{"success":true,"data":{"author":"ISLES challenge organizers (Mauricio Reyes, Ezequiel de la Rosa, Bjoern Menze et al.); mirrored by MedOtter","bucketAccessType":"public","bucketName":"MedOtter/isles2018","bucketPrefix":"","category":"Neuroscience","cloudProvider":"huggingface","createdAt":"2026-11-17T19:29:33.599Z","description":"The Ischemic Stroke Lesion Segmentation (ISLES) 2018 challenge data: CT perfusion (CTP) of acute anterior-circulation ischemic stroke for infarct core segmentation, from 103 patients and 156 CTP acquisitions pooled from the DEFUSE 2 study and a companion prospective cohort across four centers (three in the US, one in Australia). Limited z-coverage meant some patients were scanned as two slabs, each its own case folder; group on patient_id in cases.jsonl to avoid leakage. Each case has the baseline CT, the raw motion-corrected 4D CTP series (CT_4DPWI, about 1 s temporal resolution) and derived perfusion maps CT_CBF, CT_CBV, CT_MTT and CT_Tmax. TRAINING (94 cases, 63 patients) includes the OT infarct mask, drawn on MR diffusion-weighted images acquired within 3 hours of CTP and registered to CTP space; TESTING (62 cases) has no released ground truth. About 11.8 GB (measured 2026-11-17), files byte-identical to the organizers' Zenodo re-release. Load with nibabel or SimpleITK. Mirrored on Hugging Face by MedOtter from the authors' release, with a card documenting provenance and any normalization. The organizers' Zenodo record 17736412 (ISLES 2015 to 2018) is CC BY 4.0, and the archives also carry an ODbL notice; cite Cereda et al. 2016 and the ISLES 2018 challenge.","displayTitle":"ISLES 2018: CT perfusion of acute ischemic stroke with DWI-defined infarct core masks (103 patients, 156 cases)","doi":"10.5281/zenodo.17736412","huggingFaceUrl":"https://huggingface.co/datasets/MedOtter/isles2018","id":"a01cbac6-da3d-5ef9-9bb7-302561f1444c","license":"CC-BY-4.0","licenseEvidenceUrl":"https://zenodo.org/records/17736412","objectCountEstimate":2066,"probeStatus":"ok","region":"us","schemaCard":{"accessHints":[{"language":"python-datasets","snippet":"from datasets import load_dataset\nds = load_dataset(\"MedOtter/isles2018\")"},{"language":"python-hub","snippet":"from huggingface_hub import snapshot_download\nsnapshot_download(repo_id=\"MedOtter/isles2018\", repo_type=\"dataset\")"}],"caveats":["HuggingFace metadata: author=MedOtter, license:odbl, size_categories:n<1K, format:parquet, downloads=10435"],"format":"mixed","generatedAt":"2026-11-17T19:40:36.296Z","layout":{"example":"cases.jsonl","pattern":"{repoId}/{filename}"},"mountPath":"/mnt/archive","sampleHeaders":[{"bytes":"{\"case_id\": \"case_1\", \"split\": \"train\", \"smir_case_label\": \"Train_01_A\", \"patient_id\": \"Train_01\", \"slab\": \"A\", \"images\": {\"CT\": \"TRAINING/case_1/SMIR.Brain.XX.O.CT.345562/SMIR.Brain.XX.O.CT.345562.nii\", \"CT_4DPWI\": \"TRAINING/case_1/SMIR.Brain.XX.O.CT_4DPWI.345561/SMIR.Brain.XX.O.CT_4DPWI.345561.nii\", \"CT_CBF\": \"TRAINING/case_1/SMIR.Brain.XX.O.CT_CBF.345563/SMIR.Brain.XX.O.CT_CBF.345563.nii\", \"CT_CBV\": \"TRAINING/case_1/SMIR.Brain.XX.O.CT_CBV.345564/SMIR.Brain.XX.O.CT_CBV.345564.nii\", \"CT_MTT\": \"TRAINING/cas","encoding":"utf8","key":"cases.jsonl","truncatedAt":512}],"sampleObjects":["LICENSE_ODC_ODBL.txt","TESTING/case_1/SMIR.Brain.XX.O.CT.346291/SMIR.Brain.XX.O.CT.346291.json","TESTING/case_1/SMIR.Brain.XX.O.CT.346291/SMIR.Brain.XX.O.CT.346291.nii","TESTING/case_1/SMIR.Brain.XX.O.CT_4DPWI.346290/SMIR.Brain.XX.O.CT_4DPWI.346290.json","TESTING/case_1/SMIR.Brain.XX.O.CT_4DPWI.346290/SMIR.Brain.XX.O.CT_4DPWI.346290.nii"],"sidecars":[{"bytes":"---\nlicense: odbl\npretty_name: ISLES 2018 (Ischemic Stroke Lesion Segmentation Challenge)\ntask_categories:\n- image-segmentation\ntags:\n- medical\n- stroke\n- ct-perfusion\n- brain\n- 3d\nsize_categories:\n- n<1K\nconfigs:\n- config_name: default\n  data_files:\n  - split: train\n    path: data/train-*\n  - split: test\n    path: data/test-*\ndataset_info:\n  features:\n  - name: case_id\n    dtype: string\n  - name: split\n    dtype: string\n  - name: smir_case_label\n    dtype: string\n  - name: patient_id\n    dtype: string\n  - name: slab\n    dtype: string\n  - name: image\n    dtype: image\n  - name: tmax\n    dtype: image\n  - name: overlay\n    dtype: image\n  - name: num_slices\n    dtype: int32\n  - name: num_timepoints\n    dtype: int32\n  - name: spacing_mm\n    dtype: string\n  - name: has_mask\n    dtype: bool\n  - name: lesion_voxels\n    dtype: int64\n  splits:\n  - name: train\n    num_bytes: 4144300\n    num_examples: 94\n  - name: test\n    num_bytes: 2538530\n    num_examples: 62\n  download_size: 6689953\n  dataset_size: 6682830\n---\n\n# ISLES 2018 — Ischemic Stroke Lesion Segmentation Challenge 2018\n\nCT-perfusion (CTP) infarct-core segmentation in acute anterior-circulation ischemic\nstroke. 103 patients / 156 CTP acquisitions (\"cases\") pooled from the DEFUSE 2 study\nand a companion prospective cohort (4 centers: 3 US, 1 Australian; Cereda et al. 2016).\nBecause CTP z-coverage was limited (4.4-16 cm), some patients were scanned as **two\nseparate slabs (A/B)** — each slab is its own `case_N` folder.\n\n| Split | Cases | Patients | Two-slab patients | Ground truth |\n|---|---|---|---|---|\n| TRAINING | 94 | 63 | 31 | yes (`OT`) |\n| TESTING  | 62 | 40 | 22 | **withheld** (never released; SMIR leaderboard is dead) |\n\n**Grouping warning:** `case_N` is NOT a patient. Group on the `patient_id` column of\n`cases.jsonl` (derived from the SMIR sidecar descriptions, e.g. `Train_40_A` = `case_64`)\nto avoid slab-pair leakage across evaluation splits.\n\n## Per-case channels\n\n| Folder | Content |\n|---|---|\n| `CT` | baseline CT of the CTP study (~2 MB) |\n| `CT_4DPWI` | raw 4D CTP source series, motion-corrected, ~1 s temporal resolution (22-336 MB) |\n| `CT_CBF`, `CT_CBV`, `CT_MTT`, `CT_Tmax` | derived perfusion parameter maps |\n| `OT` | binary infarct-core mask (TRAINING only) |\n\nThe `OT` masks were **manually drawn on MR DWI trace images acquired within 3 h after\nCTP** (median CT-to-MRI 36 min) by a single stroke neurologist with group review,\nblinded to CTP, then registered to CTP space. DWI itself is NOT distributed — the task\nis cross-modality: CTP inputs, DWI-defined target (challenge-winner Dice ~0.51).\n\n## Layout\n\n```\nTRAINING/case_<1..94>/SMIR.Brain.XX.O.<MOD>.<smir_id>/SMIR.Brain.XX.O.<MOD>.<smir_id>.nii  (+ .json sidecar)\nTESTING/case_<1..62>/...   (same, no OT)\ncases.jsonl                 one row per case: case_id, split, smir_case_label, patient_id,\n                            slab, images{mod:path}, mask, has_mask, smir_ids\n```\n\n## Provenance & integrity\n\n- Mirrored from the **organizers' own re-release**: Zenodo DOI\n  [10.5281/zenodo.17736412](https://doi.org/10.5281/zenodo.17736412) (published 2025-11-27\n  by M. Reyes, E. de la Rosa, B. Menze after smir.ch went offline).\n  Zip md5: Training `b6d16c456e173743e30a3fbdb3bd29a5`, Testing `932a493c105024942700fe589cd441a9`.\n- Case counts verified against Hakim et al. 2021 (94/62 cases, 63/40 patients).\n- Note: the Zenodo record's prose lumps 2018 under \"multispectral MRI\" — that is a\n  boilerplate error; ISLES 2018 is CT perfusion.\n- Modifications vs the archives: the 1,030 identical 285-byte `License_ODC_ODBL.txt`\n  copies are deduplicated to the root `LICENSE_ODC_ODBL.txt`; `.DS_Store` dropped.\n  All NIfTI volumes and JSON sidecars are byte-identical to the Zenodo zips.\n- Cohort is disjoint from ISLES 2015/2016/2017 (Bern/Freiburg MRI), ISLES 2022 and\n  ISLES 2024 — no cross-edition patient overlap is documented.\n\n## License\n\nEvery SMIR data folder embeds the **Open Database License (ODbL) 1.0** (contents:\nDatabase Contents License 1.0) — see `LICENSE_ODC_ODBL.txt`. The organizers' Zenodo\nre-release is","encoding":"utf8","key":"README.md","name":"README.md","size":5151,"truncated":true}],"sourceObjectKey":"cases.jsonl","starterCell":"from datasets import load_dataset\nds = load_dataset(\"MedOtter/isles2018\")\nprint(ds)","starterSamplePath":"TESTING/case_1/SMIR.Brain.XX.O.CT.346291/SMIR.Brain.XX.O.CT.346291.nii","totalObjects":2066},"summary":"The Ischemic Stroke Lesion Segmentation (ISLES) 2018 challenge data: CT perfusion (CTP) of acute anterior-circulation ischemic stroke for infarct core segmentation, from 103…","title":"ISLES 2018: CT perfusion of acute ischemic stroke with DWI-defined infarct core masks (103 patients, 156 cases)","updatedAt":"2026-11-17T23:10:09.428Z","visibility":"public","websiteUrl":"https://huggingface.co/datasets/MedOtter/isles2018"}}