Datasets:
brain-mri-redacted
De-identified medical imaging study processed by micom.
Dataset Summary
| Property | Value |
|---|---|
| Study | brain-mri-redacted |
| Series | 19 image + 0 presentation state |
| Quality Grade | ? (?/100) |
| HIPAA Score | 100.0/100 (0 PHI findings post-redaction) |
| Scanner | GE MEDICAL SYSTEMS 3T |
| Visualizations | 13544 |
De-identification
This dataset has been de-identified using HIPAA Safe Harbor (45 CFR 164.514(b)(2)):
- 26 tags removed (address, phone, clinical trial IDs, etc.)
- 22 tags blanked (patient/physician names, institution, IDs)
- 5 UIDs rehashed (study/series/SOP instance UIDs)
- 6 dates shifted (birth, study, acquisition dates)
- Single-pass verified (each file confirmed clean in-memory)
See hipaa_compliance.json for the full pre/post compliance report.
Contents
report.html # Interactive HTML dashboard
series_stats.json # Per-series metadata + quality metrics
hipaa_compliance.json # HIPAA compliance report (pre + post redaction)
dicom_full_dump.json # All DICOM tags (redacted values)
dicom_metadata.csv # Tabular metadata export
cross_series_comparison.png # Quality comparison across series
montages/ # Multi-plane slice montages per series
histograms/ # Intensity distribution plots per series
enhanced/ # MIP/MinIP/tissue segmentation views
Quality Analysis
Each series is graded A-F based on:
- SNR estimate (signal-to-noise ratio)
- Slice intensity uniformity
- Tissue coverage percentage
- Dynamic range
- Entropy (information content)
Usage
from huggingface_hub import hf_hub_download
report = hf_hub_download("brain-mri-redacted", "report.html", repo_type="dataset")
stats = hf_hub_download("brain-mri-redacted", "series_stats.json", repo_type="dataset")
License
This dataset is shared for research purposes. All patient identifiers have been removed per HIPAA Safe Harbor. Do not attempt to re-identify subjects.
Generated by micom DICOM Extractor v5.
- Downloads last month
- 4,290