DICOMweb QIDO-RS Extractor

Point at ANY DICOMweb server and pull medical-imaging metadata into clean, structured rows — study, series or instance level — with one Actor.

QIDO-RS (Query based on ID for DICOM Objects, RESTful Services) is the DICOM PS3.18 web standard for querying imaging metadata. Every conformant PACS / VNA / cloud imaging store — Orthanc, dcm4chee, Google Cloud Healthcare API, DICOMcloud, and many more — exposes the same uniform /studies, /series, /instances wire grammar. This Actor speaks that grammar generically: give it a base URL, a query level, and optional QIDO filters, and it returns normalized rows from any of them. No pixel data is ever fetched — metadata only.

What it does

• Three query levels (mode):
  • studies — one row per matching study.
  • series — one row per series (across all studies, or scoped under a StudyInstanceUID).
  • instances — one row per SOP instance (all, under a study, or under a specific series).
• Standard QIDO filters: PatientID, PatientName, StudyDate (single or range), ModalitiesInStudy, AccessionNumber, plus includefield, fuzzymatching, and any extra query params the server supports.
• Decodes DICOM+JSON for you. Results come back tag-keyed (00100010 = PatientName) with {vr, Value[]} structures. This Actor maps the common tags to readable field names, normalizes Person-Name values (Last^First → First Last), coerces numeric attributes — and keeps the complete raw tag object in _raw plus a name-keyed tags projection, so nothing is ever lost.
• Multi-server: pass several base URLs to harvest a whole fleet in one run.
• Paginated & de-duplicated via limit/offset with per-level UID dedup.

Example input

{
  "dicomWebUrls": ["https://demo.orthanc-server.com/dicom-web"],
  "mode": "studies",
  "ModalitiesInStudy": "CT",
  "StudyDate": "20060101-20061231",
  "maxRecords": 500
}

Example output (one study row, abridged)

{
  "_source": "https://demo.orthanc-server.com/dicom-web",
  "_level": "studies",
  "StudyInstanceUID": "2.16.840.1.113669.632.20.1211.10000315526",
  "PatientName": "VIX",
  "PatientID": "vAD7q3",
  "StudyDate": "20061005",
  "ModalitiesInStudy": "CT",
  "AccessionNumber": "0",
  "NumberOfStudyRelatedSeries": 1,
  "NumberOfStudyRelatedInstances": 250,
  "tags": { "PatientName": "VIX", "Modality": "CT", "...": "..." },
  "_raw": { "00100010": { "vr": "PN", "Value": [ { "Alphabetic": "VIX" } ] }, "...": "..." }
}

Input reference

Field	Type	Notes
dicomWebUrls	array (required)	One or more DICOMweb roots (the part before /studies).
mode	select	studies | series | instances. Default studies.
studyInstanceUID	string	Scope series/instances under one study.
seriesInstanceUID	string	With a study + instances, scope under one series.
PatientID, PatientName, StudyDate, ModalitiesInStudy, AccessionNumber	string	Standard QIDO match filters.
includefield	array	Extra attributes to return, or all.
fuzzymatching	boolean	Approximate name matching (server-dependent).
extraQueryParams	object	Any additional QIDO match keys.
pageLimit	integer	Page size (limit). Default 100.
maxRecords	integer	Row cap across all sources. Default 1000.
bearerToken	string (secret)	Optional OAuth2 bearer for protected servers (e.g. Google Cloud Healthcare).
extraHeaders	object	Optional extra request headers.

Notes

• Sends Accept: application/dicom+json on every request, so servers that default to multipart XML still return JSON.
• Authentication is optional and never required or logged for public servers.
• A study/series with zero children, or a query matching nothing, is a real empty answer. A run where no source returns anything fails fast so you never get a silently-empty result.
• Metadata only. This Actor does not retrieve pixel data (that is WADO-RS, a different service).

Pricing

Pay-per-event: you are charged per record emitted. Subscription-tier discounts apply automatically.

Your spending limit is respected. If you set a maximum charge for the run, the Actor delivers only as many records as that limit pays for, then stops without making further requests to the DICOMweb server. Every row in the dataset is one charged record, and nothing beyond your limit is written. If the limit is below the price of one record, the run ends with 0 records and no charge.