What the programme covers
CT appearance changes with dose, reconstruction, slice thickness, contrast timing and field of view. A cohort suitable for one lung or cardiovascular task may be unsuitable for another even when the anatomy overlaps.
What a useful specification includes
A defensible request defines the clinical task, source evidence and acceptance criteria before patient-level data moves. The exact fields and thresholds depend on the intended model claim.
- Anatomical coverage and target finding
- Non-contrast or required contrast phases
- Scanner, tube settings and reconstruction kernel
- Slice thickness, spacing, dose and field of view
- Reference measurements, reports or outcomes
- Site and scanner holdout requirements
Quality and validation controls
CT quality review verifies series identity, acquisition constraints, completeness and the reference-standard pathway for each target label or measurement.
- Contrast and protocol classification
- Reconstruction and slice-thickness distributions
- Coverage, truncation and artefact review
- Dose and acquisition metadata completeness
- Study duplication and longitudinal handling
Availability, rights and delivery
Approved CT releases can include source DICOM, structured acquisition metadata, derived volumes, manifests and task-specific clinical evidence.
Public pages describe a sourcing and engineering capability, not guaranteed ready inventory. Each release remains subject to verified programme inventory, programme-specific authorization, privacy review, technical acceptance and buyer licence terms.
Questions, answered directly.
Can contrast and non-contrast CT be mixed?+
Only when the model task and analysis plan support that mixture. Otherwise they should be separated.
Can one chest CT support multiple AI tasks?+
Potentially, but each task requires its own inclusion rules, labels and reference standard.
Is non-contrast CT currently a ready catalogue?+
No. It is a published sourcing scope subject to verified inventory and programme controls.
Can CT data be used for external validation?+
Yes when cohort independence, acquisition coverage and ground truth meet the intended validation claim.