Brain Before Stroke (BBS)
Brain Before Stroke (BBS) prospective cohort of first-ever ischemic stroke, Bordeaux University Hospital
Early brain MRI after a first ischemic stroke with 1-year outcomes, collected to study how imaging markers relate to recovery. 428 adults from a prospective cohort in Bordeaux, France, scanned 24 to 72 hours after onset. Shared on request to the principal investigator.
Overview
Brain Before Stroke (BBS) is a prospective single-centre cohort from Bordeaux University Hospital, France, that followed consecutive patients with a first-ever ischemic stroke to study how imaging relates to recovery. It ran from June 2012 to February 2015. Its data have been used to test whether small vessel disease markers improve outcome prediction (Coutureau et al. 2021), to train and test the SHIVA-CMB microbleed detector (Tsuchida et al. 2024) and for the NeuralCup outcome prediction benchmark (Matsulevits et al. 2024).
Composition
428 adults over 18 with a minor to severe supratentorial infarct (NIHSS 1 to 25) and no prior disabling stroke, dementia or psychiatric disorder other than major depression. Mean age 67.5 years (SD 14.1) and 63.5% men. Patients were assessed for motor, cognitive and psychological outcomes, with the NeuralCup work using the 1-year scores (Fugl-Meyer, MoCA, Isaacs set, HADS). Of 361 patients with a rated SWAN scan, 92 had at least one microbleed.
Acquisition
MRI was done 24 to 72 hours after onset on a 3 T GE Discovery MR750w, with a second visit at one year. Sequences named in the sources are DWI, 3D FLAIR, 3D T1, SWAN (TR 60 ms, TE 24.3 ms, 0.43 x 0.43 x 1.6 mm) and a 2D multi-echo T2*-GRE (TE 21.7 ms echo used in SHIVA-CMB, 0.47 x 0.47 x 4.5 mm).
Annotations
Infarct masks were delineated manually. For SHIVA-CMB, 162 patients (all 92 with microbleeds and 70 random ones without) received microbleed masks on SWAN, made semi-automatically for training and manually for the test set by two expert raters, then projected to the T2*-GRE, where the test-set masks were reviewed. Coutureau et al. quantified white matter hyperintensities, lacunes, perivascular spaces, microbleeds and atrophy.
Known limitations
- Not public. Requests go to the PI and no data use terms are published.
- One centre and one scanner model.
- Counts per sequence are known only for the subsets used in each paper.
- The microbleed masks and infarct masks were made for specific studies; whether they come with a data request is not stated.
Cohort
Aggregate numbers from the sources below. Bars are relative to the 428 subjects.
Age
mean 67.5 ± 14.1
No age bins reported.
Methods, Datasets, BBS in Tsuchida et al. 2024, SHIVA-CMB, Scientific Reports (PMC11680838)
Condition
Groups can overlap
- Ischemic stroke 428 100%
- Cerebral microbleeds 92 21%
Methods, Datasets, BBS in Tsuchida et al. 2024, SHIVA-CMB, Scientific Reports (PMC11680838)
Contrast / sequence
Groups can overlap
- Susceptibility-weighted 361 84%
- Diffusion-weighted 237 55%
- FLAIR 237 55%
- T1-weighted 237 55%
- T2*-weighted 159 37%
Methods, Datasets, BBS, Table 1 in Tsuchida et al. 2024, SHIVA-CMB, Scientific Reports (PMC11680838); Methods, Dataset in Matsulevits et al. 2024, NeuralCup stroke outcome benchmark, bioRxiv preprint (PMC11507916)
Country
- France 428 100%
Methods, Datasets, BBS in Tsuchida et al. 2024, SHIVA-CMB, Scientific Reports (PMC11680838)
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
Ask the authors or the data holder. No published process or terms, and access is at their discretion
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 SHIVA-CMB data availability statement says BBS is not public because of French rules on sharing medical images, and that de-identified data may be available on request to the PI, Thomas Tourdias. No terms are published. A 1-year-outcome subset with DWI, FLAIR, T1 and infarct masks is offered separately on request as the NeuralCup challenge training set (neuralcup.bcblab.com); it is not covered by this entry.
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
Coutureau J, Asselineau J, Perez P, Kuchcinski G, Sagnier S, Renou P, Munsch F, Lopes R, Henon H, Bordet R, Dousset V, Sibon I, Tourdias T. Cerebral Small Vessel Disease MRI Features Do Not Improve the Prediction of Stroke Outcome. Neurology 96, e527-e537 (2021). doi:10.1212/WNL.0000000000011208
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.