Aller au contenu principal

DICOM

Category: Healthcare Standards
Node id: dicom

Standards: HIPAA, GDPR, SOC2

Performs DICOM medical imaging operations including parsing, validation, metadata extraction, anonymization, and format conversion

What it does​

  • DICOM file parsing with tag extraction and transfer syntax detection
  • Structural and required-tag validation
  • Clinical metadata extraction: patient, study, series, and image info
  • PHI anonymization with remove, replace, or hash methods
  • Format conversion to JPEG, PNG, NIfTI, JPEG2000
  • Configurable image quality for conversion output

Use cases​

  • Extract clinical metadata from DICOM files for radiology analytics
  • Anonymize DICOM images before sharing for research
  • Convert DICOM to standard image formats for AI model training
  • Validate DICOM files against required tag lists before archiving

Runtime​

  • Performance tier: medium

Input ports​

NameTypeRequiredDescription
dicom_datatextYesDICOM file data (base64 or raw) — single instance; may also be an array for batch processing
instancesjsonNoArray of DICOM instances for batch processing — each element has inline bytes (dicom_b64/dicom_data) and/or a file_path (e.g. from a dicomweb retrieve). When present, the node processes every instance and returns result.instances[].

Output ports​

NameTypeRequiredDescription
resultjsonNoMain result data from the DICOM operation
metadatajsonNoDICOM metadata including patient, study, and image information
successbooleanNoWhether the operation succeeded

Settings​

Basic​

SettingNameTypeRequiredDefaultAllowed valuesDescription
OperationoperationenumYesparse, validate, extract_metadata, anonymize, convertDICOM operation to perform

Advanced​

SettingNameTypeRequiredDefaultAllowed valuesDescription
Required Tagsrequired_tagsjsonNoArray of required DICOM tag names for validation
Include Pixel Datainclude_pixel_databooleanNofalseInclude pixel data info in metadata extraction
Anonymization MethodmethodenumNoremoveremove, replace, hashMethod for anonymizing DICOM data
Keep UIDskeep_uidsbooleanNofalsePreserve study/series/SOP UIDs during anonymization
Output Formatoutput_formatenumNojpeg, png, nifti, jpeg2000Target format for conversion
QualityqualitynumberNo90min 1; max 100Output quality for image conversion (1-100)

Errors​

CodeDescription
DICOM_DATA_REQUIREDdicom_data input is missing or empty
INVALID_OPERATIONSpecified operation is not supported
MISSING_PREAMBLEDICOM file is missing the required preamble
MISSING_REQUIRED_TAGSRequired DICOM tags are not present in the file
UNSUPPORTED_FORMATOutput format is not supported for conversion

Runtime notes​

  • Output shape under 'result' varies by operation: parse has tags/tag_count, validate has is_valid/errors/warnings, convert has converted_data/output_format, etc.
  • For extract_metadata, the detailed data goes into 'metadata' output (patient_info, study_info, image_info) — 'result' only contains operation name
  • Anonymize returns modified DICOM data in result.anonymized_data — UIDs are regenerated unless keep_uids is true
  • Convert renders the first pixel-data frame to result.converted_data (base64); requires PixelData in the input
  • Array mode: if the input carries a LIST of instances (the instances port, a dicom_data array, or an upstream dicomweb retrieve result.instances[] with file_path/inline bytes), the node processes each and returns result.instances[] (an array) plus result.instance_count/success_count/is_array. file_path instances are read from disk. Per-instance shapes: anonymize → {study_instance_uid, series_instance_uid, sop_instance_uid, file_path} (anonymized DICOM written to a new temp file, no inline bytes); convert → {sop_instance_uid, converted_data (base64), output_format, ...UIDs}; extract_metadata/parse/validate → the same record the single-item path returns, per instance. A single dicom_data string still returns the legacy single-item shape.

See Build any workflow for how to put nodes together and Ports, types and data flow for how port types connect.

Quelque chose n'est pas clair ? Envoyez vos commentaires sur la documentation par courriel.