Now at SIIM 2026 · Startup Kiosk 3 · June 10–12

Carina AI · clinical-research software for academic medical centers

Meet CuratAI — the AI-Native Multimodal Data Curation Platform

On-premise data curation for clinical research. Pulls from PACS and EHR, removes PHI, hands back structured cohorts that stay inside the institution.

Before Original
After CuratAI defaced
Drag to compare
Facial features come off; brain, skull base, and surrounding anatomy stay intact.
  • 10+ Academic medical centers
    Mayo · JHU · Penn · UVA · Duke · Emory · Colorado · Brown · UTSW · Kentucky
  • 2 FDA 510(k)-cleared products
    INTContour and INTDose · shipping since 2022
  • 40+ Peer-reviewed publications
    Reviewed by the same people who use the software

How it works

What CuratAI does, end to end

Multi-site clinical research lives or dies on data that's safe to share — de-identified at the pixel and traceable end to end. Carina AI builds CuratAI, the on-premise platform for that work, plus the FDA-cleared clinical AI products built on top of it.

  1. 1

    Retrieval

    PACS · EHR · Pathology

  2. 2

    Ingest

    Unstructured → Structured · Local LLM

  3. 3

    De-identify

    Pixel · Metadata · Text

  4. 4

    Annotate

    ROI · Measure · QA

  5. 5

    AI Plugins

    Segmentation · Classification · Custom

  6. 6

    Collaborate

    In-platform · Cross-institution

Explore CuratAI in depth →

What customers build

Real outcomes on the same stack

Mayo Clinic Platform

Uses CuratAI's mask-based defacing to process imaging data on the Mayo Clinic Platform — identity removed, intracranial anatomy preserved.

Mayo Clinic Arizona · Radiation Oncology

Retrained INTContour on local prostate protocols (iodinated spacers, rectal balloons). Blinded review by 6 GU radiation oncologists and 2 RO residents found AI contours ≥ manual in ≥50% of cases across every structure.

Read the full case study → · Duan et al., Medical Physics, 2023

University of Colorado · Anschutz

Uses CuratAI to build foundation models and accelerate multi-modal data retrieval and processing across the institutional research pipeline.

University of Pennsylvania

Benchmarked four defacing algorithms against CuratAI's. CuratAI delivered the best privacy–utility trade-off — preserving organ volumes with no measurable changes across all structures, while still suppressing facial re-identification.

Read the full case study → · Wang et al., Red Journal, 2026

Beyond CuratAI

The full Carina AI suite

Clinical AI products built on the same on-premise stack — two FDA-cleared, two NIH-supported in active research.

FDA 510(k) K212274 · cleared 2022

INTContour

AI-Native Auto-Contouring Platform

AI auto-contouring for radiation oncology — 120+ OARs across CT, MRI, and PET/CT with automatic quality assurance.

FDA 510(k) K213137 · cleared 2022

INTDose

Independent Dose Verification Platform

Monte Carlo 3-D secondary dose verification. Machine-specific compatibility, customizable DVH and gamma analysis.

NIH-supported · multi-modality

OncoAI Suite

AI-Native Oncology Imaging Suite

Detection, segmentation, measurement, and longitudinal tracking on CT, MRI, and PET/CT — LymphNode, BrainTumor, LungNodule.

CT and MRI · Evans Index

QuantBrain

AI-Native Brain Quantification Platform

Automatic ventricle segmentation and quantitative volumetric measurement for hydrocephalus on CT and MRI.

From research data to clinical reality

One on-premise stack — from the first de-identified cohort to the FDA-cleared product.