Datasets:

hfigueiras
/

LMBM

This dataset contains reconstructed longitudinal breast DCE-MRI volumes and patient-level tabular metadata for the MAMA-MIA patients whose source data came from ISPY1, ISPY2, and Breast MRI NACT Pilot. Patients excluded during local QC are not included in this public upload.

Dataset Description

This is a multimodal, longitudinal breast cancer response dataset. The imaging modality is volumetric DCE-MRI stored as NIfTI phase files (volume_phase*.nii.gz). The tabular modality contains patient-level clinical, demographic/body-habitus, treatment, tumor biology, outcome, scanner, and acquisition metadata inherited from MAMA-MIA, plus reconstruction fields added for this longitudinal release.

The main response label inherited from MAMA-MIA is pcr, pathological complete response to neoadjuvant chemotherapy. Longitudinal structure is represented by multiple reconstructed MRI exams/timepoints per patient when the source TCIA data contained additional exams. Each patient folder contains one or more exam-date folders, and each exam folder contains the available DCE phase volumes for that timepoint.

Active patients: 925
Active exams/timepoints: 3418
Active NIfTI phase files: 21452
Source cohorts: ISPY1, ISPY2, and Breast MRI NACT Pilot
Timepoints per patient: 15 patients have 1 exam/timepoint, 52 patients have 2 exams/timepoints, 133 patients have 3 exams/timepoints, and 725 patients have 4 exams/timepoints
DCE phase files per exam: 1-11

Clinical And Imaging Table

metadata/clinical_and_imaging_info_active_reconstructed.csv contains one row per active released patient. It has 58 public columns: the 50 MAMA-MIA clinical/imaging fields for the active reconstructed cohort plus 8 reconstruction/linkage fields added for this longitudinal release.

Using the grouping below, the public table includes 19 clinical/treatment/ outcome/tumor-biology variables, 8 demographic/body-habitus variables, 21 imaging/acquisition/scanner/TCIA-series variables, 8 reconstruction/linkage variables, and 2 identifier/source-provenance variables.

Column group	Count	Columns
Identifiers/source provenance	2	`patient_id`, `dataset`
Clinical, treatment, outcome, and tumor biology	19	`bilateral_breast_cancer`, `multifocal_cancer`, `nac_agent`, `endocrine_therapy`, `anti_her2_neu_therapy`, `pcr`, `mastectomy_post_nac`, `days_to_follow_up`, `days_to_recurrence`, `days_to_metastasis`, `days_to_death`, `hr`, `er`, `pr`, `her2`, `mammaprint`, `oncotype_score`, `nottingham_grade`, `tumor_subtype`
Demographics and body habitus	8	`age`, `menopause`, `ethnicity`, `has_implant`, `weight`, `patient_size`, `bmi_group`, `breast_density`
Imaging, acquisition, scanner, and TCIA series metadata	21	`view`, `bilateral_mri`, `num_phases`, `fat_suppressed`, `field_strength`, `image_rows`, `image_columns`, `num_slices`, `pixel_spacing`, `slice_thickness`, `site`, `manufacturer`, `scanner_model`, `high_bit`, `window_center`, `window_width`, `echo_time`, `repetition_time`, `acquisition_times`, `acquisition_date`, `tcia_series_uid`
Reconstruction and longitudinal linkage	8	`active_reconstructed_patient_id`, `active_volume_root`, `active_patient_rel_path`, `num_reconstructed_exams`, `num_reconstructed_extra_exams`, `num_reconstructed_phase_files`, `reconstructed_exam_dates`, `reconstructed_exam_dates_iso`

Dataset Organization

.
|-- README.md
|-- metadata/
|   |-- clinical_and_imaging_info_active_reconstructed.csv
|   |-- patient_manifest.csv
|   |-- exam_manifest.csv
|   |-- phase_manifest.csv
|   `-- data_dictionary.csv
`-- Reconstructed_Datasets/
    |-- ISPY1-Volumes/                         # 145 patients, 549 exams, 1705 phase files
    |   `-- ISPY1_1001/
    |       `-- 10-13-1984/
    |           |-- volume_phase0.nii.gz
    |           |-- volume_phase1.nii.gz
    |           `-- volume_phase2.nii.gz
    |-- ISPY2-Volumes/                         # 716 patients, 2680 exams, 19159 phase files
    |   `-- ISPY2-100899/
    |       `-- 10-26-2002/
    |           |-- volume_phase0.nii.gz
    |           |-- volume_phase1.nii.gz
    |           `-- ...
    `-- NACT-Volumes/                          # 64 patients, 189 exams, 588 phase files
        `-- NACT_01/
            `-- <MM-DD-YYYY>/
                |-- volume_phase0.nii.gz
                |-- volume_phase1.nii.gz
                `-- ...

Public-facing patient_id values in the metadata follow MAMA-MIA naming. ISPY2 folders use hyphens locally, for example ISPY2-100899; the metadata also records the MAMA-MIA-style identifier ISPY2_100899.

Metadata

metadata/clinical_and_imaging_info_active_reconstructed.csv
metadata/patient_manifest.csv
metadata/exam_manifest.csv
metadata/phase_manifest.csv
metadata/data_dictionary.csv

The phase manifest contains project-relative paths to every uploaded NIfTI file.

Sources And Citations

This release is derived from public TCIA/MAMA-MIA source data. Users should cite MAMA-MIA, the original source datasets represented here, and TCIA.

MAMA-MIA

@article{garrucho2025,
  title={A large-scale multicenter breast cancer DCE-MRI benchmark dataset with expert segmentations},
  author={Garrucho, Lidia and others},
  journal={Scientific Data},
  year={2025},
  doi={10.1038/s41597-025-04707-4},
  pages={453},
  number={1},
  volume={12}
}

Original Source Datasets

@dataset{newitt2016ispy1,
  title={Multi-center breast DCE-MRI data and segmentations from patients in the I-SPY 1/ACRIN 6657 trials},
  author={Newitt, David and Hylton, Nola and I-SPY 1 Network and ACRIN 6657 Trial Team},
  year={2016},
  publisher={The Cancer Imaging Archive},
  doi={10.7937/K9/TCIA.2016.HdHpgJLK}
}

@dataset{newitt2016nact,
  title={Single site breast DCE-MRI data and segmentations from patients undergoing neoadjuvant chemotherapy},
  author={Newitt, David and Hylton, Nola},
  year={2016},
  version={3},
  publisher={The Cancer Imaging Archive},
  doi={10.7937/K9/TCIA.2016.QHsyhJKy}
}

@dataset{li2022ispy2,
  title={I-SPY 2 Breast Dynamic Contrast Enhanced MRI Trial (ISPY2)},
  author={Li, W. and Newitt, D. C. and Gibbs, J. and Wilmes, L. J. and Jones, E. F. and Arasu, V. A. and Strand, F. and Onishi, N. and Nguyen, A. A.-T. and Kornak, J. and Joe, B. N. and Price, E. R. and Ojeda-Fournier, H. and Eghtedari, M. and Zamora, K. W. and Woodard, S. A. and Umphrey, H. and Bernreuter, W. and Nelson, M. and Hylton, N. M.},
  year={2022},
  version={1},
  publisher={The Cancer Imaging Archive},
  doi={10.7937/TCIA.D8Z0-9T85}
}

TCIA

@article{clark2013tcia,
  title={The Cancer Imaging Archive (TCIA): Maintaining and Operating a Public Information Repository},
  author={Clark, K. and Vendt, B. and Smith, K. and Freymann, J. and Kirby, J. and Koppel, P. and Moore, S. and Phillips, S. and Maffitt, D. and Pringle, M. and Tarbox, L. and Prior, F.},
  journal={Journal of Digital Imaging},
  volume={26},
  number={6},
  pages={1045--1057},
  year={2013},
  doi={10.1007/s10278-013-9622-7}
}

Use Restrictions

Do not attempt to identify or contact individual participants. Follow TCIA and MAMA-MIA data usage policies and attribution requirements. This dataset is shared under a non-commercial Creative Commons license to match the MAMA-MIA license.

Downloads last month: 90

Reconstructed MAMA-MIA Longitudinal DCE-MRI