正文
获取原文件references/tcia_tool_reference.md
All 10 tools live-tested against the real TCIA API
(https://services.cancerimagingarchive.net/nbia-api/) on 2026-09-17.
Every response below is real output (trimmed for length), not fabricated.
TCIA scope: public, de-identified cancer imaging metadata. These tools
return metadata only — collection/patient/study/series descriptors,
never pixel data. Dates are frequently shifted for de-identification
(observed: StudyDate: "2000-01-01 00:00:00.0" on a real LIDC-IDRI study).
1. TCIA_list_collections
No parameters.
tu run TCIA_list_collections '{}'
{"status": "success", "data": [
{"Collection": "4D-Lung"},
{"Collection": "A091105"},
{"Collection": "ACNS0332"},
{"Collection": "ACRIN-6698"},
{"Collection": "ACRIN-Contralateral-Breast-MR"}
]}
Hundreds of collections exist. TCGA-linked collections follow
TCGA-<CANCERTYPE> (e.g. TCGA-GBM, TCGA-LUAD, TCGA-BRCA); others use
study-specific names (e.g. LIDC-IDRI for the Lung Image Database
Consortium lung-nodule CT collection).
2. TCIA_get_patients
Params: Collection (optional; omit for all collections).
tu run TCIA_get_patients '{"Collection":"LIDC-IDRI"}'
{"status": "success", "data": [
{"PatientId": "LIDC-IDRI-0001", "PatientName": "", "Collection": "LIDC-IDRI",
"Phantom": "NO", "SpeciesCode": "337915000", "SpeciesDescription": "Homo sapiens"},
{"PatientId": "LIDC-IDRI-0002", "PatientName": "", "Collection": "LIDC-IDRI",
"Phantom": "NO", "SpeciesCode": "337915000", "SpeciesDescription": "Homo sapiens"}
]}
Phantom: "YES" marks QA test objects, not real patients — filter these
out when the user wants a human-subject cohort count.
3. TCIA_get_patient_studies
Params: Collection, PatientID, StudyInstanceUID (all optional; combine
to narrow).
tu run TCIA_get_patient_studies '{"Collection":"LIDC-IDRI","PatientID":"LIDC-IDRI-0001"}'
{"status": "success", "data": [
{"StudyInstanceUID": "1.3.6.1.4.1.14519.5.2.1.6279.6001.175012972118199124641098335511",
"StudyDate": "2000-01-01 00:00:00.0", "PatientID": "LIDC-IDRI-0001",
"PatientName": "", "Collection": "LIDC-IDRI", "SeriesCount": 1},
{"StudyInstanceUID": "1.3.6.1.4.1.14519.5.2.1.6279.6001.298806137288633453246975630178",
"StudyDate": "2000-01-01 00:00:00.0", "PatientID": "LIDC-IDRI-0001",
"PatientName": "", "Collection": "LIDC-IDRI", "SeriesCount": 9}
], "count": 2}
One patient here had two studies: a 1-series study and a 9-series study —
always check SeriesCount rather than assuming one series per study.
4. TCIA_get_series
Params: Collection, PatientID, StudyInstanceUID, Modality,
BodyPartExamined (all optional; combine to narrow). This is the main
filterable search across series.
tu run TCIA_get_series '{"Collection":"LIDC-IDRI","PatientID":"LIDC-IDRI-0001"}'
{"status": "success", "data": [
{"SeriesInstanceUID": "1.3.6.1.4.1.14519.5.2.1.6279.6001.141365756818074696859567662357",
"StudyInstanceUID": "1.3.6.1.4.1.14519.5.2.1.6279.6001.175012972118199124641098335511",
"Modality": "DX", "SeriesDate": "2000-01-01 00:00:00.0",
"BodyPartExamined": "CHEST", "SeriesNumber": 3000923,
"AnnotationsFlag": true, "Collection": "LIDC-IDRI",
"PatientID": "LIDC-IDRI-0001", "Manufacturer": "GE MEDICAL SYSTEMS",
"ManufacturerModelName": "Revolution XQi ADS_28.2",
"ImageCount": 2, "FileSize": 16357620,
"LicenseName": "Creative Commons Attribution 3.0 Unported License",
"LicenseURI": "http://creativecommons.org/licenses/by/3.0/",
"CollectionURI": "https://doi.org/10.7937/K9/TCIA.2015.LO9QL9SX"}
]}
Note this patient's first study is a chest DX (digital radiograph), not a CT — confirm modality per series rather than assuming a collection is uniformly one modality.
5. TCIA_get_modality_values
Params: Collection, BodyPartExamined (both optional).
tu run TCIA_get_modality_values '{"Collection":"LIDC-IDRI"}'
{"status": "success", "data": [
{"Modality": "CR"}, {"Modality": "CT"}, {"Modality": "DX"},
{"Modality": "SEG"}, {"Modality": "..."}
]}
LIDC-IDRI itself is not CT-only — it also has CR/DX radiographs and SEG
(DICOM segmentation objects, i.e. annotation overlays), not just the raw
CT series.
6. TCIA_get_body_part_values
Params: Collection, Modality (both optional).
tu run TCIA_get_body_part_values '{"Modality":"CT"}'
{"status": "success", "data": [
{}, {"BodyPartExamined": "ABDOMEN"}, {"BodyPartExamined": "ABDOMENPELVIS"},
{"BodyPartExamined": "ADRENAL"}, {"BodyPartExamined": "AORTA"},
{"BodyPartExamined": "BLADDER"}
]}
Note the real response includes an empty {} entry (series with no
BodyPartExamined tag set) — handle this rather than assuming every
element has the key.
7. TCIA_get_manufacturer_values
Params: Collection, Modality, BodyPartExamined (all optional).
tu run TCIA_get_manufacturer_values '{"Collection":"LIDC-IDRI","Modality":"CT"}'
{"status": "success", "data": [
{"Manufacturer": "GE MEDICAL SYSTEMS"}, {"Manufacturer": "Philips"},
{"Manufacturer": "SIEMENS"}
]}
8. TCIA_get_series_metadata
Params: SeriesInstanceUID (required).
tu run TCIA_get_series_metadata '{"SeriesInstanceUID":"1.3.6.1.4.1.14519.5.2.1.6279.6001.141365756818074696859567662357"}'
{"status": "success", "data": [
{"Series UID": "1.3.6.1.4.1.14519.5.2.1.6279.6001.141365756818074696859567662357",
"Collection": "LIDC-IDRI",
"Data Description URI": "https://doi.org/10.7937/K9/TCIA.2015.LO9QL9SX",
"Subject ID": "LIDC-IDRI-0001",
"Study UID": "1.3.6.1.4.1.14519.5.2.1.6279.6001.175012972118199124641098335511",
"Study Date": "01-01-2000", "Manufacturer": "GE MEDICAL SYSTEMS",
"Modality": "DX", "SOP Class UID": "1.2.840.10008.5.1.4.1.1.1.1",
"Number of Images": "2", "File Size": "16357620",
"Series Number": "3000923.000000",
"License Name": "Creative Commons Attribution 3.0 Unported License"}
]}
Note the field naming here ("Series UID", "Study Date", spaced keys)
differs from TCIA_get_series's camelCase field names (SeriesInstanceUID,
StudyDate) — this is a different underlying NBIA endpoint with its own
field-naming convention; do not assume the two tools share a schema.
9. TCIA_get_series_size
Params: SeriesInstanceUID (required). Use before recommending a download.
tu run TCIA_get_series_size '{"SeriesInstanceUID":"1.3.6.1.4.1.14519.5.2.1.6279.6001.141365756818074696859567662357"}'
{"status": "success", "data": [{"TotalSizeInBytes": 16357620, "ObjectCount": 2}], "count": 1}
Matches the FileSize/ImageCount fields already visible in
TCIA_get_series's response for the same series (16357620 bytes, 2
images) — both are legitimate ways to get the same numbers.
10. TCIA_get_sop_instance_uids
Params: SeriesInstanceUID (required). Deepest granularity — one entry per
individual DICOM image/slice.
tu run TCIA_get_sop_instance_uids '{"SeriesInstanceUID":"1.3.6.1.4.1.14519.5.2.1.6279.6001.141365756818074696859567662357"}'
{"status": "success", "data": [
{"SOPInstanceUID": "1.3.6.1.4.1.14519.5.2.1.6279.6001.307896144859643716158189196068"},
{"SOPInstanceUID": "1.3.6.1.4.1.14519.5.2.1.6279.6001.257944242390114100388269195181"}
]}
Two SOP instances for this series, matching ImageCount: 2 from
TCIA_get_series — consistent across tools.