Breast lesion DCE-MRI radiomics (Naples)
Radiomic and Artificial Intelligence Analysis with Textural Metrics, Morphological and Dynamic Perfusion Features Extracted by Dynamic Contrast-Enhanced Magnetic Resonance Imaging in the Classification of Breast Lesions
Breast dynamic contrast-enhanced MRI of 85 women with histologically proven lesions, used to tell benign from malignant lesions with texture, shape and perfusion features. Only per-lesion shape features are open; the images are not shared.
Overview
A retrospective single-centre cohort from the National Cancer Institute of Naples (IRCCS Fondazione G. Pascale), analysed together with the University of Naples Federico II. The study asked how well features computed from dynamic contrast-enhanced breast MRI separate benign from malignant lesions. Texture metrics, shape descriptors and time-intensity curve parameters were extracted from manually segmented lesions and tested one at a time with ROC analysis and jointly with linear discriminant analysis, decision trees, k-nearest neighbours and support vector machines. The authors posted the 15 shape features per lesion on Zenodo under CC BY 4.0. The MR images are not available.
Composition
85 women (mean age 52 years, standard deviation 11, range 26 to 78) with histologically proven breast lesions were enrolled between October 2017 and April 2018. Together they had 91 lesions: 35 benign (mostly ductal hyperplasia, fibroadenoma and fibrosis) and 56 malignant (ductal carcinoma in situ, infiltrating lobular and ductal carcinoma and a few rarer types). Exclusion criteria included breast implants, metal implants, pregnancy, renal disease and ongoing chemotherapy.
Acquisition
All examinations ran on one 1.5 T Siemens Magnetom Symphony with a 16-channel breast coil, patients prone. The protocol had axial turbo spin-echo T2-weighted and fat-suppressed T1-weighted images and a coronal T1-weighted 3D FLASH dynamic series: one acquisition before and nine after 0.1 mmol/kg Gd-DOTA, 56 seconds apart, with 0.885 mm in-plane pixels and 1.6 mm slices. Subtraction images were generated automatically.
Annotations
Two breast radiologists with 22 and 15 years of experience outlined every lesion slice by slice on the third subtracted dynamic series, first separately and then in consensus. The label is histology: surgical specimens for malignant lesions, and surgery, fine needle aspiration or core biopsy for benign ones. The Zenodo spreadsheet lists patient and lesion ids, BI-RADS category, histology and the shape features.
Known limitations
- Images, segmentations, texture and dynamic features are not shared; only the shape features are open.
- Single centre and single scanner, small sample, no external validation set.
- The paper's lesion table counts two intraductal papillomas among the malignant lesions.
Cohort
Aggregate numbers from the sources below. Bars are relative to the 85 subjects.
Scanner vendor
- Siemens Healthineers 85 100%
Section 2.2 in Fusco et al. 2021, Applied Sciences 11(4), 1880
Age
mean 52 ± 11, range 26 to 78
No age bins reported.
Section 2.1 in Fusco et al. 2021, Applied Sciences 11(4), 1880
Contrast / sequence
Groups can overlap
- DCE perfusion 85 100%
Section 2.2 in Fusco et al. 2021, Applied Sciences 11(4), 1880
License and access
Our reading of the license, not legal advice. Before you use the data, read the original license and confirm that your use is allowed. We take no responsibility for how you use a dataset. Full disclaimer
The data are not available outside the institution or study that holds them
The Zenodo record holds one spreadsheet with the 15 morphological features of the 91 lesions, with BI-RADS and histology per lesion. It is openly downloadable under CC BY 4.0. The paper states that no additional data are available, and no route to the MR images is published.
Different parts of the data carry different licenses. The summary on the right shows the most restrictive answer per rule.
No published license or data use terms
The data holder has published no license and no data use terms. Any use, sharing or commercial right has to be agreed with the data holder, and nothing can be assumed to be allowed. Default copyright and data protection law still apply.
Neither the Zenodo record nor the paper offers the images or states terms for them.
What you can do
- Commercial use Not stated
- Not stated
- Train ML models Not stated
- Create derived data Not stated
- Publish results Not stated
What you can share
- Share the data Not stated
- Share derived data Not stated
- Share trained models Not stated
What you must do
- Cite or credit Not stated
- Share alike Not stated
- Sign an agreement Not stated
- Ethics approval Not stated
- Manuscript review Not stated
- Release code Not stated
- Return results Not stated
- Delete after use Not stated
Limits
- No re-identification Not stated
- Location limits Not stated
Creative Commons Attribution 4.0 International
Use, share and adapt the data for any purpose, including commercial use, as long as you credit the creators.
What you can do
- Yes
- Yes
- Yes
- Yes
- Yes
What you can share
- Yes
- Yes
- Yes
What you must do
- Yes
- Share alike No
- No
- No
- Manuscript review No
- Release code No
- Return results No
- Delete after use No
Limits
- No
- Location limits No
Citation
Fusco R, Piccirillo A, Sansone M, Granata V, Vallone P, Barretta ML, Petrosino T, Siani C, Di Giacomo R, Di Bonito M, Botti G, Petrillo A. Radiomic and Artificial Intelligence Analysis with Textural Metrics, Morphological and Dynamic Perfusion Features Extracted by Dynamic Contrast-Enhanced Magnetic Resonance Imaging in the Classification of Breast Lesions. Applied Sciences 11(4), 1880 (2021). doi:10.3390/app11041880
Sources
Every number on this page comes from one of these documents. Each chart names the table or page it is taken from. The raw numbers are in stats.csv.