fidoptMRM NeAt mouse brain atlas
T2-weighted MR of the head of an adult male C57BL/6J mouse at 9.4 T, alive, with 20 brain structures labelled a side (MRM NeAt in vivo atlas, Ma et al. 2008, as re-oriented by Ma at UCL).
10 test casesmouseimage MRresearch only
Read before comparing (7)
- Pool. Ma et al. imaged 12 male C57BL/6J mice alive at 9.4 T (T2-weighted 3D spin echo, 0.1 mm); the repository holds 10 of them (A1 and A10 are absent), the test cases. The development cases are the 10 formalin-fixed, excised brains of the 2005 atlas (17.6 T, 0.047 mm), other animals and another setting: no skull, finer voxels, other contrast.
- Truth. On each brain an atlas was registered, then the structures refined by hand in Amira beside histological atlases; the papers give no agreement between readers. Graded semi-automatic.
- Labels. 1 to 20 are on the right and 21 to 40 on the left, as the header puts them on every case (olfactory bulbs anterior, neocortex dorsal, a proper rotation); docs/structure_label_list.csv agrees, docs/structure_label_list_hemisphere_separated.csv says the reverse. 2 (corpus callosum and external capsule), 10 (ventricles) and 17 (brainstem) cover both sides: the provider merged 22, 30 and 37 into them. The smallest structures, the anterior commissure and the fimbria, hold 0.3 to 1.2 µl.
- Images. The repository's templates byte for byte: in vivo the whole head, values without unit from -339 to 70,061; ex vivo the excised brain on a near-zero background. The repository's brain masks are not given.
- Start. Over the 10 test cases a whole-image mask scores a mean Dice over the 37 structures of 0.006 at the median, an Otsu mask 0.027 (0.024 to 0.037).
- Licence. The repository is under GPL-3.0, its README asking to cite Ma et al.: research board only.
- Overlap. No other dataset of the corpus holds NeAt; MouseBrainMR is MMB-MRI, other mice. RS2-Net and the authors' multi-atlas tools used these brains as atlases: a method built on them may have seen the test cases.
No method has a result on every case of this dataset yet.
Case by case
Click a score to look at that result: the image, the truth in green and the method's map in red. Underlined scores have their image; for the others you see the case without the result.
About the data
- Test cases
- The 10 in vivo mice of the repository are the test cases, the 10 ex vivo brains of the 2005 atlas, other animals, the development cases. One case a mouse.
- Truth
- contours by a program, verified by people the atlas's 20 structures, split into left and right by D. Ma, the corpus callosum and external capsule, the ventricles and the brainstem kept whole: on each brain an atlas registered and then refined by hand in Amira, with histological atlases beside (Ma et al. 2005 and 2008)
- Source
- github.com/dama-lab/mouse-brain-atlas/tree/61d2e0155fc8a0cb77a21fdd7156436574083178/NeAt
doi.org/10.3389/neuro.05.001.2008
doi.org/10.1016/j.neuroscience.2005.07.014
doi.org/10.1371/journal.pone.0086576 - Licence
- GNU GPL 3.0
10 cases for tuning, never tested
exvivo_A7, exvivo_A9, exvivo_A5, exvivo_A0, exvivo_A1, exvivo_A3, exvivo_A4, exvivo_A8, exvivo_A2, exvivo_A6