Datasets:

MedOtter
/

Adrenal-ACC-Ki67-Seg

license: cc-by-4.0
task_categories:
  - image-segmentation
tags:
  - medical
  - ct
  - adrenal
  - adrenocortical-carcinoma
  - ki-67
  - tumor-segmentation
  - dicom
  - tcia
pretty_name: Adrenal-ACC-Ki67-Seg
size_categories:
  - n<1K
dataset_info:
  features:
    - name: patient_id
      dtype: string
    - name: study_uid
      dtype: string
    - name: ct_series_uid
      dtype: string
    - name: seg_series_uid
      dtype: string
    - name: num_ct_slices
      dtype: int32
    - name: slice_index
      dtype: int32
    - name: tumor_voxels
      dtype: int64
    - name: image
      dtype: image
    - name: mask
      dtype: image
    - name: overlay
      dtype: image
  splits:
    - name: preview
      num_bytes: 14238577
      num_examples: 53
  download_size: 14247684
  dataset_size: 14238577
configs:
  - config_name: default
    data_files:
      - split: preview
        path: data/preview-*

Adrenal-ACC-Ki67-Seg

Voxel-level segmentation of pathologically-proven Adrenocortical Carcinoma (ACC) tumors on contrast-enhanced CT, with matched Ki-67 proliferation-marker labels.

Dataset Details

Field	Value
Modality	CT (contrast-enhanced)
Body part	Abdomen — adrenal gland (tumor)
Task	3D binary tumor segmentation
Patients	53
Studies	65
CT series	124
SEG series	53 (one per patient)
Images	18,255 DICOM slices
Format	DICOM (images) + DICOM SEG (segmentations) + XLSX (clinical)
License	CC BY 4.0

No predefined train/val/test split — single cohort of 53 patients with pathologically confirmed ACC imaged between 2006 and 2018. Tumor volumes range from 2.6 to 7,864 cm³ (median ~177 cm³).

Structure

images/<PatientID>/<StudyInstanceUID>/<SeriesInstanceUID>/*.dcm
segmentations/<PatientID>/<StudyInstanceUID>/<SeriesInstanceUID>/*.dcm
Adrenal-ACC-Ki67-Seg_SupportingData_20230522.xlsx   # clinical & Ki-67 data
series_to_patient.json                              # series-level metadata index

Segmentation DICOMs are DICOM SEG objects and reference their source CT series via DICOM metadata. Clinical metadata (demographics, Ki-67 index, follow-up, tumor measurements) is provided in the XLSX spreadsheet.

Source

TCIA collection: https://www.cancerimagingarchive.net/collection/adrenal-acc-ki67-seg/
DOI: 10.7937/1FPG-VM46

Citation

@article{ahmed2020radiomic,
  author  = {Ahmed, Ahmed A. and Elmohr, Mohab M. and Fuentes, David and
             Habra, Mouhammed A. and Fisher, Sarah B. and Perrier, Nancy D. and
             Zhang, Meng and Elsayes, Khaled M.},
  title   = {Radiomic mapping model for prediction of Ki-67 expression in
             adrenocortical carcinoma},
  journal = {Clinical Radiology},
  volume  = {75},
  number  = {6},
  pages   = {479.e17--479.e22},
  year    = {2020},
  doi     = {10.1016/j.crad.2020.01.012}
}

@misc{adrenalacc2022tcia,
  author    = {Ahmed, A. A. and Elmohr, M. M. and Fuentes, D. and
               Habra, M. A. and Fisher, S. B. and Perrier, N. D. and
               Zhang, M. and Elsayes, K. M.},
  title     = {Adrenal-ACC-Ki67-Seg [Dataset]},
  year      = {2022},
  publisher = {The Cancer Imaging Archive},
  doi       = {10.7937/1FPG-VM46}
}