The study is the unit, not the slice
An MRI or CT examination can contain hundreds or thousands of DICOM instances across localizers, diagnostic sequences, repeats and derived maps. Counting files inflates scale and hides missing series. Dataset inventory is therefore reported in patients, examinations and series.
Series classification
Classification combines DICOM fields with geometry and controlled rules. Free-text SeriesDescription alone is unreliable because naming varies by site and scanner. Ambiguous series are flagged for review instead of silently assigned.
- Study and series grouping with pseudonymous identifiers
- Sequence or reconstruction classification with confidence and evidence
- Slice ordering, orientation, spacing and duplicate checks
- DICOM-to-NIfTI conversion with source mapping
- Preview, intensity and metadata summaries for quality review
Reproducible outputs
Every transformation is versioned. Dataset manifests include input checksums, tool versions, exclusions and the relationship between source DICOM, derived volumes, labels and annotations. A buyer can therefore audit what changed without receiving identity-bearing source records.
Questions, answered directly.
Why keep source DICOM?+
DICOM contains acquisition and provenance information that may be lost in a derived volume. Whether source DICOM is delivered depends on the agreement.
Is NIfTI always required?+
No. It is common for 3D model workflows, but some buyers need source DICOM or both formats.
Can sequence names be inferred from filenames?+
Not reliably. Classification should use approved DICOM fields, geometry and review rules.