Page History

Breast MRI is a common image modality to assess extent of disease in breast cancer patients. Recent studies show that MRI has a potential in prognosis of patient sort and long term outcomes as well as predicting pathological and genomic features of the tumors. However, large, well annotated datasets are needed to make further progress in the field. We share such a dataset here.

In this dataset, we share MRI imaging and other data for 922 patients with invasive breast cancer. Their prone position axial breast MRI images were acquired by 1.5T or 3T scanners. Following MRI sequences are shared: a non-fat saturated T1-weighted sequence, a fat-saturated gradient echo T1-weighted pre-contrast sequence, and mostly three to four post-contrast sequences.

The images are associated with a variety of metadata. Specifically, we are sharing:

1. Demographic, clinical, pathology, treatment, outcomes, and genomic data. This data has been collected from a variety of sources including clinical notes and has served as a source for multiple published papers on radiogenomics, outcomes prediction, and other.
2. 529 imaging features which represent a variety of imaging characteristics including size, shape, texture, and enhancement of both the tumor and the surrounding tissue, which is combined of features commonly published in the literature, as well as the features developed in our lab. These features were used in our previous paper (https://doi.org/10.1016/j.eswa.2017.06.029)
3. Annotation boxes provided by radiologists that indicate locations of the lesions in the images.

For more information see this site https://sites.duke.edu/mazurowski/resources/breast-cancer-mri-dataset/, the related publication, or contact TCIA Helpdesk at help@cancerimagingarchive.net. 

A large dataset of dynamic contrast-enhanced magnetic resonance images of the breast for breast cancer patients. Tumors annotated by radiologists. Extensive meta-data including patient demographics, treatment, outcomes, pathology, genomics.

This dataset is the dataset used in the following paper:

The dataset contains four components: (1) DICOM Images, (2) Annotation_Boxes, (3) Clinical_and_Other_Features, and (4) Imaging_Features.  Breast MRI images of 922 patients, including the following sequences: fat saturated pre-contrast sequence, fat saturated post-contrast sequences, and non-fat saturated T1 weighted sequence. Tumor annotations in the form of boxes drawn by radiologists, six values in the spreadsheet represent the location of the start row, end row, start column, end column, start slice, and end slice of the box. The annotation boxes can be applied directly to pre and post sequences. 

Accompanying the DICOM are two spreadsheets: a spreadsheet containing demographic, clinical, pathology, outcomes, genomic, and other information for the patients. The legend and column explanation are included in the spreadsheet (different tabs). Also, a spreadsheet that contains 529 imaging features extracted from the images. These are the features that were used in the paper listed above (Saha et al., BJR 2018).

Acknowledgements

We would like to acknowledge the individuals and institutions that have provided data for this collection: