|
1
|
|
|
#!/usr/bin/env python |
|
2
|
|
|
|
|
3
|
|
|
"""Tests for Image.""" |
|
4
|
|
|
|
|
5
|
|
|
import copy |
|
6
|
|
|
import torch |
|
7
|
|
|
import numpy as np |
|
8
|
|
|
from torchio import ScalarImage, LabelMap, Subject, INTENSITY, LABEL, STEM |
|
9
|
|
|
from ..utils import TorchioTestCase |
|
10
|
|
|
from torchio import RandomFlip, RandomAffine |
|
11
|
|
|
|
|
12
|
|
|
|
|
13
|
|
|
class TestImage(TorchioTestCase): |
|
14
|
|
|
"""Tests for `Image`.""" |
|
15
|
|
|
|
|
16
|
|
|
def test_image_not_found(self): |
|
17
|
|
|
with self.assertRaises(FileNotFoundError): |
|
18
|
|
|
ScalarImage('nopath') |
|
19
|
|
|
|
|
20
|
|
|
def test_wrong_path_value(self): |
|
21
|
|
|
with self.assertRaises(RuntimeError): |
|
22
|
|
|
ScalarImage('~&./@#"!?X7=+') |
|
23
|
|
|
|
|
24
|
|
|
def test_wrong_path_type(self): |
|
25
|
|
|
with self.assertRaises(TypeError): |
|
26
|
|
|
ScalarImage(5) |
|
27
|
|
|
|
|
28
|
|
|
def test_wrong_affine(self): |
|
29
|
|
|
with self.assertRaises(TypeError): |
|
30
|
|
|
ScalarImage(5, affine=1) |
|
31
|
|
|
|
|
32
|
|
|
def test_tensor_flip(self): |
|
33
|
|
|
sample_input = torch.ones((4, 30, 30, 30)) |
|
34
|
|
|
RandomFlip()(sample_input) |
|
35
|
|
|
|
|
36
|
|
|
def test_tensor_affine(self): |
|
37
|
|
|
sample_input = torch.ones((4, 10, 10, 10)) |
|
38
|
|
|
RandomAffine()(sample_input) |
|
39
|
|
|
|
|
40
|
|
|
def test_crop_attributes(self): |
|
41
|
|
|
cropped = self.sample.crop((1, 1, 1), (5, 5, 5)) |
|
42
|
|
|
self.assertIs(self.sample.t1['pre_affine'], cropped.t1['pre_affine']) |
|
43
|
|
|
|
|
44
|
|
|
def test_crop_does_not_create_wrong_path(self): |
|
45
|
|
|
data = torch.ones((10, 10, 10)) |
|
46
|
|
|
image = ScalarImage(tensor=data) |
|
47
|
|
|
cropped = image.crop((1, 1, 1), (5, 5, 5)) |
|
48
|
|
|
self.assertIs(cropped.path, None) |
|
49
|
|
|
|
|
50
|
|
|
def test_scalar_image_type(self): |
|
51
|
|
|
data = torch.ones((10, 10, 10)) |
|
52
|
|
|
image = ScalarImage(tensor=data) |
|
53
|
|
|
self.assertIs(image.type, INTENSITY) |
|
54
|
|
|
|
|
55
|
|
|
def test_label_map_type(self): |
|
56
|
|
|
data = torch.ones((10, 10, 10)) |
|
57
|
|
|
label = LabelMap(tensor=data) |
|
58
|
|
|
self.assertIs(label.type, LABEL) |
|
59
|
|
|
|
|
60
|
|
|
def test_wrong_scalar_image_type(self): |
|
61
|
|
|
data = torch.ones((10, 10, 10)) |
|
62
|
|
|
with self.assertRaises(ValueError): |
|
63
|
|
|
ScalarImage(tensor=data, type=LABEL) |
|
64
|
|
|
|
|
65
|
|
|
def test_wrong_label_map_type(self): |
|
66
|
|
|
data = torch.ones((10, 10, 10)) |
|
67
|
|
|
with self.assertRaises(ValueError): |
|
68
|
|
|
LabelMap(tensor=data, type=INTENSITY) |
|
69
|
|
|
|
|
70
|
|
|
def test_crop_scalar_image_type(self): |
|
71
|
|
|
data = torch.ones((10, 10, 10)) |
|
72
|
|
|
image = ScalarImage(tensor=data) |
|
73
|
|
|
cropped = image.crop((1, 1, 1), (5, 5, 5)) |
|
74
|
|
|
self.assertIs(cropped.type, INTENSITY) |
|
75
|
|
|
|
|
76
|
|
|
def test_crop_label_map_type(self): |
|
77
|
|
|
data = torch.ones((10, 10, 10)) |
|
78
|
|
|
label = LabelMap(tensor=data) |
|
79
|
|
|
cropped = label.crop((1, 1, 1), (5, 5, 5)) |
|
80
|
|
|
self.assertIs(cropped.type, LABEL) |
|
81
|
|
|
|
|
82
|
|
|
def test_no_input(self): |
|
83
|
|
|
with self.assertRaises(ValueError): |
|
84
|
|
|
image = ScalarImage() |
|
85
|
|
|
|
|
86
|
|
|
def test_bad_key(self): |
|
87
|
|
|
with self.assertRaises(ValueError): |
|
88
|
|
|
image = ScalarImage(path='', data=5) |
|
89
|
|
|
|
|
90
|
|
|
def test_repr(self): |
|
91
|
|
|
sample = Subject(t1=ScalarImage(self.get_image_path('repr_test'))) |
|
92
|
|
|
assert 'shape' not in repr(sample['t1']) |
|
93
|
|
|
sample.load() |
|
94
|
|
|
assert 'shape' in repr(sample['t1']) |
|
95
|
|
|
|
|
96
|
|
|
def test_data_tensor(self): |
|
97
|
|
|
sample = copy.deepcopy(self.sample) |
|
98
|
|
|
sample.load() |
|
99
|
|
|
self.assertIs(sample.t1.data, sample.t1.tensor) |
|
100
|
|
|
|
|
101
|
|
|
def test_bad_affine(self): |
|
102
|
|
|
with self.assertRaises(ValueError): |
|
103
|
|
|
ScalarImage(tensor=torch.rand(1, 2, 3, 4), affine=np.eye(3)) |
|
104
|
|
|
|
|
105
|
|
|
def test_nans_tensor(self): |
|
106
|
|
|
tensor = np.random.rand(1, 2, 3, 4) |
|
107
|
|
|
tensor[0, 0, 0, 0] = np.nan |
|
108
|
|
|
with self.assertWarns(UserWarning): |
|
109
|
|
|
image = ScalarImage(tensor=tensor) |
|
110
|
|
|
image.set_check_nans(False) |
|
111
|
|
|
|
|
112
|
|
|
def test_nans_file(self): |
|
113
|
|
|
image = ScalarImage(self.get_image_path('repr_test', add_nans=True)) |
|
114
|
|
|
with self.assertWarns(UserWarning): |
|
115
|
|
|
image.load() |
|
116
|
|
|
|
|
117
|
|
|
def test_get_center(self): |
|
118
|
|
|
tensor = torch.rand(1, 3, 3, 3) |
|
119
|
|
|
image = ScalarImage(tensor=tensor) |
|
120
|
|
|
ras = image.get_center() |
|
121
|
|
|
lps = image.get_center(lps=True) |
|
122
|
|
|
self.assertEqual(ras, (1, 1, 1)) |
|
123
|
|
|
self.assertEqual(lps, (-1, -1, 1)) |
|
124
|
|
|
|
|
125
|
|
|
def test_with_list_of_missing_files(self): |
|
126
|
|
|
with self.assertRaises(FileNotFoundError): |
|
127
|
|
|
ScalarImage(path=['nopath', 'error']) |
|
128
|
|
|
|
|
129
|
|
|
def test_with_a_list_of_paths(self): |
|
130
|
|
|
shape = (5, 5, 5) |
|
131
|
|
|
path1 = self.get_image_path('path1', shape=shape) |
|
132
|
|
|
path2 = self.get_image_path('path2', shape=shape) |
|
133
|
|
|
image = ScalarImage(path=[path1, path2]) |
|
134
|
|
|
self.assertEqual(image.shape, (2, 5, 5, 5)) |
|
135
|
|
|
self.assertEqual(image[STEM], ['path1', 'path2']) |
|
136
|
|
|
|
|
137
|
|
|
def test_with_a_list_of_images_with_different_shapes(self): |
|
138
|
|
|
path1 = self.get_image_path('path1', shape=(5, 5, 5)) |
|
139
|
|
|
path2 = self.get_image_path('path2', shape=(7, 5, 5)) |
|
140
|
|
|
image = ScalarImage(path=[path1, path2]) |
|
141
|
|
|
with self.assertRaises(RuntimeError): |
|
142
|
|
|
image.load() |
|
143
|
|
|
|
|
144
|
|
|
def test_with_a_list_of_images_with_different_affines(self): |
|
145
|
|
|
path1 = self.get_image_path('path1', spacing=(1, 1, 1)) |
|
146
|
|
|
path2 = self.get_image_path('path2', spacing=(1, 2, 1)) |
|
147
|
|
|
image = ScalarImage(path=[path1, path2]) |
|
148
|
|
|
with self.assertWarns(RuntimeWarning): |
|
149
|
|
|
image.load() |
|
150
|
|
|
|
|
151
|
|
|
def test_with_a_list_of_2d_paths(self): |
|
152
|
|
|
shape = (5, 6) |
|
153
|
|
|
path1 = self.get_image_path('path1', shape=shape, suffix='.nii') |
|
154
|
|
|
path2 = self.get_image_path('path2', shape=shape, suffix='.img') |
|
155
|
|
|
path3 = self.get_image_path('path3', shape=shape, suffix='.hdr') |
|
156
|
|
|
image = ScalarImage(path=[path1, path2, path3]) |
|
157
|
|
|
self.assertEqual(image.shape, (3, 5, 6, 1)) |
|
158
|
|
|
self.assertEqual(image[STEM], ['path1', 'path2', 'path3']) |
|
159
|
|
|
|
|
160
|
|
|
def test_axis_name_2d(self): |
|
161
|
|
|
path = self.get_image_path('im2d', shape=(5, 6)) |
|
162
|
|
|
image = ScalarImage(path) |
|
163
|
|
|
height_idx = image.axis_name_to_index('h') |
|
164
|
|
|
width_idx = image.axis_name_to_index('w') |
|
165
|
|
|
self.assertEqual(image.height, image.shape[height_idx]) |
|
166
|
|
|
self.assertEqual(image.width, image.shape[width_idx]) |
|
167
|
|
|
|
fepegar /
torchio
