Planning Station.
Use INTContour's comparison tool to view annotation differences across series within a specific study.

INTContour browser viewer with contour-refinement overlay on an axial CT slice

Advanced Editing & Analysis

Calculate gross total volume of lymph nodes and primary tumors from PET/CT studies.
AI-assisted slice-to-slice interpolation lets clinicians contour a few key slices and have the intermediate ones filled in, cutting annotation time.

INTContour PET/CT contouring with target volume calculation

Segmentation Protocol Customization

Develop new protocols by adjusting regions of interest, naming conventions, and morphological operations to match specific needs.
Customize Treatment Planning System (TPS) settings for new protocols to seamlessly align with your workflow preferences.

INTContour protocol customization panel — adjust ROI names, expansions, and TPS export labels

Incremental Learning

INTContour trains new models using your on-site data.
Optimize your models with a robust backbone network structure to support any structures you need.^*
Allows for fused multi-modality inputs with registration.

^* Tools provided as a research-use feature. INTContour does not assure the accuracy and performance of newly trained models.

INTContour incremental learning workflow — train new models from on-site cases

Batch Study Processing

Leverage your workstation to efficiently parse local folders with multiple studies and output specified contours for all data at once.
Enhance image quality using CBCT to CT translation and generate synthetic CT from MRI.
Replace Protected Health Information (PHI) with anonymized values in longitudinal studies while preserving all DICOM-RT data links.

INTContour batch processing workflow — DICOM in, contours out across a queue of studies

Award-winning algorithm lineage

1st place · 2019 AAPM MRI Head & Neck Segmentation Challenge
1st place · 2018 MICCAI Brain Tumor Segmentation Challenge (BraTS)
2nd place · 2017 AAPM Thoracic Organ Segmentation Challenge

Proof point · Mayo Clinic

When the vendor model isn't enough

The strongest evidence we have for incremental learning.

Setting. Prostate planning with iodinated spacer and rectal balloon — anatomy not represented in vendor training data. Baseline. Three major vendors' built-in deep-learning auto-segmentation models underperformed on prostate, seminal vesicles, bladder, rectum, and femur heads. Intervention. INTContour incremental retraining on 100 cases. Readout. Blinded 5-point rating by 6 GU radiation oncologists plus 2 RO residents, 115 test cases.

≥ manual in 50%+

of cases across every group and reviewer — range 63.8% (SV, junior resident) to 100% (bladder, RO1).

Read the full case study →

Duan et al., Medical Physics, 2023.
Duan et al., Radiotherapy & Oncology, 2025.