def test(self):
normal_gt = {
'fake_metric_1': {
'std': 1.0,
'mean': 1.0
},
'fake_metric_2': {
'std': 0.5,
'mean': 0.5
}
}
self._test(generate_streamlines(self._tmp_dir), normal_gt)
gt_density_weighting = {
'fake_metric_1': {
'mean': 1.2,
'std': 0.979795897113
},
'fake_metric_2': {
'mean': 0.4,
'std': 0.489897948557
}
}
self._test(generate_streamlines_with_duplicates(self._tmp_dir),
normal_gt, gt_density_weighting)
def test(self):
self._test(
generate_streamlines(self._tmp_dir), {
'fake_metric_1': {
'mean': [1.0] * 5,
'std': [1.0] * 5,
'tractprofile':
[[[0.0, 2.0, 0.0, 0.0, 2.0, 0.0, 2.0, 2.0]] * 5]
},
'fake_metric_2': {
'mean': [0.5] * 5,
'std': [0.5] * 5,
'tractprofile':
[[[1.0, 0.0, 1.0, 1.0, 0.0, 1.0, 0.0, 0.0]] * 5]
}
})
self._test(
generate_streamlines_with_duplicates(self._tmp_dir), {
'fake_metric_1': {
'mean': [1.2] * 5,
'std': [0.979795897113] * 5,
'tractprofile':
[[[2.0, 0.0, 2.0, 2.0, 0.0, 2.0, 2.0, 0.0, 2.0, 0.0]] * 5]
},
'fake_metric_2': {
'mean': [0.4] * 5,
'std': [0.489897948557] * 5,
'tractprofile':
[[[0.0, 1.0, 0.0, 0.0, 1.0, 0.0, 0.0, 1.0, 0.0, 1.0]] * 5]
}
})
def test_non_weighted_with_itselt(self):
reference = generate_fa(self._tmp_dir)
streamlines = generate_streamlines(self._tmp_dir)
gt_dict = {'dice': 1.0}
with RedirectStdOut() as output:
self.call(main_weighted_dice,
streamlines, streamlines, reference)
output_dict = json.loads('\n'.join(output))
self.assertEqual(
output_dict, gt_dict,
"Wrong dice coeff. {} != {}.".format(output_dict['dice'],
gt_dict['dice']))
def test(self):
spacing = (1.0, 1.0, 1.0)
tracts_path = generate_streamlines(self._tmp_dir, spacing)
image_path = os.path.join(self._tmp_dir, 'input.nii.gz')
output_path = os.path.join(self._tmp_dir, 'output.nii.gz')
generate_cross_image((10, 10, 10), spacing, image_path)
self.call(main, tracts_path, image_path, output_path,
tp='trackvis')
self.assertTrue(_check(nib.load(output_path).get_data(), 1),
'Wrong voxels values')
self.call(main, tracts_path, image_path, output_path, '-f',
tp='trackvis', binary=42)
self.assertTrue(_check(nib.load(output_path).get_data(), 42),
'Wrong voxels values')
def test_weighted(self):
reference = generate_fa(self._tmp_dir)
streamlines = generate_streamlines(self._tmp_dir)
streamlines_missing_corner =\
generate_streamlines_missing_corner(self._tmp_dir)
gt_dice = 0.8571 # Confirmed using ANTs ImageMath 3 DiceAndMinDistSum
with RedirectStdOut() as output:
self.call(main_weighted_dice, '--weighted',
streamlines, streamlines_missing_corner, reference)
output_dict = json.loads('\n'.join(output))
output_dice = float(output_dict['dice'])
self.assertAlmostEqual(
output_dice, gt_dice, 4,
"Wrong dice coeff. {} != {}.".format(output_dice,
gt_dice))
def test_distance_weighting(self):
fake1 = {'mean': 2.0 / 3.0, 'std': 0.942809041582}
fake2 = {'mean': 2.0 / 3.0, 'std': 0.471404520791}
bundle_path = generate_streamlines(self._tmp_dir)
self._create_maps(bundle_path)
gt = {
'fake_metric_1': {l: fake1
for l in self.labels},
'fake_metric_2': {l: fake2
for l in self.labels}
}
with RedirectStdOut() as output:
self.call(main_meanstdperpoint, '--distance_weighting',
bundle_path, self.label_map, self.distance_map,
*self.metrics)
self._compare(bundle_path, gt, output)
def test_no_weighting(self):
fake1 = {'mean': 1.0, 'std': 1.0}
fake2 = {'mean': 0.5, 'std': 0.5}
bundle_path = generate_streamlines(self._tmp_dir)
self._create_maps(bundle_path)
gt = {
'fake_metric_1': {l: fake1
for l in self.labels},
'fake_metric_2': {l: fake2
for l in self.labels}
}
with RedirectStdOut() as output:
self.call(main_meanstdperpoint, bundle_path, self.label_map,
self.distance_map, *self.metrics)
self._compare(bundle_path, gt, output)