Cao 2026 In-house cSS Test Set
In-house cortical superficial siderosis test set of Cao et al. (MICCAI 2026 SASHIMI workshop)
In-house test set of 10 patients with confirmed cortical superficial siderosis on SWI or T2*-weighted MRI, with manual lesion delineations validated by a neuroradiologist. Used to evaluate synthetic-data training for cSS segmentation; not released.
Overview
This is a small clinical test set of patients with cortical superficial siderosis (cSS), the curvilinear hemosiderin deposits along the cortical surface that are a marker of cerebral amyloid angiopathy. Cao and colleagues used it to check whether a segmentation network trained only on procedurally generated images with synthetic cSS and microbleed labels works on real scans. The paper calls the set in-house. The authors are from Friedrich-Alexander-Universität Erlangen-Nürnberg, Otto von Guericke University Magdeburg and the German Centre for Neurodegenerative Diseases, but the paper does not say where the patients were scanned.
Composition
10 patients with confirmed cSS, 6 of them women, with a median age of 74.5 years (interquartile range 67 to 78). The set serves only as a test set. The paper reports no diagnoses beyond cSS and no information on how patients were selected.
Acquisition
Each patient has a susceptibility-weighted or T2*-weighted MRI. Field strength, scanner, vendor and voxel size of the real scans are not reported, nor how many patients had each sequence.
Annotations
A trained investigator delineated the cSS lesions by hand, and an experienced neuroradiologist validated the delineations. The paper evaluates voxel-level segmentation (AUPRC and AUROC) against these masks.
Known limitations
- Not publicly available, and no data availability statement or access route is published.
- Only 10 patients, one reader with neuroradiologist validation.
- Acquisition details and the holding site are not reported.
- The paper's microbleed evaluation used 13 ALFA scans from the VALDO 2021 challenge, which are covered by the VALDO 2021 entry, not by this one.
Cohort
Aggregate numbers from the sources below. Bars are relative to the 10 subjects.
Sex
- Female 6 100%
Covers 6 of 10 subjects.
Condition
subjects, values can overlap
- Superficial siderosis 10 100%
Split
subjects
- Test 10 100%
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 paper calls the set in-house and has no data availability statement; no repository or data terms were found. The paper's own license covers the text only.
What you can do
- Commercial use 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
Cao Y, Dash S, Zachariadis A, Schreiber S, Neumann K, Bernal J. Synthetic training for long-tail haemorrhagic lesion segmentation in data-scarce settings. arXiv:2610.01542 (2026).
All numbers
Every number on this page, as stored in stats.csv, with its source.
| Measure | Breakdown | Value | Source |
|---|---|---|---|
| Subjects | total patients with confirmed cSS | 10 | cao2026 Section 3.2 |
| Subjects | condition=superficial_siderosis | 10 | cao2026 Section 3.2 |
| Subjects | split=test used only for evaluation | 10 | cao2026 Section 3.2 |
| Subjects | sex=female | 6 | cao2026 Section 3.2 |
| Median age | total IQR 67 to 78 years | 74.5 | cao2026 Section 3.2 |
Sources
The keys used in the table above.