For Buyers¶
For organizations acquiring annotated medical imaging datasets — medical device manufacturers (medtech), AI/ML developers and startups, pharmaceutical companies, government health agencies, and academic groups commissioning external data work.
The procurement gap¶
Medical imaging dataset procurement currently relies on vendor claims and manual post-delivery review. There is no shared, machine-checkable definition of what counts as documented provenance, adequate quality, or acceptance-ready structure.
Without an objective acceptance criterion, dataset quality is typically assessed after integration begins — when remediation is most expensive and timelines are already committed.
A compliance benchmark of four widely used public medical imaging datasets, scored against the VIDS specification, illustrates the scale of the gap:
| Dataset | VIDS Compliance |
|---|---|
| BraTS | 39% |
| MSD | 30% |
| LIDC-IDRI | 27% |
| CheXpert | 20% |
Source: VIDS benchmarks repository.
If publicly maintained reference datasets score in this range, privately commissioned datasets — which receive far less external scrutiny — are unlikely to score higher.
Recommended action¶
Include a VIDS validation requirement as an acceptance condition in your next dataset procurement.
The reference contract language is drop-in. It defines acceptance as "zero failed rules under the VIDS Reference Validator at the specified Profile" — an objective, automated, vendor-agnostic criterion.
Reference Procurement Language →
CC BY 4.0. Free to use, adapt, and redistribute.
How VIDS works for buyers¶
The buyer-side adoption surface has four parts. Each is purpose-built and citable in contracts and audits.
Reference Procurement Language — drop-in clauses that make VIDS PASS the acceptance condition for dataset deliveries.
Compliance Evaluation — diagnostic assessment of an existing dataset against the 22 VIDS dimensions. Used to surface gaps in already-acquired datasets or vendor sample deliveries before contracting.
Scoring Rubric — the 22-dimension scoring framework used in compliance evaluations. Published openly so any score is traceable to documented criteria.
Validation Attestation — the signed artifact issued for datasets that pass full validation. A vendor attaches it to a delivery and a buyer forwards it internally as proof of acceptance, supported by the underlying machine-readable Validation Report.
What VIDS gives a buyer¶
An objective acceptance criterion. The validator runs on the delivered dataset and produces a binary pass/fail result. No subjective interpretation, no checklist negotiation, no "trust us" deliveries.
A reference framework citable in contracts and audits. EU AI Act, IMDRF GMLP, and the FDA AI/ML SaMD Action Plan all point toward machine-readable provenance. VIDS provides the structure that makes those requirements concrete.
Vendor independence. The specification is open (CC BY 4.0). The validator is open-source (Apache 2.0). No supplier lock-in at the standard level. Datasets curated under VIDS can be exported to nnU-Net, MONAI, COCO, or flat NIfTI without losing the underlying data or provenance.
Lower cost of failure. A vendor delivery that fails validation is identified at acceptance, not after integration. Remediation occurs while the contractual obligation is still active and the cost falls on the vendor.
Scope of validation. VIDS validates dataset structure, annotation provenance, quality documentation, and ML readiness. It does not assess the clinical correctness of individual annotations, the diagnostic accuracy of any AI model trained on the dataset, or the dataset's suitability for a specific clinical use case. A successful VIDS validation establishes that a dataset is structurally sound, audit-ready, and provenance-traceable — not that the underlying data is clinically correct or fit for any particular purpose.
About operators¶
VIDS is an open standard. Compliance evaluations and attestation issuance are performed by approved evaluation operators. Princeton Medical Systems is the current sole operator and maintains the reference validator, evaluation rubric, and attestation format.
For evaluation requests or procurement-specific inquiries: info@princetonmedicalsystems.com
Reference materials¶
LIDC-Hybrid-100 reference dataset — 100 lung CT volumes with consensus segmentations from four radiologist reads, scored 21/21 on the VIDS Full profile. The publicly available worked example of what a VIDS-compliant dataset looks like. CC BY 4.0.
VIDS arXiv preprint — "VIDS: A Verified Imaging Dataset Standard for Medical AI." Includes the 22-dimension scoring rubric, the benchmark of four widely used public datasets, and the design rationale behind the validation profiles.