Update README.md
Browse files
README.md
CHANGED
|
@@ -1,27 +1,49 @@
|
|
| 1 |
# BreastDCEDL_ISPY Dataset
|
| 2 |
|
| 3 |
-
|
| 4 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5 |
|
| 6 |
📄 **Related Paper**:
|
| 7 |
[BreastDCEDL: Curating a Comprehensive DCE-MRI Dataset and Developing a Transformer Implementation](https://arxiv.org/abs/2506.12190)
|
| 8 |
|
| 9 |
📦 **Zenodo Archive**:
|
| 10 |
-
[Download
|
| 11 |
-
|
| 12 |
|
| 13 |
## Contents
|
| 14 |
|
| 15 |
-
- 3D DCE-MRI scans in NIfTI format (3 time points per patient
|
| 16 |
- Tumor segmentation masks
|
| 17 |
-
- Clinical metadata including
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 18 |
|
| 19 |
-
This dataset supports benchmark tasks such as:
|
| 20 |
- Prediction of pathological complete response (pCR)
|
| 21 |
- Hormone receptor (HR) status classification
|
| 22 |
- HER2 status classification
|
| 23 |
- Tumor segmentation
|
| 24 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 25 |
---
|
| 26 |
-
|
| 27 |
-
|
|
|
|
| 1 |
# BreastDCEDL_ISPY Dataset
|
| 2 |
|
| 3 |
+
**BreastDCEDL_ISPY** is a curated dataset of pre-treatment 3D dynamic contrast-enhanced MRI (DCE-MRI) scans from **1,154 breast cancer patients**, combining data from two major clinical trials:
|
| 4 |
+
- **I-SPY1**: 172 patients
|
| 5 |
+
- **I-SPY2**: 982 patients
|
| 6 |
+
|
| 7 |
+
Each patient includes **3 selected DCE-MRI scans**:
|
| 8 |
+
- Pre-contrast
|
| 9 |
+
- Early post-contrast
|
| 10 |
+
- Late post-contrast
|
| 11 |
+
|
| 12 |
+
This minimal version was derived from the full TCIA archives and optimized for deep learning applications in breast cancer imaging and outcome prediction.
|
| 13 |
|
| 14 |
📄 **Related Paper**:
|
| 15 |
[BreastDCEDL: Curating a Comprehensive DCE-MRI Dataset and Developing a Transformer Implementation](https://arxiv.org/abs/2506.12190)
|
| 16 |
|
| 17 |
📦 **Zenodo Archive**:
|
| 18 |
+
[Download BreastDCEDL_ISPY Dataset on Zenodo](https://zenodo.org/records/17274053)
|
|
|
|
| 19 |
|
| 20 |
## Contents
|
| 21 |
|
| 22 |
+
- 3D DCE-MRI scans in NIfTI format (3 time points per patient)
|
| 23 |
- Tumor segmentation masks
|
| 24 |
+
- Clinical metadata including:
|
| 25 |
+
- Pathological complete response (pCR)
|
| 26 |
+
- Hormone receptor (HR) status
|
| 27 |
+
- HER2 status
|
| 28 |
+
- Age and race
|
| 29 |
+
|
| 30 |
+
## Benchmark Tasks Supported
|
| 31 |
|
|
|
|
| 32 |
- Prediction of pathological complete response (pCR)
|
| 33 |
- Hormone receptor (HR) status classification
|
| 34 |
- HER2 status classification
|
| 35 |
- Tumor segmentation
|
| 36 |
|
| 37 |
+
## Source Citations
|
| 38 |
+
|
| 39 |
+
- **I-SPY1**:
|
| 40 |
+
*ACRIN 6657/I-SPY1 Breast MRI Database*, The Cancer Imaging Archive (TCIA).
|
| 41 |
+
[https://doi.org/10.7937/K9/TCIA.2016.3L9TD9TK](https://doi.org/10.7937/K9/TCIA.2016.3L9TD9TK)
|
| 42 |
+
|
| 43 |
+
- **I-SPY2**:
|
| 44 |
+
*ACRIN 6698/I-SPY2 Breast MRI Database*, The Cancer Imaging Archive (TCIA).
|
| 45 |
+
[https://doi.org/10.7937/TCIA.2020.1UPK9QEY](https://doi.org/10.7937/TCIA.2020.1UPK9QEY)
|
| 46 |
+
|
| 47 |
---
|
| 48 |
+
|
| 49 |
+
**License**: CC-BY
|