Harvard Dataverse hosts a multiparametric breast MRI dataset comprising 52 patient studies. The data includes T2-weighted, diffusion-weighted, native T1-weighted, ultrafast DCE, and regular high-resolution T1-weighted post-contrast sequences, acquired on a Siemens 3.0 Tesla scanner. Author Mbabazi, Ainekirabo contributed this dataset, which was last updated on June 11, 2026.
Use Cases
- Developing image classification models for lesion detection based on multiparametric MRI sequences.
- Training segmentation algorithms for breast anatomy based on T2-weighted and T1-weighted images.
- Researching contrast uptake kinetics based on ultrafast and regular DCE-MRI sequences.
- Benchmarking de-identification techniques for medical imaging data based on the described anonymization pipeline.
Strengths
- Dataset includes 52 patient studies, providing a foundation for model development.
- Data is comprehensively de-identified, with PHI removed via a pipeline that replaces identifiers with pseudonyms and shifts dates.
- Includes multiple clinically relevant MRI sequences: T2-weighted, DWI, native T1-weighted, ultrafast DCE, and high-resolution post-contrast T1-weighted.
Limitations
- Row count and dataset size are unknown, which may limit suitability assessment.
- Column-level documentation is absent; field semantics must be inferred after download.
- Data may reflect temporal or scanner bias inherent to the single-source collection method.
Provenance
- Source
- Harvard Dataverse
- Collection Method
- Acquired on a Siemens 3.0 Tesla scanner following a standardized ultrafast abbreviated breast MRI (AB-MRI) protocol.
- Freshness
- Last updated 2026-06-11 13:46:06; freshness should be verified.