def test_affine(self, input_param, input_data, expected_val):
g = Affined(**input_param)
result = g(input_data)["img"]
self.assertEqual(isinstance(result, torch.Tensor), isinstance(expected_val, torch.Tensor))
if isinstance(result, torch.Tensor):
np.testing.assert_allclose(result.cpu().numpy(), expected_val.cpu().numpy(), rtol=1e-4, atol=1e-4)
else:
np.testing.assert_allclose(result, expected_val, rtol=1e-4, atol=1e-4)
def test_affine(self, input_param, input_data, expected_val):
input_copy = deepcopy(input_data)
g = Affined(**input_param)
result = g(input_data)
test_local_inversion(g, result, input_copy, dict_key="img")
assert_allclose(result["img"],
expected_val,
rtol=1e-4,
atol=1e-4,
type_test="tensor")
TESTS.append((
"RandRotated 3d",
"3D",
1e-1,
RandRotated(KEYS, *[random.uniform(np.pi / 6, np.pi) for _ in range(3)],
1), # type: ignore
))
TESTS.append((
"Affine 3d",
"3D",
1e-1,
Affined(
KEYS,
spatial_size=[155, 179, 192],
rotate_params=[np.pi / 6, -np.pi / 5, np.pi / 7],
shear_params=[0.5, 0.5],
translate_params=[10, 5, -4],
scale_params=[0.8, 1.3],
),
))
TESTS.append((
"RandAffine 3d",
"3D",
1e-1,
RandAffined(
KEYS,
[155, 179, 192],
prob=1,
padding_mode="zeros",
rotate_range=[np.pi / 6, -np.pi / 5, np.pi / 7],
def test_affine(self, input_param, input_data, expected_val):
g = Affined(**input_param)
result = g(input_data)["img"]
assert_allclose(result, expected_val, rtol=1e-4, atol=1e-4)
def image_mixing(data, seed=None):
#random.seed(seed)
file_list = [x for x in data if int(x['_label']) == 1]
random.shuffle(file_list)
crop_foreground = CropForegroundd(keys=["image"],
source_key="image",
margin=(0, 0, 0),
select_fn=lambda x: x != 0)
WW, WL = 1500, -600
ct_window = CTWindowd(keys=["image"], width=WW, level=WL)
resize2 = Resized(keys=["image"],
spatial_size=(int(512 * 0.75), int(512 * 0.75), -1),
mode="area")
resize1 = Resized(keys=["image"],
spatial_size=(-1, -1, 40),
mode="nearest")
gauss = GaussianSmooth(sigma=(1., 1., 0))
gauss2 = GaussianSmooth(sigma=(2.0, 2.0, 0))
affine = Affined(keys=["image"],
scale_params=(1.0, 2.0, 1.0),
padding_mode='zeros')
common_transform = Compose([
LoadImaged(keys=["image"]),
ct_window,
CTSegmentation(keys=["image"]),
AddChanneld(keys=["image"]),
affine,
crop_foreground,
resize1,
resize2,
SqueezeDimd(keys=["image"]),
])
dirs = setup_directories()
data_dir = dirs['data']
mixed_images_dir = os.path.join(data_dir, 'mixed_images')
if not os.path.exists(mixed_images_dir):
os.mkdir(mixed_images_dir)
for img1, img2 in itertools.combinations(file_list, 2):
img1 = {'image': img1["image"], 'seg': img1['seg']}
img2 = {'image': img2["image"], 'seg': img2['seg']}
img1_data = common_transform(img1)["image"]
img2_data = common_transform(img2)["image"]
img1_mask, img2_mask = (img1_data > 0), (img2_data > 0)
img_presek = np.logical_and(img1_mask, img2_mask)
img = np.maximum(img_presek * img1_data, img_presek * img2_data)
multi_slice_viewer(img, "img1")
loop = True
while loop:
save = input("Save image [y/n/e]: ")
if save.lower() == 'y':
loop = False
k = str(time.time()).encode('utf-8')
h = blake2b(key=k, digest_size=16)
name = h.hexdigest() + '.nii.gz'
out_path = os.path.join(mixed_images_dir, name)
write_nifti(img, out_path, resample=False)
elif save.lower() == 'n':
loop = False
break
elif save.lower() == 'e':
print("exeting")
exit()
else:
print("wrong input!")
