File Types

How is data ingested into Kaapana and what file types are supported?

Files

Binary models, documents, PDFs, and other non-DICOM files are uploaded through the MinIO web interface (see Store), where they land in a bucket and are managed as ordinary objects. Workflows access them through the same underlying storage.

DICOM

DICOM file is stored in the internal PACS via DIMSE C-STORE or DICOMweb STOW-RS. Both paths converge on the Clinical Trial Processor (CTP), which receives the DICOM before the service-process-incoming-dcm ingestion workflow (Airflow DAG) picks it up, validates it, extracts and indexes metadata, sets the correct access rights for the series, and creates a thumbnail. Other automatic pipelines can be autotriggered here — pseudoanonymization, whole-slide microscopy conversion, or radiomics feature extraction. The ingestion DAG then hands the DICOM off to the PACS for storage and indexing.

Data API (Work in Progress)

The Data API will eventually catalog all file types as data entities — a filesystem path, an S3 object, PACS coordinates, or a plain URL — with metadata validated against a JSON Schema. Workflow will be responsible for getting the data from the coordinates provided by the Data API. This gives all file types the same guarantees: schema-validated metadata, a stable identifier, and discoverability.