Breast MRI molecular cancer subtype

This data set is part of the public development data for the 2023 Automated Universal Classification Challenge (AUC23). The data set concerns the classification of breast cancer molecular subtypes on dynamic contrast-enhanced magnetic resonance imaging (MRI) and was derived from Duke Hospital. The data set is a subset of the data originally introduced and described by Saha et al. (2018), with no additional images or patient information. Data was restructured in compliance with the AUC23 challenge format. The dataset is a single-institutional, retrospective collection of 737 biopsy-confirmed patients from 1 January 2000 to 23 March 2014 with invasive breast cancer and available pre-operative MRI at Duke Hospital.

Images are 3D tensors:

• 0: 3D T1-subtraction dynamic contrast-enhanced MRI

Classification labels:

• 0: Luminal A, estrogen-receptor (ER) and/or progesterone-receptor (PR) positive, human epidermal growth factor receptor 2 (HER2) negative
• 1: Luminal B, ER and/or PR negative, HER2 positive
• 2: HER2, ER and PR negative, HER2 positive
• 3: Triple negative, ER, PR, and HER2 negative

Folder structure:

imagesTr (root folder with all patients and studies)
    ├── Breast_MRI_0001_0000.mha  (3D T1-subtraction MRI imaging for study 0001)
    ├── Breast_MRI_0003_0000.mha  (3D T1-subtraction MRI imaging for study 0003)
    ├── ...

Please cite the following article if you are using the Duke-Breast-Cancer-MRI Dataset:

A. Saha, M. R. Harowicz, L. J. Grimm, C. E. Kim, S. V. Ghate, R. Walsh, M. A. Mazurowski, "A machine learning approach to radiogenomics of breast cancer: a study of 922 subjects and 529 DCE-MRI features". Br J Cancer. 2018 Aug;119(4):508-516. doi: 10.1038/s41416-018-0185-8. Epub 2018 Jul 23. PMID: 30033447; PMCID: PMC6134102.