A Pan-Cancer PDX Histology Image Repository with Genomic and Pathologic Annotation

This dataset corresponds to a collection of images and/or image-derived data available from the
National Cancer Institute Imaging Data Commons (IDC).
This dataset was converted into DICOM representation and ingested by the IDC team.
You can explore and visualize the corresponding images using the
IDC Portal.
You can use the manifests included in this Zenodo record to download the collection following
the Download instructions below.

A patient-derived xenograft (PDX) involves implanting a human tumor biopsy into an immunodeficient
mouse. PDXs model human intra- and inter-tumoral heterogeneity within the intact tissue of the
mouse. This dataset contains histologic hematoxylin and eosin (H&E) whole slide images of PDX
samples and, in some cases, the human progenitor samples from which they are derived. These images
were curated as part of the National Cancer Institute's PDX Development and Trial Centers Research
Network (PDXNet) program, a collaborative initiative focused on pre-clinical model development and
testing of targeted therapeutic agents. They were contributed by Baylor College of Medicine (BCM),
Huntsman Cancer Institute, MD Anderson Cancer Center (MDACC), The Wistar Institute (WISTAR),
Washington University in St Louis (WUSTL), and The Jackson Laboratory (JAX).

Images are provided in Digital Imaging and Communications in Medicine (DICOM) format and are
available from the National Cancer Institute Imaging Data Commons (IDC). The original images in TIFF
and SVS format were provided to the IDC team for archival purposes and were converted into
DICOM Whole Slide Microscopy (SM) representation using custom open source scripts and tools at
idc-wsi-conversion.
Clinical data accompanying the images are available at
pdxnet-image-analysis-aacr2022.

The repository contains >1,000 PDX H&E images and >100 matched human progenitor tumor images.
The dataset encompasses a wide variety of cancer types and anatomic sites, including breast, lung,
colorectal, pancreatic, melanoma, sarcoma, prostate, ovarian, and many others. Cancer types span
adenocarcinomas, squamous cell carcinomas, sarcomas, neuroendocrine tumors, and other histologies.
Most images include pathologic assessment of tumor stage and slide-level proportions of cancer,
stromal, and necrotic regions. A subset of images has associated HoVer-Net cell segmentations and
detailed pathologic annotations of neoplastic, stromal, and necrotic regions.

Genomic and transcriptomic data (RNA-seq, WES) and clinical metadata are linked to the images
via the PDXNet Portal.

Files included

A manifest file's name indicates the IDC data release in which a version of collection data was first introduced. For example, pdxnet-idc_v22-aws.s5cmd corresponds to the contents of the pdxnet collection introduced in IDC data release v22.

pdxnet-idc_v24-aws.s5cmd: AWS download manifest

pdxnet-idc_v24-gcs.s5cmd: GCS download manifest

pdxnet-idc_v24-dcf.dcf: DCF download manifest

Manifest files ending in -aws.s5cmd reference files in Amazon Web Services (AWS) buckets; -gcs.s5cmd reference files in Google Cloud Storage. The actual files are identical and mirrored between AWS and GCP.

Download instructions

Each manifest file includes instructions in its header on how to download the included files.

To download the files using .s5cmd manifests:

Install idc-index:
pip install --upgrade idc-index

Download the files referenced by a manifest included in this dataset:
idc download manifest.s5cmd

To download files using a .dcf manifest, see the manifest header.

For questions or help, contact [e-mail]
or post on the IDC Forum.