def test_inverse_array(self, use_compose, dtype, device):
img: MetaTensor
tr = Compose([
AddChannel(),
Orientation("RAS"),
Flip(1),
Spacing([1.0, 1.2, 0.9], align_corners=False)
])
num_invertible = len(
[i for i in tr.transforms if isinstance(i, InvertibleTransform)])
# forward
img = tr(self.get_image(dtype, device))
self.assertEqual(len(img.applied_operations), num_invertible)
# inverse with Compose
if use_compose:
img = tr.inverse(img)
self.assertEqual(len(img.applied_operations), 0)
# inverse individually
else:
_tr: InvertibleTransform
num_to_inverse = num_invertible
for _tr in tr.transforms[::-1]:
if isinstance(_tr, InvertibleTransform):
img = _tr.inverse(img)
num_to_inverse -= 1
self.assertEqual(len(img.applied_operations),
num_to_inverse)
def test_inverse_inferred_seg(self, extra_transform):
test_data = []
for _ in range(20):
image, label = create_test_image_2d(100, 101)
test_data.append({
"image": image,
"label": label.astype(np.float32)
})
batch_size = 10
# num workers = 0 for mac
num_workers = 2 if sys.platform == "linux" else 0
transforms = Compose([
AddChanneld(KEYS),
SpatialPadd(KEYS, (150, 153)), extra_transform
])
dataset = CacheDataset(test_data, transform=transforms, progress=False)
loader = DataLoader(dataset,
batch_size=batch_size,
shuffle=False,
num_workers=num_workers)
device = "cuda" if torch.cuda.is_available() else "cpu"
model = UNet(spatial_dims=2,
in_channels=1,
out_channels=1,
channels=(2, 4),
strides=(1, )).to(device)
data = first(loader)
self.assertEqual(data["image"].shape[0], batch_size * NUM_SAMPLES)
labels = data["label"].to(device)
self.assertIsInstance(labels, MetaTensor)
segs = model(labels).detach().cpu()
segs_decollated = decollate_batch(segs)
self.assertIsInstance(segs_decollated[0], MetaTensor)
# inverse of individual segmentation
seg_metatensor = first(segs_decollated)
# test to convert interpolation mode for 1 data of model output batch
convert_applied_interp_mode(seg_metatensor.applied_operations,
mode="nearest",
align_corners=None)
# manually invert the last crop samples
xform = seg_metatensor.applied_operations.pop(-1)
shape_before_extra_xform = xform["orig_size"]
resizer = ResizeWithPadOrCrop(spatial_size=shape_before_extra_xform)
with resizer.trace_transform(False):
seg_metatensor = resizer(seg_metatensor)
with allow_missing_keys_mode(transforms):
inv_seg = transforms.inverse({"label": seg_metatensor})["label"]
self.assertEqual(inv_seg.shape[1:], test_data[0]["label"].shape)
def test_inverse_inferred_seg(self):
test_data = []
for _ in range(20):
image, label = create_test_image_2d(100, 101)
test_data.append({
"image": image,
"label": label.astype(np.float32)
})
batch_size = 10
# num workers = 0 for mac
num_workers = 2 if sys.platform != "darwin" else 0
transforms = Compose([
AddChanneld(KEYS),
SpatialPadd(KEYS, (150, 153)),
CenterSpatialCropd(KEYS, (110, 99))
])
num_invertible_transforms = sum(1 for i in transforms.transforms
if isinstance(i, InvertibleTransform))
dataset = CacheDataset(test_data, transform=transforms, progress=False)
loader = DataLoader(dataset,
batch_size=batch_size,
shuffle=False,
num_workers=num_workers)
device = "cuda" if torch.cuda.is_available() else "cpu"
model = UNet(
dimensions=2,
in_channels=1,
out_channels=1,
channels=(2, 4),
strides=(2, ),
).to(device)
data = first(loader)
labels = data["label"].to(device)
segs = model(labels).detach().cpu()
label_transform_key = "label" + InverseKeys.KEY_SUFFIX.value
segs_dict = {
"label": segs,
label_transform_key: data[label_transform_key]
}
segs_dict_decollated = decollate_batch(segs_dict)
# inverse of individual segmentation
seg_dict = first(segs_dict_decollated)
with allow_missing_keys_mode(transforms):
inv_seg = transforms.inverse(seg_dict)["label"]
self.assertEqual(len(data["label_transforms"]),
num_invertible_transforms)
self.assertEqual(len(seg_dict["label_transforms"]),
num_invertible_transforms)
self.assertEqual(inv_seg.shape[1:], test_data[0]["label"].shape)
def test_transforms(self):
key = "im"
_, im = self.get_im()
tr = Compose([ToMetaTensord(key), BorderPadd(key, 1), DivisiblePadd(key, 16), FromMetaTensord(key)])
num_tr = len(tr.transforms)
data = {key: im, PostFix.meta(key): {"affine": torch.eye(4)}}
# apply one at a time
for i, _tr in enumerate(tr.transforms):
data = _tr(data)
is_meta = isinstance(_tr, (ToMetaTensord, BorderPadd, DivisiblePadd))
if is_meta:
self.assertEqual(len(data), 1 if not config.USE_META_DICT else 2) # im, im_transforms, compatibility
self.assertIsInstance(data[key], MetaTensor)
n_applied = len(data[key].applied_operations)
else:
self.assertEqual(len(data), 3) # im, im_meta_dict, im_transforms
self.assertIsInstance(data[key], torch.Tensor)
self.assertNotIsInstance(data[key], MetaTensor)
n_applied = len(data[PostFix.transforms(key)])
self.assertEqual(n_applied, i + 1)
# inverse one at a time
for i, _tr in enumerate(tr.transforms[::-1]):
data = _tr.inverse(data)
is_meta = isinstance(_tr, (FromMetaTensord, BorderPadd, DivisiblePadd))
if is_meta:
self.assertEqual(len(data), 1) # im
self.assertIsInstance(data[key], MetaTensor)
n_applied = len(data[key].applied_operations)
else:
self.assertEqual(len(data), 3) # im, im_meta_dict, im_transforms
self.assertIsInstance(data[key], torch.Tensor)
self.assertNotIsInstance(data[key], MetaTensor)
n_applied = len(data[PostFix.transforms(key)])
self.assertEqual(n_applied, num_tr - i - 1)
# apply all in one go
data = tr({key: im, PostFix.meta(key): {"affine": torch.eye(4)}})
self.assertEqual(len(data), 3) # im, im_meta_dict, im_transforms
self.assertIsInstance(data[key], torch.Tensor)
self.assertNotIsInstance(data[key], MetaTensor)
n_applied = len(data[PostFix.transforms(key)])
self.assertEqual(n_applied, num_tr)
# inverse all in one go
data = tr.inverse(data)
self.assertEqual(len(data), 3) # im, im_meta_dict, im_transforms
self.assertIsInstance(data[key], torch.Tensor)
self.assertNotIsInstance(data[key], MetaTensor)
n_applied = len(data[PostFix.transforms(key)])
self.assertEqual(n_applied, 0)
def test_invert(self):
set_determinism(seed=0)
im_fname = make_nifti_image(create_test_image_3d(101, 100, 107, noise_max=100)[1]) # label image, discrete
data = [im_fname for _ in range(12)]
transform = Compose(
[
LoadImage(image_only=True),
EnsureChannelFirst(),
Orientation("RPS"),
Spacing(pixdim=(1.2, 1.01, 0.9), mode="bilinear", dtype=np.float32),
RandFlip(prob=0.5, spatial_axis=[1, 2]),
RandAxisFlip(prob=0.5),
RandRotate90(prob=0, spatial_axes=(1, 2)),
RandZoom(prob=0.5, min_zoom=0.5, max_zoom=1.1, keep_size=True),
RandRotate(prob=0.5, range_x=np.pi, mode="bilinear", align_corners=True, dtype=np.float64),
RandAffine(prob=0.5, rotate_range=np.pi, mode="nearest"),
ResizeWithPadOrCrop(100),
CastToType(dtype=torch.uint8),
]
)
# num workers = 0 for mac or gpu transforms
num_workers = 0 if sys.platform != "linux" or torch.cuda.is_available() else 2
dataset = Dataset(data, transform=transform)
self.assertIsInstance(transform.inverse(dataset[0]), MetaTensor)
loader = DataLoader(dataset, num_workers=num_workers, batch_size=1)
inverter = Invert(transform=transform, nearest_interp=True, device="cpu")
for d in loader:
d = decollate_batch(d)
for item in d:
orig = deepcopy(item)
i = inverter(item)
self.assertTupleEqual(orig.shape[1:], (100, 100, 100))
# check the nearest interpolation mode
torch.testing.assert_allclose(i.to(torch.uint8).to(torch.float), i.to(torch.float))
self.assertTupleEqual(i.shape[1:], (100, 101, 107))
# check labels match
reverted = i.detach().cpu().numpy().astype(np.int32)
original = LoadImage(image_only=True)(data[-1])
n_good = np.sum(np.isclose(reverted, original.numpy(), atol=1e-3))
reverted_name = i.meta["filename_or_obj"]
original_name = original.meta["filename_or_obj"]
self.assertEqual(reverted_name, original_name)
print("invert diff", reverted.size - n_good)
self.assertTrue((reverted.size - n_good) < 300000, f"diff. {reverted.size - n_good}")
set_determinism(seed=None)
def test_inverse_inferred_seg(self, extra_transform):
test_data = []
for _ in range(20):
image, label = create_test_image_2d(100, 101)
test_data.append({"image": image, "label": label.astype(np.float32)})
batch_size = 10
# num workers = 0 for mac
num_workers = 2 if sys.platform == "linux" else 0
transforms = Compose([AddChanneld(KEYS), SpatialPadd(KEYS, (150, 153)), extra_transform])
num_invertible_transforms = sum(1 for i in transforms.transforms if isinstance(i, InvertibleTransform))
dataset = CacheDataset(test_data, transform=transforms, progress=False)
loader = DataLoader(dataset, batch_size=batch_size, shuffle=False, num_workers=num_workers)
device = "cuda" if torch.cuda.is_available() else "cpu"
model = UNet(spatial_dims=2, in_channels=1, out_channels=1, channels=(2, 4), strides=(2,)).to(device)
data = first(loader)
self.assertEqual(len(data["label_transforms"]), num_invertible_transforms)
self.assertEqual(data["image"].shape[0], batch_size * NUM_SAMPLES)
labels = data["label"].to(device)
segs = model(labels).detach().cpu()
label_transform_key = "label" + InverseKeys.KEY_SUFFIX
segs_dict = {"label": segs, label_transform_key: data[label_transform_key]}
segs_dict_decollated = decollate_batch(segs_dict)
# inverse of individual segmentation
seg_dict = first(segs_dict_decollated)
# test to convert interpolation mode for 1 data of model output batch
convert_inverse_interp_mode(seg_dict, mode="nearest", align_corners=None)
with allow_missing_keys_mode(transforms):
inv_seg = transforms.inverse(seg_dict)["label"]
self.assertEqual(len(data["label_transforms"]), num_invertible_transforms)
self.assertEqual(len(seg_dict["label_transforms"]), num_invertible_transforms)
self.assertEqual(inv_seg.shape[1:], test_data[0]["label"].shape)
# Inverse of batch
batch_inverter = BatchInverseTransform(transforms, loader, collate_fn=no_collation, detach=True)
with allow_missing_keys_mode(transforms):
inv_batch = batch_inverter(segs_dict)
self.assertEqual(inv_batch[0]["label"].shape[1:], test_data[0]["label"].shape)
def test_invert(self):
set_determinism(seed=0)
im_fname, seg_fname = (
make_nifti_image(i)
for i in create_test_image_3d(101, 100, 107, noise_max=100))
transform = Compose([
LoadImaged(KEYS, image_only=True),
EnsureChannelFirstd(KEYS),
Orientationd(KEYS, "RPS"),
Spacingd(KEYS,
pixdim=(1.2, 1.01, 0.9),
mode=["bilinear", "nearest"],
dtype=np.float32),
ScaleIntensityd("image", minv=1, maxv=10),
RandFlipd(KEYS, prob=0.5, spatial_axis=[1, 2]),
RandAxisFlipd(KEYS, prob=0.5),
RandRotate90d(KEYS, prob=0, spatial_axes=(1, 2)),
RandZoomd(KEYS,
prob=0.5,
min_zoom=0.5,
max_zoom=1.1,
keep_size=True),
RandRotated(KEYS,
prob=0.5,
range_x=np.pi,
mode="bilinear",
align_corners=True,
dtype=np.float64),
RandAffined(KEYS, prob=0.5, rotate_range=np.pi, mode="nearest"),
ResizeWithPadOrCropd(KEYS, 100),
CastToTyped(KEYS, dtype=[torch.uint8, np.uint8]),
CopyItemsd("label",
times=2,
names=["label_inverted", "label_inverted1"]),
CopyItemsd("image",
times=2,
names=["image_inverted", "image_inverted1"]),
])
data = [{"image": im_fname, "label": seg_fname} for _ in range(12)]
# num workers = 0 for mac or gpu transforms
num_workers = 0 if sys.platform != "linux" or torch.cuda.is_available(
) else 2
dataset = Dataset(data, transform=transform)
transform.inverse(dataset[0])
loader = DataLoader(dataset, num_workers=num_workers, batch_size=1)
inverter = Invertd(
# `image` was not copied, invert the original value directly
keys=["image_inverted", "label_inverted"],
transform=transform,
orig_keys=["label", "label"],
nearest_interp=True,
device="cpu",
)
inverter_1 = Invertd(
# `image` was not copied, invert the original value directly
keys=["image_inverted1", "label_inverted1"],
transform=transform,
orig_keys=["image", "image"],
nearest_interp=[True, False],
device="cpu",
)
expected_keys = [
"image", "image_inverted", "image_inverted1", "label",
"label_inverted", "label_inverted1"
]
# execute 1 epoch
for d in loader:
d = decollate_batch(d)
for item in d:
item = inverter(item)
item = inverter_1(item)
self.assertListEqual(sorted(item), expected_keys)
self.assertTupleEqual(item["image"].shape[1:], (100, 100, 100))
self.assertTupleEqual(item["label"].shape[1:], (100, 100, 100))
# check the nearest interpolation mode
i = item["image_inverted"]
torch.testing.assert_allclose(
i.to(torch.uint8).to(torch.float), i.to(torch.float))
self.assertTupleEqual(i.shape[1:], (100, 101, 107))
i = item["label_inverted"]
torch.testing.assert_allclose(
i.to(torch.uint8).to(torch.float), i.to(torch.float))
self.assertTupleEqual(i.shape[1:], (100, 101, 107))
# check the case that different items use different interpolation mode to invert transforms
d = item["image_inverted1"]
# if the interpolation mode is nearest, accumulated diff should be smaller than 1
self.assertLess(
torch.sum(
d.to(torch.float) -
d.to(torch.uint8).to(torch.float)).item(), 1.0)
self.assertTupleEqual(d.shape, (1, 100, 101, 107))
d = item["label_inverted1"]
# if the interpolation mode is not nearest, accumulated diff should be greater than 10000
self.assertGreater(
torch.sum(
d.to(torch.float) -
d.to(torch.uint8).to(torch.float)).item(), 10000.0)
self.assertTupleEqual(d.shape, (1, 100, 101, 107))
# check labels match
reverted = item["label_inverted"].detach().cpu().numpy().astype(
np.int32)
original = LoadImaged(KEYS, image_only=True)(data[-1])["label"]
n_good = np.sum(np.isclose(reverted, original, atol=1e-3))
reverted_name = item["label_inverted"].meta["filename_or_obj"]
original_name = data[-1]["label"]
self.assertEqual(reverted_name, original_name)
print("invert diff", reverted.size - n_good)
# 25300: 2 workers (cpu, non-macos)
# 1812: 0 workers (gpu or macos)
# 1821: windows torch 1.10.0
self.assertTrue((reverted.size - n_good) < 40000,
f"diff. {reverted.size - n_good}")
set_determinism(seed=None)