Datasets:

Angelou0516
/

BONBID-HIE

subject_id stringlengths 11 11	split stringclasses 1 value	num_slices int32 16 64	lesion_voxels int32 0 214k	preview_slice_idx int32 4 37
MGHNICU_010	train	23	13,038	14
MGHNICU_014	train	23	14,741	13
MGHNICU_015	train	23	12,325	13
MGHNICU_028	train	23	2,711	13
MGHNICU_062	train	19	27,552	11
MGHNICU_064	train	46	57,868	15
MGHNICU_069	train	45	54,405	33
MGHNICU_070	train	48	2,459	27
MGHNICU_071	train	48	45	5
MGHNICU_072	train	45	213,718	23
MGHNICU_074	train	42	33,689	23
MGHNICU_077	train	40	43,376	28
MGHNICU_101	train	43	91,094	25
MGHNICU_105	train	48	138,226	25
MGHNICU_109	train	40	212,444	19
MGHNICU_125	train	46	12,683	37
MGHNICU_144	train	49	41	26
MGHNICU_153	train	46	922	9
MGHNICU_173	train	64	6,825	32
MGHNICU_174	train	64	753	21
MGHNICU_178	train	56	3,922	15
MGHNICU_197	train	45	612	18
MGHNICU_209	train	28	0	14
MGHNICU_214	train	25	317	15
MGHNICU_215	train	27	0	13
MGHNICU_231	train	54	0	27
MGHNICU_234	train	52	36,459	24
MGHNICU_262	train	60	3,513	27
MGHNICU_263	train	50	38	15
MGHNICU_264	train	52	827	33
MGHNICU_272	train	25	408	7
MGHNICU_275	train	54	40	14
MGHNICU_286	train	64	1,515	21
MGHNICU_290	train	54	52	17
MGHNICU_299	train	60	1,565	28
MGHNICU_303	train	18	370	8
MGHNICU_305	train	60	51,515	30
MGHNICU_306	train	57	34,244	30
MGHNICU_309	train	50	11	23
MGHNICU_310	train	59	90	21
MGHNICU_312	train	50	23	10
MGHNICU_323	train	55	122	11
MGHNICU_329	train	50	0	25
MGHNICU_331	train	50	14	16
MGHNICU_336	train	50	75	8
MGHNICU_338	train	45	1,379	8
MGHNICU_346	train	56	23	19
MGHNICU_348	train	52	75	21
MGHNICU_350	train	34	3,729	20
MGHNICU_355	train	50	94	19
MGHNICU_358	train	50	144	21
MGHNICU_359	train	52	38	19
MGHNICU_362	train	50	12	18
MGHNICU_370	train	53	24	18
MGHNICU_373	train	17	393	12
MGHNICU_376	train	24	56	5
MGHNICU_377	train	16	14	5
MGHNICU_378	train	24	245	8
MGHNICU_379	train	21	2,139	6
MGHNICU_393	train	52	70	19
MGHNICU_403	train	52	41	31
MGHNICU_405	train	23	6,733	11
MGHNICU_415	train	49	59,736	25
MGHNICU_432	train	17	0	8
MGHNICU_433	train	23	283	8
MGHNICU_435	train	28	202	11
MGHNICU_437	train	19	1,162	10
MGHNICU_438	train	16	644	4
MGHNICU_439	train	28	106	17
MGHNICU_440	train	20	609	5
MGHNICU_441	train	18	2,521	10
MGHNICU_442	train	25	0	12
MGHNICU_444	train	25	52	18
MGHNICU_445	train	20	491	11
MGHNICU_446	train	16	189	10
MGHNICU_447	train	19	21	11
MGHNICU_448	train	18	11	5
MGHNICU_449	train	20	140	7
MGHNICU_452	train	20	13	7
MGHNICU_453	train	20	13	8
MGHNICU_454	train	17	61	6
MGHNICU_455	train	22	117	11
MGHNICU_456	train	17	3,526	5
MGHNICU_457	train	20	44	11
MGHNICU_458	train	28	0	14

BONBID-HIE

BONBID-HIE (BOston Neonatal Brain Injury Dataset for Hypoxic Ischemic Encephalopathy) is a curated MRI dataset for neonatal HIE lesion segmentation, released as part of the MICCAI 2023 BONBID-HIE challenge.

Dataset Summary

Field	Details
Modality	Diffusion MRI (ADC-derived maps)
Body Part	Neonatal brain (term/late-preterm with HIE)
Subjects (Train)	85
Subjects (Val)	4 (Docker sanity-check split)
Subjects (Test)	44 (password-encrypted, not redistributed here)
Format	MetaImage `.mha` (3D volumes)
Total Size	~1.3 GB
Scanners	GE 1.5T Signa, Siemens 3T Trio
License	CC BY-NC-ND 4.0

Data Structure

Each split contains three parallel directories:

1ADC_ss/ — skull-stripped Apparent Diffusion Coefficient map (model input)
2Z_ADC/ — Z-score normalized ADC map (additional input/aid; NOT ground truth)
3LABEL/ — manual expert lesion annotation (recommended ground truth, train/val only)

File naming:

1ADC_ss/MGHNICU_{ID}-VISIT_01-ADC_ss.mha
2Z_ADC/Zmap_MGHNICU_{ID}-VISIT_01-ADC_smooth2mm_clipped10.mha
3LABEL/MGHNICU_{ID}-VISIT_01_lesion.mha

Plus BONBID2023_clinicaldata_val.xlsx (clinical metadata for the val split).

Ground Truth

The recommended ground truth is the manual expert lesion annotation in 3LABEL/, drawn by a trained physician using MRICroN. For 27 uncertain cases, consensus was reached among three pediatric neuroradiologists. The 2Z_ADC/ map is provided as an algorithm-development aid and is NOT a ground-truth annotation.

Notes

Test split is omitted: it was distributed only to MICCAI 2023 challenge participants and is password-encrypted on Zenodo. Training samples (n=85) plus validation (n=4) are reproduced here.
Val split is small (n=4) — intended as a Docker sanity-check, not a statistical validation set. Cross-validation on the train split is the typical evaluation strategy.

Citation

@article{bao2024bonbid,
  title   = {{BOston Neonatal Brain Injury Data for Hypoxic Ischemic Encephalopathy (BONBID-HIE): I. MRI and Lesion Labeling}},
  author  = {Bao, Rina and Song, Ya'nan and Bates, Sara V. and others},
  journal = {Scientific Data},
  publisher = {Nature},
  year    = {2024},
  doi     = {10.1038/s41597-024-03986-7},
  url     = {https://www.nature.com/articles/s41597-024-03986-7}
}

Source

Original release: Zenodo record 10602767 (V3, paper-cited) Challenge portal: bonbid-hie2023.grand-challenge.org

Downloads last month: 28

Number of rows:

Total file size:

1.32 GB