Whole adult woman body ct scan and mri?

Support
1

Hi everyone,

Can anyone please tell me where I could download a whole (young healthy) adult woman body ct scan and mri?

I would be used to start building a libre 3D atlas of the woman’s anatomy.

2

Not really sure which one might get you there, but it seems worth taking a check from the following sites;

The Visible Human Project - Getting the Data.

3

If you are looking for Visible Human, it is available from IDC: https://portal.imaging.datacommons.cancer.gov/explore/filters/?collection_id=nlm&collection_id=nlm_visible_human_project

Best way to get started with IDC is via agentic interfaces described here: Start here | IDC User Guide

4

The first link opens another website and contains 135 files for 30 patients. There is no way to filter by sex or to previsualize the content. I chose the first dataset of the list, put it in my ‘cart’ and have to choose wich serie to download (again without previsualization).

I am again downloading 23 Gb of data crossing my fingers to be lucky and find a whole body of adult woman…

Edit: 40 minutes later, I have 1228 folders containing about 100 segmentations each.
I only need a CT scan and an MRI of an adult woman…

image
image1723×788 87 KB

5

I am not sure what you are referring to by “the first link”, and I don’t know what is the screenshot that you included. But if you use IDC Portal, you can visualize and download individual series, see tutorial slides here: Portal tutorial | IDC User Guide.

As I mentioned earlier, the easiest is to set up your LLM agent with IDC skill/MCP. Here’s the response on how to download Visible Human data from IDC: Claude (the actual response is below for completeness).

Happy to respond to any questions about getting data from IDC. We also have dedicated IDC user forum here: https://discourse.canceridc.dev/.

Listing and Downloading Visible Human Modalities from IDC

The Visible Human Project is in the NCI Imaging Data Commons as collection nlm_visible_human_project (2 subjects: VHP-M and VHP-F), on data release v24.

What’s in the collection

Modality What it is Series Size
CT Whole-body axial CT 11 ~2.1 GB
MR T1/T2, axial and coronal, by body region 24 ~0.25 GB
XC Digitized cryosection photographs (70mm film + full-color direct digital) 4 ~303 GB

Note the size skew — XC is 97% of the collection, so download it deliberately rather than by accident.

Setup

pip install idc-index

1. List what’s available

from idc_index import IDCClient
client = IDCClient()

client.sql_query("""
    SELECT Modality, PatientID, SeriesDescription, SeriesInstanceUID,
           ROUND(series_size_MB, 1) AS size_MB
    FROM index
    WHERE collection_id = 'nlm_visible_human_project'
    ORDER BY Modality, PatientID, SeriesDescription
""")

A modality-level summary first, if you just want the shape of it:

client.sql_query("""
    SELECT Modality,
           COUNT(DISTINCT SeriesInstanceUID) AS series,
           ROUND(SUM(series_size_MB)/1024, 2) AS size_GB
    FROM index
    WHERE collection_id = 'nlm_visible_human_project'
    GROUP BY Modality
""")

No-install alternative, same data over REST:

curl -s https://api.imaging.datacommons.cancer.gov/v3/sql \
  -H 'content-type: application/json' \
  -d '{"sql":"SELECT Modality, COUNT(DISTINCT SeriesInstanceUID) n FROM index WHERE collection_id = '\''nlm_visible_human_project'\'' GROUP BY 1"}'

2. Download one modality

download_from_selection takes downloadDir first and filter keywords after — this trips people up because the sibling method download_dicom_series reverses that order.

# All CT (~2.1 GB)
client.download_from_selection(
    downloadDir="./vhp/ct",
    collection_id="nlm_visible_human_project",
    seriesInstanceUID=list(client.sql_query("""
        SELECT SeriesInstanceUID FROM index
        WHERE collection_id = 'nlm_visible_human_project' AND Modality = 'CT'
    """)['SeriesInstanceUID'].values)
)

Swap 'CT' for 'MR' or 'XC'. To narrow further, add AND PatientID = 'VHP-M', or filter on SeriesDescription LIKE 'T2%' for just the T2 MR series.

Files land as <crdc_instance_uuid>.dcm under a %collection_id/%PatientID/%StudyInstanceUID/%Modality_%SeriesInstanceUID tree; the real DICOM UIDs live inside the file headers, not the filenames.

From the shell instead:

idc download <SeriesInstanceUID> --download-dir ./vhp/ct

3. Or pull directly from S3

Every series has a public series_aws_url. With s5cmd installed:

# get the URLs
curl -s https://api.imaging.datacommons.cancer.gov/v3/cohort/manifest.txt \
  -H 'content-type: application/json' \
  -d '{"filters":{"terms":{"collection_id":["nlm_visible_human_project"],"Modality":["MR"]}}}' \
  -o vhp_mr_manifest.txt

# fetch them
cat vhp_mr_manifest.txt | xargs -I {} s5cmd --no-sign-request cp "{}" ./vhp/mr/

No credentials, no egress charges. Add --endpoint-url https://storage.googleapis.com if you’d rather pull from GCS.

Before you preview or publish

Browse a single series without downloading via client.get_viewer_URL(seriesInstanceUID=uid), which opens OHIF in a browser.

On licensing: this collection is not CC-BY like most of IDC. All 39 series carry the National Library of Medicine Terms and Conditions (May 21, 2019). A license is no longer required for access, but NLM asks that you not imply their endorsement of anything you build. Generate attributions with client.citations_from_selection(collection_id="nlm_visible_human_project").

Metadata current as of IDC data release v24 (idc-index-data 24.2.2).