|
1
|
|
|
import urllib.parse |
|
2
|
|
|
import torch |
|
3
|
|
|
from torchvision.datasets.utils import download_and_extract_archive |
|
4
|
|
|
from ...utils import get_torchio_cache_dir, compress |
|
5
|
|
|
from ... import ScalarImage, LabelMap, DATA |
|
6
|
|
|
from .mni import SubjectMNI |
|
7
|
|
|
|
|
8
|
|
|
|
|
9
|
|
|
class ICBM2009CNonlinearSymmetric(SubjectMNI): |
|
10
|
|
|
r"""ICBM template. |
|
11
|
|
|
|
|
12
|
|
|
More information can be found in the `website |
|
13
|
|
|
<http://www.bic.mni.mcgill.ca/ServicesAtlases/ICBM152NLin2009>`_. |
|
14
|
|
|
|
|
15
|
|
|
.. image:: http://www.bic.mni.mcgill.ca/uploads/ServicesAtlases/mni_icbm152_sym_09c_small.jpg |
|
16
|
|
|
:alt: ICBM 2009c Nonlinear Symmetric |
|
17
|
|
|
|
|
18
|
|
|
Args: |
|
19
|
|
|
load_4d_tissues: If ``True``, the tissue probability maps will be loaded |
|
20
|
|
|
together into a 4D image. Otherwise, they will be loaded into |
|
21
|
|
|
independent images. |
|
22
|
|
|
|
|
23
|
|
|
Example: |
|
24
|
|
|
>>> import torchio |
|
25
|
|
|
>>> icbm = torchio.datasets.ICBM2009CNonlinearSymmetric() |
|
26
|
|
|
>>> icbm |
|
27
|
|
|
ICBM2009CNonlinearSymmetric(Keys: ('t1', 'eyes', 'face', 'brain', 't2', 'pd', 'tissues'); images: 7) |
|
28
|
|
|
>>> icbm = torchio.datasets.ICBM2009CNonlinearSymmetric(load_4d_tissues=False) |
|
29
|
|
|
>>> icbm |
|
30
|
|
|
ICBM2009CNonlinearSymmetric(Keys: ('t1', 'eyes', 'face', 'brain', 't2', 'pd', 'gm', 'wm', 'csf'); images: 9) |
|
31
|
|
|
|
|
32
|
|
|
""" |
|
33
|
|
|
def __init__(self, load_4d_tissues: bool = True): |
|
34
|
|
|
self.name = 'mni_icbm152_nlin_sym_09c_nifti' |
|
35
|
|
|
self.url_base = 'http://www.bic.mni.mcgill.ca/~vfonov/icbm/2009/' |
|
36
|
|
|
self.filename = f'{self.name}.zip' |
|
37
|
|
|
self.url = urllib.parse.urljoin(self.url_base, self.filename) |
|
38
|
|
|
download_root = get_torchio_cache_dir() / self.name |
|
39
|
|
|
if download_root.is_dir(): |
|
40
|
|
|
print(f'Using cache found in {download_root}') # noqa: T001 |
|
41
|
|
|
else: |
|
42
|
|
|
download_and_extract_archive( |
|
43
|
|
|
self.url, |
|
44
|
|
|
download_root=download_root, |
|
45
|
|
|
filename=self.filename, |
|
46
|
|
|
remove_finished=True, |
|
47
|
|
|
) |
|
48
|
|
|
|
|
49
|
|
|
files_dir = download_root / 'mni_icbm152_nlin_sym_09c' |
|
50
|
|
|
|
|
51
|
|
|
p = files_dir / 'mni_icbm152' |
|
52
|
|
|
m = 'tal_nlin_sym_09c' |
|
53
|
|
|
s = '.nii.gz' |
|
54
|
|
|
|
|
55
|
|
|
tissues_path = files_dir / f'{p}_tissues_{m}.nii.gz' |
|
56
|
|
|
if not tissues_path.is_file(): |
|
57
|
|
|
gm = LabelMap(f'{p}_gm_{m}.nii') |
|
58
|
|
|
wm = LabelMap(f'{p}_wm_{m}.nii') |
|
59
|
|
|
csf = LabelMap(f'{p}_csf_{m}.nii') |
|
60
|
|
|
gm[DATA] = torch.cat((gm[DATA], wm[DATA], csf[DATA])) |
|
|
|
|
|
|
61
|
|
|
gm.save(tissues_path) |
|
62
|
|
|
|
|
63
|
|
|
for fp in files_dir.glob('*.nii'): |
|
64
|
|
|
compress(fp, fp.with_suffix('.nii.gz')) |
|
65
|
|
|
fp.unlink() |
|
66
|
|
|
|
|
67
|
|
|
subject_dict = dict( |
|
68
|
|
|
t1=ScalarImage(f'{p}_t1_{m}{s}'), |
|
69
|
|
|
eyes=LabelMap(f'{p}_t1_{m}_eye_mask{s}'), |
|
70
|
|
|
face=LabelMap(f'{p}_t1_{m}_face_mask{s}'), |
|
71
|
|
|
brain=LabelMap(f'{p}_t1_{m}_mask{s}'), |
|
72
|
|
|
t2=ScalarImage(f'{p}_t2_{m}{s}'), |
|
73
|
|
|
pd=ScalarImage(f'{p}_csf_{m}{s}'), |
|
74
|
|
|
) |
|
75
|
|
|
if load_4d_tissues: |
|
76
|
|
|
subject_dict['tissues'] = LabelMap(tissues_path) |
|
77
|
|
|
else: |
|
78
|
|
|
subject_dict['gm'] = LabelMap(f'{p}_gm_{m}{s}') |
|
79
|
|
|
subject_dict['wm'] = LabelMap(f'{p}_wm_{m}{s}') |
|
80
|
|
|
subject_dict['csf'] = LabelMap(f'{p}_csf_{m}{s}') |
|
81
|
|
|
|
|
82
|
|
|
super().__init__(subject_dict) |
|
83
|
|
|
|
fepegar /
torchio
Loading history...