Synthsis Labs

Perception

The scan, and the doctor’s reasoning.

De-identified imaging paired with what the physician saw, considered and decided, for clinical AI.

See the data

One read, start to finish.

  1. Study

    Chest CT, follow-up for a known nodule.

  2. Impression

    Nodule stable, with a new ground-glass opacity nearby.

  3. Differential

    Infection considered. Early malignancy not excluded.

  4. Decision

    Short-interval follow-up ordered.

  5. Outcome

    Linked to the next study when it happens.

An illustrative example, simplified.

Every read holds

  • Imaging, in DICOM.
  • Region markings and measurements.
  • Impression, differential and decision.
  • Clinical context and later outcomes.

Train on the read, not just the image.

Collected with hospitals. Never scraped.

Collected with partner hospitals under institutional agreements and ethics review. Physicians who annotate are paid for each read. Nothing leaves our systems until it has been de-identified and checked by a person.

Privacy rules

  • Identifiers removed with the safe harbor method.
  • Demographic combinations checked for re-identification risk.
  • Automated screens, then human review, before any delivery.
  • Prospective collection only under ethics review.
How provenance works

What it trains

  • Diagnostic models
  • Report generation
  • Clinical decision support

In the marketplace

All of the marketplace

  • CT reads with reasoning

    De-identified studies paired with the radiologist’s impression, differentials and follow-up decision at the time of the read.

    DICOM, reads, differentials

  • Radiographs with reports

    X-rays across orthopedics, chest and emergency care, with structured reports and demographic context.

    X-ray, structured reports

  • Ultrasound with sonographer notes

    Full sweep recordings with probe settings, measurements and annotated findings.

    ultrasound video, measurements, notes

  • Hospital operations events

    De-identified workflow events from working hospitals: orders, handoffs and escalations.

    event streams

  • Prospective clinical cohorts

    Collected against your protocol, from modality and population to annotation schema, under ethics review.

    protocol-defined

Need a cohort we don’t have?

Prospective collection against your protocol, from modality and population to annotation schema.