Angelou0516/Figshare_Brain_Tumor

Figshare Brain Tumor Dataset (Cheng et al. 2015)

Brain tumor MRI dataset from Cheng et al., hosted on Figshare. 3064 T1-weighted contrast-enhanced MRI slices from 233 patients with three kinds of brain tumor:

• Meningioma — 708 slices
• Glioma — 1426 slices
• Pituitary tumor — 930 slices

Each slice ships with both a per-pixel binary tumor mask and the original hand-delineated tumor border polyline.

Schema

Column	Type	Description
id	string	Slice ID, "1" to "3064"
pid	string	Patient ID (used to keep splits patient-disjoint)
label_id	int32	1 = meningioma, 2 = glioma, 3 = pituitary
label_name	string	"meningioma" / "glioma" / "pituitary"
fold	int32	Pre-computed 5-fold CV index (1..5, patient-disjoint)
image	Image	16-bit grayscale PNG, 512×512, raw T1-CE MRI
mask	Image	8-bit grayscale PNG, 512×512, 0 = bg / 255 = tumor
tumor_border	list[float]	Polyline coordinates [x1, y1, x2, y2, ...] of the manual border

Splits

Split	Source	Count
train	folds 2-5	2522
test	fold 1	542

The fold column is preserved on every row so users can run other CV protocols (rotate which fold is held out, etc.). Fold assignment is patient-level, so no patient appears in both train and test of any fold.

Acquisition

T1-weighted contrast-enhanced MRI (after Gd-DTPA injection, 0.1 mmol/kg @ 2 ml/s) acquired between 2005-09 and 2010-10 at Nanfang Hospital (Guangzhou) and General Hospital of Tianjin Medical University, China. In-plane 512×512, pixel spacing 0.49×0.49 mm, slice thickness 6 mm, slice gap 1 mm.

Mask sources

The upstream .mat files ship two semantically equivalent encodings of the same manual annotation:

1. cjdata.tumorBorder — raw polyline coordinates as drawn by the radiologist.
2. cjdata.tumorMask — the convenience rasterisation of that polyline as a binary 512×512 image.

This mirror exposes both: mask is the rasterised version (recommended ground truth for segmentation), tumor_border is the raw polyline.

License

CC BY 4.0 — free use including commercial, attribution required.

Citation

@article{cheng2015enhanced,
  title   = {Enhanced Performance of Brain Tumor Classification via Tumor Region Augmentation and Partition},
  author  = {Cheng, Jun and Huang, Wei and Cao, Shuangliang and Yang, Ru and Yang, Wei and Yun, Zhaoqiang and Wang, Zhijian and Feng, Qianjin},
  journal = {PLOS ONE},
  volume  = {10},
  number  = {10},
  pages   = {e0140381},
  year    = {2015},
  doi     = {10.1371/journal.pone.0140381}
}

Source

• Figshare: https://figshare.com/articles/dataset/brain_tumor_dataset/1512427
• DOI: 10.6084/m9.figshare.1512427
• Paper: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0140381