CBIS-DDSM-R: A Curated Radiomic Feature Dataset for Breast Cancer Classification

CBIS-DDSM-R is a curated and reproducible version of the Curated Breast Imaging Subset of DDSM (CBIS-DDSM). This project provides a fully automated pipeline for downloading, processing, and extracting radiomic features from CBIS-DDSM, enabling reproducible experiments in breast cancer imaging research.

Project Structure

cbis-ddsm-r/
│
├── consts/                # Project-wide constants and paths
│   ├── const.py
│   └── paths.py
│
├── data/
│   ├── img/               # DICOM images and ROI masks
│   └── csv/               # Final output CSVs with radiomic features
│
├── dataset/               # Dataset construction and metadata processing
│   ├── constructor.py     # Merges radiomic features with CBIS-DDSM metadata
│   ├── downloader.py      # Downloads CBIS-DDSM from TCIA
│   └── processor.py       # Processes and parses CBIS-DDSM metadata
│
├── features/
│   └── radiomics_features.py # Handles radiomic feature extraction using PyRadiomics
│
├── manifest/
│   └── CBIS-DDSM-All.tcia   # Manifest file to download dataset via NBIA Data Retriever
│
├── metadata/             # Original CBIS-DDSM metadata files from TCIA
│   ├── calc_case_description_test_set.csv
│   ├── calc_case_description_train_set.csv
│   ├── mass_case_description_test_set.csv
│   └── mass_case_description_train_set.csv
│
├── pipeline/             # Orchestrates the full processing pipeline
│   ├── run_downloader.py
│   ├── run_metadata_processor.py
│   ├── run_image_processing.py
│   └── run_radiomics.py
│
├── preprocessing/        # Image preprocessing utilities and filters
│   ├── image_processing.py
│   └── image_utils.py
│
├── storage/              # Handles saving of local data
│   └── local.py

What's Included

• Automated download of CBIS-DDSM via manifest
• DICOM image conversion and pectoral muscle removal
• ROI extraction and alignment
• Full PyRadiomics feature extraction pipeline
• Metadata processing and merging with radiomic features

Requirements

• Python 3.9

Python Environment

conda create -n cbis-ddsm-r
conda activate cbis-ddsm-r
conda install -r requirements.txt

Usage

1. Download the dataset

python pipeline/run_downloader.py

2. Process the metadata

python pipeline/run_metadata_processor.py

3. Preprocess images

python pipeline/run_image_processing.py

4. Extract radiomic features using PyRadiomics

python pipeline/run_radiomics.py

5. Create the dataset

python pipeline/run_dataset_construction.py

Automatic pipeline

In case you want to run just one script to get dataset without modifying the pipeline, run the following command:

python pipeline/run_all.py

Dataset on HuggingFace

In case you don't want to run any python script, you can find the dataset at: HuggingFace

Cite this paper

If you use this dataset, please cite the following paper.

@Article{data10110179,
AUTHOR = {Sánchez-Femat, Erika and Galván-Tejada, Carlos E. and Galván-Tejada, Jorge I. and Gamboa-Rosales, Hamurabi and Luna-García, Huizilopoztli and Flores-Chaires, Luis Alberto and Saldívar-Pérez, Javier and Reveles-Martínez, Rafael and Celaya-Padilla, José M.},
TITLE = {CBIS-DDSM-R: A Curated Radiomic Feature Dataset for Breast Cancer Classification},
JOURNAL = {Data},
VOLUME = {10},
YEAR = {2025},
NUMBER = {11},
ARTICLE-NUMBER = {179},
URL = {https://www.mdpi.com/2306-5729/10/11/179},
ISSN = {2306-5729},
ABSTRACT = {Early and accurate breast cancer detection is critical for patient outcomes. The Curated Breast Imaging Subset of the Digital Database for Screening Mammography (CBIS-DDSM) has been instrumental for computer-aided diagnosis (CAD) systems. However, the lack of a standardized preprocessing pipeline and consistent metadata has limited its utility for reproducible quantitative imaging or radiomics. This paper introduces CBIS-DDSM-R, an open-source, radiomics-ready extension of the original dataset. It provides an automated pipeline for preprocessing mammograms and extracts a standardized set of 93 radiomics features per lesion, adhering to Image Biomarker Standardisation Initiative (IBSI) guidelines using PyRadiomics. The resulting dataset combines clinical and radiomics data into a unified format, offering a robust benchmark for developing and validating reproducible radiomics models for breast cancer characterization.},
DOI = {10.3390/data10110179}
}