1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
|
import astra
import numpy as np
import pytest
DET_SPACING_X = 1.0
DET_SPACING_Y = 1.0
DET_ROW_COUNT = 20
DET_COL_COUNT = 45
N_ANGLES = 180
ANGLES = np.linspace(0, 2 * np.pi, N_ANGLES, endpoint=False)
SOURCE_ORIGIN = 100
ORIGIN_DET = 100
N_ROWS = 40
N_COLS = 30
N_SLICES = 50
VOL_SHIFT = 1, 2, 3
VOL_GEOM = astra.create_vol_geom(
N_ROWS, N_COLS, N_SLICES,
-N_COLS/2 + VOL_SHIFT[0], N_COLS/2 + VOL_SHIFT[0],
-N_ROWS/2 + VOL_SHIFT[1], N_ROWS/2 + VOL_SHIFT[1],
-N_SLICES/2 + VOL_SHIFT[2], N_SLICES/2 + VOL_SHIFT[2]
)
DATA_INIT_VALUE = 1.0
@pytest.fixture
def proj_geom(request):
geometry_type = request.param
if geometry_type == 'parallel3d':
return astra.create_proj_geom('parallel3d', DET_SPACING_X, DET_SPACING_Y,
DET_ROW_COUNT, DET_COL_COUNT, ANGLES)
elif geometry_type == 'parallel3d_vec':
geom = astra.create_proj_geom('parallel3d', DET_SPACING_X, DET_SPACING_Y,
DET_ROW_COUNT, DET_COL_COUNT, ANGLES)
return astra.geom_2vec(geom)
elif geometry_type == 'cone':
return astra.create_proj_geom('cone', DET_SPACING_X, DET_SPACING_Y,
DET_ROW_COUNT, DET_COL_COUNT, ANGLES,
SOURCE_ORIGIN, ORIGIN_DET)
elif geometry_type == 'cone_vec':
geom = astra.create_proj_geom('cone', DET_SPACING_X, DET_SPACING_Y,
DET_ROW_COUNT, DET_COL_COUNT, ANGLES,
SOURCE_ORIGIN, ORIGIN_DET)
return astra.geom_2vec(geom)
elif geometry_type == 'short_scan':
cone_angle = np.arctan2(0.5 * DET_COL_COUNT * DET_SPACING_X, SOURCE_ORIGIN + ORIGIN_DET)
angles = np.linspace(0, np.pi + 2 * cone_angle, 180)
return astra.create_proj_geom('cone', DET_SPACING_X, DET_SPACING_Y,
DET_ROW_COUNT, DET_COL_COUNT, angles,
SOURCE_ORIGIN, ORIGIN_DET)
def _fourier_space_filter(proj_geom):
# The full filter size should be the smallest power of two that is at least
# twice the number of detector pixels
full_filter_size = int(2 ** np.ceil(np.log2(2 * proj_geom['DetectorColCount'])))
half_filter_size = full_filter_size // 2 + 1
return np.linspace(0, 1, half_filter_size).reshape(1, -1)
def _real_space_filter(proj_geom):
n = proj_geom['DetectorColCount']
kernel = np.zeros([1, n])
for i in range(n//4):
f = np.pi * (2*i + 1)
val = -2.0 / (f * f)
kernel[0, n//2 + (2*i+1)] = val
kernel[0, n//2 - (2*i+1)] = val
kernel[0, n//2] = 0.5
return kernel
@pytest.fixture
def custom_filter(proj_geom, request):
filter_type = request.param
if filter_type == 'projection':
kernel = _fourier_space_filter(proj_geom)
elif filter_type == 'sinogram':
weights = np.random.rand(N_ANGLES)
kernel = np.outer(weights, _fourier_space_filter(proj_geom))
elif filter_type == 'rprojection':
kernel = _real_space_filter(proj_geom)
elif filter_type == 'rsinogram':
weights = np.random.rand(N_ANGLES)
kernel = np.outer(weights, _real_space_filter(proj_geom))
dummy_geom = astra.create_proj_geom('parallel', 1, kernel.shape[1], np.zeros(kernel.shape[0]))
filter_data_id = astra.data2d.create('-sino', dummy_geom, kernel)
yield filter_type, filter_data_id
astra.data2d.delete(filter_data_id)
@pytest.fixture
def sinogram_mask(proj_geom):
mask = np.random.rand(DET_ROW_COUNT, N_ANGLES, DET_COL_COUNT) > 0.1
mask_data_id = astra.data3d.create('-sino', proj_geom, mask)
yield mask_data_id
astra.data3d.delete(mask_data_id)
@pytest.fixture
def reconstruction_mask():
mask = np.random.rand(N_SLICES, N_ROWS, N_COLS) > 0.1
mask_data_id = astra.data3d.create('-vol', VOL_GEOM, mask)
yield mask_data_id
astra.data3d.delete(mask_data_id)
def make_algorithm_config(algorithm_type, proj_geom, options=None):
algorithm_config = astra.astra_dict(algorithm_type)
vol_data_id = astra.data3d.create('-vol', VOL_GEOM, DATA_INIT_VALUE)
if algorithm_type.startswith('FP'):
algorithm_config['VolumeDataId'] = vol_data_id
proj_data_id = astra.data3d.create('-sino', proj_geom, DATA_INIT_VALUE)
else:
algorithm_config['ReconstructionDataId'] = vol_data_id
# Make reconstruction contain negative and large numbers for testing
# min/max constraint options
proj_data = -10 * np.ones([DET_ROW_COUNT, N_ANGLES, DET_COL_COUNT])
proj_data[DET_ROW_COUNT//4:-DET_ROW_COUNT//4:, DET_COL_COUNT//4:-DET_COL_COUNT//4] = 10
proj_data_id = astra.data3d.create('-sino', proj_geom, proj_data)
algorithm_config['ProjectionDataId'] = proj_data_id
if options is not None:
algorithm_config['option'] = options
return algorithm_config
def get_algorithm_output(algorithm_config, n_iter=None):
if n_iter is None:
if algorithm_config['type'] in ['SIRT3D_CUDA', 'CGLS3D_CUDA']:
n_iter = 2
else:
n_iter = 1
algorithm_id = astra.algorithm.create(algorithm_config)
astra.algorithm.run(algorithm_id, n_iter)
if algorithm_config['type'].startswith('FP'):
output = astra.data3d.get(algorithm_config['ProjectionDataId'])
astra.data3d.delete(algorithm_config['VolumeDataId'])
else:
output = astra.data3d.get(algorithm_config['ReconstructionDataId'])
astra.data3d.delete(algorithm_config['ReconstructionDataId'])
astra.data3d.delete(algorithm_config['ProjectionDataId'])
astra.algorithm.delete(algorithm_id)
return output
@pytest.mark.parametrize(
'proj_geom,', ['parallel3d', 'parallel3d_vec', 'cone', 'cone_vec'], indirect=True
)
@pytest.mark.parametrize(
'algorithm_type', ['FP3D_CUDA', 'BP3D_CUDA', 'FDK_CUDA', 'SIRT3D_CUDA', 'CGLS3D_CUDA'],
)
def test_algorithms(proj_geom, algorithm_type):
if algorithm_type == 'FDK_CUDA' and proj_geom['type'] not in ['cone', 'cone_vec']:
pytest.xfail('Not implemented')
algorithm_config = make_algorithm_config(algorithm_type, proj_geom)
output = get_algorithm_output(algorithm_config)
assert not np.allclose(output, DATA_INIT_VALUE)
class TestOptions:
@pytest.mark.parametrize('proj_geom,', ['parallel3d', 'cone'], indirect=True)
@pytest.mark.parametrize('algorithm_type', ['FP3D_CUDA', 'SIRT3D_CUDA', 'CGLS3D_CUDA'])
def test_detector_supersampling_fp(self, proj_geom, algorithm_type):
if algorithm_type == 'FP3D_CUDA':
pytest.xfail('Known bug')
algorithm_no_supersampling = make_algorithm_config(algorithm_type, proj_geom)
algorithm_with_supersampling = make_algorithm_config(algorithm_type, proj_geom,
options={'DetectorSuperSampling': 3})
output_no_supersampling = get_algorithm_output(algorithm_no_supersampling)
output_with_supersampling = get_algorithm_output(algorithm_with_supersampling)
assert not np.allclose(output_with_supersampling, DATA_INIT_VALUE)
assert not np.allclose(output_with_supersampling, output_no_supersampling)
@pytest.mark.parametrize('proj_geom,', ['parallel3d', 'cone'], indirect=True)
@pytest.mark.parametrize(
'algorithm_type', ['BP3D_CUDA', 'FDK_CUDA', 'SIRT3D_CUDA', 'CGLS3D_CUDA']
)
def test_voxel_supersampling(self, proj_geom, algorithm_type):
if algorithm_type in ['BP3D_CUDA', 'FDK_CUDA']:
pytest.xfail('Known bug')
if algorithm_type == 'FDK_CUDA' and proj_geom['type'] == 'parallel3d':
pytest.xfail('Not implemented')
algorithm_no_supersampling = make_algorithm_config(algorithm_type, proj_geom)
algorithm_with_supersampling = make_algorithm_config(algorithm_type, proj_geom,
options={'VoxelSuperSampling': 3})
reconstruction_no_supersampling = get_algorithm_output(algorithm_no_supersampling)
reconstruction_with_supersampling = get_algorithm_output(algorithm_with_supersampling)
assert not np.allclose(reconstruction_with_supersampling, DATA_INIT_VALUE)
assert not np.allclose(reconstruction_with_supersampling, reconstruction_no_supersampling)
@pytest.mark.parametrize('proj_geom', ['cone'], indirect=True)
@pytest.mark.parametrize('filter_type', ['ram-lak', 'none'])
def test_fbp_filters_basic(self, proj_geom, filter_type):
algorithm_config = make_algorithm_config(algorithm_type='FDK_CUDA', proj_geom=proj_geom,
options={'FilterType': filter_type})
reconstruction = get_algorithm_output(algorithm_config)
assert not np.allclose(reconstruction, DATA_INIT_VALUE)
@pytest.mark.parametrize('proj_geom', ['cone'], indirect=True)
@pytest.mark.parametrize('filter_type', ['tukey', 'gaussian', 'blackman', 'kaiser'])
def test_fbp_filter_parameter(self, proj_geom, filter_type):
algorithm_config = make_algorithm_config(
algorithm_type='FDK_CUDA', proj_geom=proj_geom,
options={'FilterType': filter_type, 'FilterParameter': -1.0}
)
reconstruction = get_algorithm_output(algorithm_config)
assert not np.allclose(reconstruction, DATA_INIT_VALUE)
@pytest.mark.parametrize('proj_geom', ['cone'], indirect=True)
@pytest.mark.parametrize('filter_type', ['shepp-logan', 'cosine', 'hamming', 'hann'])
def test_fbp_filter_d(self, proj_geom, filter_type):
algorithm_config = make_algorithm_config(
algorithm_type='FDK_CUDA', proj_geom=proj_geom,
options={'FilterType': filter_type, 'FilterD': 1.0}
)
reconstruction = get_algorithm_output(algorithm_config)
assert not np.allclose(reconstruction, DATA_INIT_VALUE)
@pytest.mark.parametrize('proj_geom', ['cone'], indirect=True)
@pytest.mark.parametrize(
'custom_filter', ['projection', 'sinogram', 'rprojection', 'rsinogram'], indirect=True
)
def test_fbp_custom_filters(self, proj_geom, custom_filter):
filter_type, filter_data_id = custom_filter
algorithm_config = make_algorithm_config(
algorithm_type='FDK_CUDA', proj_geom=proj_geom,
options={'FilterType': filter_type, 'FilterSinogramId': filter_data_id}
)
reconstruction = get_algorithm_output(algorithm_config)
assert not np.allclose(reconstruction, DATA_INIT_VALUE)
@pytest.mark.parametrize('proj_geom', ['short_scan'], indirect=True)
def test_short_scan(self, proj_geom):
algorithm_no_short_scan = make_algorithm_config('FDK_CUDA', proj_geom)
algorithm_with_short_scan = make_algorithm_config('FDK_CUDA', proj_geom,
options={'ShortScan': True})
reconstruction_no_short_scan = get_algorithm_output(algorithm_no_short_scan)
reconstruction_with_short_scan = get_algorithm_output(algorithm_with_short_scan)
assert not np.allclose(reconstruction_with_short_scan, DATA_INIT_VALUE)
assert not np.allclose(reconstruction_with_short_scan, reconstruction_no_short_scan)
@pytest.mark.parametrize('proj_geom,', ['parallel3d'], indirect=True)
def test_min_max_constraint(self, proj_geom):
algorithm_no_constrains = make_algorithm_config('SIRT3D_CUDA', proj_geom)
algorithm_with_constrains = make_algorithm_config(
'SIRT3D_CUDA', proj_geom, options={'MinConstraint': 0.0, 'MaxConstraint': 0.125}
)
reconstruction_no_constrains = get_algorithm_output(algorithm_no_constrains)
reconstruction_with_constrains = get_algorithm_output(algorithm_with_constrains)
assert reconstruction_no_constrains.min() < 0.0
assert reconstruction_no_constrains.max() > 0.125
assert reconstruction_with_constrains.min() == 0.0
assert reconstruction_with_constrains.max() == 0.125
@pytest.mark.parametrize('proj_geom,', ['parallel3d'], indirect=True)
@pytest.mark.parametrize('algorithm_type', ['SIRT3D_CUDA', 'CGLS3D_CUDA'])
def test_reconstruction_mask(self, proj_geom, reconstruction_mask, algorithm_type):
algorithm_config = make_algorithm_config(
algorithm_type, proj_geom, options={'ReconstructionMaskId': reconstruction_mask}
)
reconstruction = get_algorithm_output(algorithm_config)
assert not np.allclose(reconstruction, DATA_INIT_VALUE)
mask = (astra.data3d.get(reconstruction_mask) > 0)
assert np.allclose(reconstruction[~mask], DATA_INIT_VALUE)
@pytest.mark.parametrize('proj_geom,', ['parallel3d'], indirect=True)
def test_sinogram_mask(self, proj_geom, sinogram_mask):
algorithm_no_mask = make_algorithm_config('SIRT3D_CUDA', proj_geom)
algorithm_with_sino_mask = make_algorithm_config('SIRT3D_CUDA', proj_geom,
options={'SinogramMaskId': sinogram_mask})
reconstruction_no_mask = get_algorithm_output(algorithm_no_mask)
reconstruction_with_sino_mask = get_algorithm_output(algorithm_with_sino_mask)
assert not np.allclose(reconstruction_with_sino_mask, DATA_INIT_VALUE)
assert not np.allclose(reconstruction_with_sino_mask, reconstruction_no_mask)
@pytest.mark.parametrize('proj_geom,', ['parallel3d'], indirect=True)
@pytest.mark.parametrize('algorithm_type', ['SIRT3D_CUDA', 'CGLS3D_CUDA'])
def test_get_res_norm(self, proj_geom, algorithm_type):
algorithm_config = make_algorithm_config(algorithm_type, proj_geom)
algorithm_id = astra.algorithm.create(algorithm_config)
astra.algorithm.run(algorithm_id, 2)
res_norm = astra.algorithm.get_res_norm(algorithm_id)
astra.algorithm.delete(algorithm_id)
assert res_norm > 0.0
|
