Skip to main content
Open Standard

VIDS

The Verified Imaging Dataset Standard

An open standard for medical imaging dataset documentation, stewarded by Princeton Medical Systems.

Documentation that can be checked

VIDS defines a structured way to document dataset identity, imaging metadata, annotations, provenance, quality information, machine-learning splits and version identity.

VIDS verifies that documentation is present and structured; it does not certify a dataset. It does not judge clinical quality, model performance, bias, authenticity or fitness for a particular use.

How the standard is organized

The authoritative specification and rule documentation remain at vidsstandard.org.

Profiles

POC covers core dataset structure, imaging metadata and annotation provenance. Full adds required quality, agreement, class-distribution and machine-learning split documentation.

Validator

The open reference validator reports PASS, FAIL, WARN or SKIP per rule. Install validator 1.2.1 with pip install vids-validator.

Extensions

Extensions adapt the documentation framework to additional modalities and use cases. VIDS-Fundus 1.0 is the first published modality extension.

Reference implementations

Public reference implementations show how the specification is applied and preserve the supporting artifacts with persistent DOI records.

Structure → Document → Validate → Version

The validator checks the documentation required by the selected profile. It does not decide whether a dataset is good.

StructureDataset files DocumentRequired metadata ValidateRule results VersionDataset identity

Research and reference implementations

VIDS preprint

Muthu and Shalen. VIDS: A Verified Imaging Dataset Standard for Medical AI. arXiv 2604.17525.

LIDC-Hybrid-100

A public VIDS-Full reference implementation with its supporting records. DOI 10.5281/zenodo.19582717.

VIDS-Fundus 1.0

A modality extension and metadata-only reference implementation. Specification DOI · Metadata DOI.

VIDS is governed in public

Read the governance process, follow decisions on GitHub, or contact the standards team at standards@vidsstandard.org.

How VIDS Is GovernedVIDS on GitHub

Read the full specification

The specification, validator documentation and governance records are maintained at vidsstandard.org.

Read the Standard