Skip to content
Open Imaging Index
MRI Brain

UCSF Radiation-Induced Microbleed 7 T SWI Set

UCSF 7 T SWI dataset of radiation-induced cerebral microbleeds in glioma patients

Ultra-high-field susceptibility-weighted brain MRI of glioma patients with microbleeds caused by radiotherapy, used to train a network that separates true microbleeds from false detections. 91 scans of 73 patients at 7 T. No data release or sharing route.

Go to dataset Edit Report

Overview

This dataset comes from the Lupo lab at the University of California San Francisco and was used to train a 3D deep residual network that separates true cerebral microbleeds from false positive candidates produced by an earlier, rule-based detector. All patients had gliomas and had received brain radiotherapy. The data are described in a 2019 paper only; the paper has no data availability statement, and no download or request process has been published.

Composition

73 glioma patients, each treated with radiation to a maximum dose of 50 to 60 Gy and each with confirmed radiation-induced microbleeds. Twelve patients were scanned more than once, giving 91 scans. The paper split the patients into 54 for training, 7 for validation and 12 for testing. The paper does not report age or sex.

Acquisition

All scans were acquired at 7 T on a GE scanner with an 8- or 32-channel phased-array head coil. 31 patients had a 4-echo 3D TOF-SWI sequence (TE 2.4, 12, 14.3 and 20.3 ms, TR 40 ms, 0.5 x 0.5 x 1 mm) and 49 had a standard flow-compensated 3D SWI sequence (TE 16 ms, TR 50 ms, 0.5 x 0.5 x 2 mm). These two numbers add up to more than 73, which the paper does not explain.

Annotations

The earlier detector proposed 19,762 candidates. A research scientist experienced in reading microbleeds, guided beforehand by a neuroradiologist, marked each candidate as a true microbleed or a false positive with a labelling tool, yielding 2,835 true microbleeds and 16,927 false positives. A neuroradiologist's ratings were used separately to check the network's likelihood scores. The labels are candidate-level decisions, not voxel masks.

Known limitations

  • No data release, license or data use terms exist; access, if any, is at the authors' discretion.
  • The population is narrow: glioma patients after radiotherapy, scanned at 7 T, which differs from the usual 1.5 T and 3 T clinical setting.
  • Ground truth comes from a single reader.
  • The publisher issued a correction stating the article was published open access by mistake.

Cohort

Aggregate numbers from the sources below. Bars are relative to the 73 subjects.

Contrast / sequence

Groups can overlap

  • Susceptibility-weighted 73 100%

Subjects and Image Acquisition in Chen et al. 2019, Journal of Digital Imaging (PMC6737152)

Scanner vendor

  • GE HealthCare 73 100%

Subjects and Image Acquisition in Chen et al. 2019, Journal of Digital Imaging (PMC6737152)

Split

  • Training 54 74%
  • Test 12 16%
  • Validation 7 10%

Table 1 in Chen et al. 2019, Journal of Digital Imaging (PMC6737152)

Condition

Groups can overlap

  • Cerebral microbleeds 73 100%
  • Glioma 73 100%

Subjects and Image Acquisition in Chen et al. 2019, Journal of Digital Imaging (PMC6737152)

Field strength

  • 7 T 73 100%

Subjects and Image Acquisition in Chen et al. 2019, Journal of Digital Imaging (PMC6737152)

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

Access
On request

Ask the authors or the data holder. No published process or terms, and access is at their discretion

Access page

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.

The paper has no data availability statement and no data sharing route is published. The corresponding author is Janine Lupo, UCSF.

Original license text Checked 2026-10-08

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

Citation

Chen Y, Villanueva-Meyer JE, Morrison MA, Lupo JM. Toward Automatic Detection of Radiation-Induced Cerebral Microbleeds Using a 3D Deep Residual Network. J Digit Imaging 32(5), 766-772 (2019). doi:10.1007/s10278-018-0146-z

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.