XNAT Imaging Extractor

Point at any public XNAT server and pull structured neuroimaging metadata over the standard XNAT REST API. XNAT is the de-facto open-source platform for neuroimaging & clinical imaging research data management, and every install — the NITRC image repository, the BMIA public archive, central.xnat.org, and countless institutional deployments — speaks the identical /data/... REST surface. So there is no per-server scraper: give a base URL and a mode, get clean rows.

Modes

mode	Returns	Endpoint
projects	one row per project (whole server)	/data/projects
subjects	one row per subject in a project	/data/projects/{projectId}/subjects
experiments	one row per imaging session in a project	/data/projects/{projectId}/experiments
scans	one row per scan in a session	/data/experiments/{experimentId}/scans

• subjects and experiments need a projectId.
• In experiments mode, add a subjectId to narrow to one subject.
• scans needs an experimentId.

Input

{
  "serverUrl": "https://www.nitrc.org/ir",
  "mode": "projects",
  "maxRecords": 1000
}

Drill down further:

{ "serverUrl": "https://www.nitrc.org/ir", "mode": "subjects",    "projectId": "ixi" }
{ "serverUrl": "https://www.nitrc.org/ir", "mode": "experiments", "projectId": "studyforrest_rev003" }
{ "serverUrl": "https://www.nitrc.org/ir", "mode": "scans",       "experimentId": "NITRC_IR_E07478" }

Field	Type	Notes
serverUrl	string (required)	XNAT base URL. Include any path prefix — NITRC serves under /ir.
mode	enum	projects | subjects | experiments | scans (default projects).
projectId	string	Required for subjects / experiments.
subjectId	string	Optional drill-down in experiments mode.
experimentId	string	Required for scans.
maxRecords	integer	Row cap (default 1000).
bearer	string (secret)	Optional bearer token for authenticated servers.
extraHeaders	object	Optional extra HTTP headers (e.g. a session cookie).

Output

Each row carries the mapped, human-friendly fields for its mode (id, name, label, project, date, scan type, series description, …) plus a lossless _raw object with the complete original API record, so nothing is dropped even when a column key differs between XNAT versions.

{
  "id": "NITRC_IR_E07478",
  "label": "sub001",
  "project": "studyforrest_rev003",
  "date": "",
  "xsiType": "xnat:mrSessionData",
  "subjectId": "NITRC_IR_E07478",
  "uri": "/data/experiments/NITRC_IR_E07478",
  "_server": "https://www.nitrc.org/ir",
  "_mode": "experiments",
  "_raw": { "...": "complete original record" }
}

Notes on public vs gated servers

?format=json is a request, not a guarantee. Some XNAT servers gate public browse behind an interactive login and answer with an HTML login page at HTTP 200 (e.g. central.xnat.org, cnda.wustl.edu). This actor detects that (non-JSON body / no ResultSet envelope) and fails fast rather than emitting an empty dataset — it never fabricates rows and never reports a gated server as an honest zero. Servers that expose anonymous public projects (NITRC-IR, BMIA) return real JSON and work out of the box.

Pricing

Pay-per-event: $0.0005 per record (one project / subject / experiment / scan row). You only pay for rows actually extracted.