def get_data(keys):
"""Get the example data to be used.
Use MarsAtlas as it only contains 1 image for quick download and
that image is parcellated.
"""
cache_dir = os.environ.get("MONAI_DATA_DIRECTORY") or tempfile.mkdtemp()
fname = "MarsAtlas-MNI-Colin27.zip"
url = "https://www.dropbox.com/s/ndz8qtqblkciole/" + fname + "?dl=1"
out_path = os.path.join(cache_dir, "MarsAtlas-MNI-Colin27")
zip_path = os.path.join(cache_dir, fname)
download_and_extract(url, zip_path, out_path)
image, label = sorted(glob(os.path.join(out_path, "*.nii")))
data = {CommonKeys.IMAGE: image, CommonKeys.LABEL: label}
transforms = Compose([
LoadImaged(keys),
AddChanneld(keys),
ScaleIntensityd(CommonKeys.IMAGE),
Rotate90d(keys, spatial_axes=[0, 2])
])
data = transforms(data)
max_size = max(data[keys[0]].shape)
padder = SpatialPadd(keys, (max_size, max_size, max_size))
return padder(data)
def run_inference_test(root_dir, device=torch.device("cuda:0")):
images = sorted(glob(os.path.join(root_dir, "im*.nii.gz")))
segs = sorted(glob(os.path.join(root_dir, "seg*.nii.gz")))
val_files = [{"img": img, "seg": seg} for img, seg in zip(images, segs)]
# define transforms for image and segmentation
val_transforms = Compose([
LoadNiftid(keys=["img", "seg"]),
AsChannelFirstd(keys=["img", "seg"], channel_dim=-1),
# resampling with align_corners=True or dtype=float64 will generate
# slight different results between PyTorch 1.5 an 1.6
Spacingd(keys=["img", "seg"],
pixdim=[1.2, 0.8, 0.7],
mode=["bilinear", "nearest"],
dtype=np.float32),
ScaleIntensityd(keys=["img", "seg"]),
ToTensord(keys=["img", "seg"]),
])
val_ds = monai.data.Dataset(data=val_files, transform=val_transforms)
# sliding window inferene need to input 1 image in every iteration
val_loader = monai.data.DataLoader(val_ds, batch_size=1, num_workers=4)
dice_metric = DiceMetric(include_background=True,
to_onehot_y=False,
sigmoid=True,
reduction="mean")
model = UNet(
dimensions=3,
in_channels=1,
out_channels=1,
channels=(16, 32, 64, 128, 256),
strides=(2, 2, 2, 2),
num_res_units=2,
).to(device)
model_filename = os.path.join(root_dir, "best_metric_model.pth")
model.load_state_dict(torch.load(model_filename))
model.eval()
with torch.no_grad():
metric_sum = 0.0
metric_count = 0
# resampling with align_corners=True or dtype=float64 will generate
# slight different results between PyTorch 1.5 an 1.6
saver = NiftiSaver(output_dir=os.path.join(root_dir, "output"),
dtype=np.float32)
for val_data in val_loader:
val_images, val_labels = val_data["img"].to(
device), val_data["seg"].to(device)
# define sliding window size and batch size for windows inference
sw_batch_size, roi_size = 4, (96, 96, 96)
val_outputs = sliding_window_inference(val_images, roi_size,
sw_batch_size, model)
value = dice_metric(y_pred=val_outputs, y=val_labels)
not_nans = dice_metric.not_nans.item()
metric_count += not_nans
metric_sum += value.item() * not_nans
val_outputs = (val_outputs.sigmoid() >= 0.5).float()
saver.save_batch(val_outputs, val_data["img_meta_dict"])
metric = metric_sum / metric_count
return metric
def test_samples(self):
testing_dir = os.path.join(os.path.dirname(os.path.realpath(__file__)), "testing_data")
keys = "image"
xforms = Compose(
[
LoadImaged(keys=keys),
AddChanneld(keys=keys),
ScaleIntensityd(keys=keys),
RandSpatialCropSamplesd(keys=keys, roi_size=(8, 8, 5), random_size=True, num_samples=10),
]
)
image_path = os.path.join(testing_dir, "anatomical.nii")
xforms.set_random_state(0)
ims = xforms({keys: image_path})
fig, mat = matshow3d(
[im[keys] for im in ims], title=f"testing {keys}", figsize=(2, 2), frames_per_row=5, every_n=2, show=False
)
self.assertTrue(mat.dtype == np.float32)
with tempfile.TemporaryDirectory() as tempdir:
tempimg = f"{tempdir}/matshow3d_patch_test.png"
fig.savefig(tempimg)
comp = compare_images(f"{testing_dir}/matshow3d_patch_test.png", tempimg, 5e-2, in_decorator=True)
if comp:
print("not none comp: ", comp) # matplotlib 3.2.2
np.testing.assert_allclose(comp["rms"], 30.786983, atol=1e-3, rtol=1e-3)
else:
self.assertIsNone(comp, f"value of comp={comp}") # None indicates test passed
def test_factor_scale(self):
key = "img"
for p in TEST_NDARRAYS:
scaler = ScaleIntensityd(keys=[key], minv=None, maxv=None, factor=0.1)
result = scaler({key: p(self.imt)})
expected = (self.imt * (1 + 0.1)).astype(np.float32)
assert_allclose(result[key], p(expected), type_test="tensor")
def _get_loader(self, folders):
images = []
segs = []
for folder in folders:
images += glob(os.path.join(folder, "*_im.nii.gz"))
segs += glob(os.path.join(folder, "*_seg.nii.gz"))
images = sorted(images, key=os.path.basename)
segs = sorted(segs, key=os.path.basename)
files = [{"img": img, "seg": seg} for img, seg in zip(images, segs)]
transforms = Compose([
LoadImaged(keys=["img", "seg"]),
AsChannelFirstd(keys=["img", "seg"], channel_dim=-1),
ScaleIntensityd(keys="img"),
ToTensord(keys=["img", "seg"]),
])
ds = CacheDataset(data=files, transform=transforms)
loader = DataLoader(ds,
batch_size=1,
num_workers=4,
collate_fn=list_data_collate)
return loader
def test_3d_rgb(self):
testing_dir = os.path.join(os.path.dirname(os.path.realpath(__file__)), "testing_data")
keys = "image"
xforms = Compose(
[
LoadImaged(keys=keys),
AddChanneld(keys=keys),
ScaleIntensityd(keys=keys),
# change to RGB color image
RepeatChanneld(keys=keys, repeats=3),
]
)
image_path = os.path.join(testing_dir, "anatomical.nii")
ims = xforms({keys: image_path})
fig = pyplot.figure() # external figure
fig, _ = matshow3d(
volume=ims[keys],
fig=fig,
figsize=(2, 2),
frames_per_row=5,
every_n=2,
frame_dim=-1,
channel_dim=0,
show=False,
)
with tempfile.TemporaryDirectory() as tempdir:
tempimg = f"{tempdir}/matshow3d_rgb_test.png"
fig.savefig(tempimg)
comp = compare_images(f"{testing_dir}/matshow3d_rgb_test.png", tempimg, 5e-2)
self.assertIsNone(comp, f"value of comp={comp}") # None indicates test passed
def run_inference_test(root_dir, device=torch.device("cuda:0")):
images = sorted(glob(os.path.join(root_dir, "im*.nii.gz")))
segs = sorted(glob(os.path.join(root_dir, "seg*.nii.gz")))
val_files = [{"img": img, "seg": seg} for img, seg in zip(images, segs)]
# define transforms for image and segmentation
val_transforms = Compose([
LoadNiftid(keys=["img", "seg"]),
AsChannelFirstd(keys=["img", "seg"], channel_dim=-1),
ScaleIntensityd(keys=["img", "seg"]),
ToTensord(keys=["img", "seg"]),
])
val_ds = monai.data.Dataset(data=val_files, transform=val_transforms)
# sliding window inferene need to input 1 image in every iteration
val_loader = DataLoader(val_ds,
batch_size=1,
num_workers=4,
collate_fn=list_data_collate,
pin_memory=torch.cuda.is_available())
model = UNet(
dimensions=3,
in_channels=1,
out_channels=1,
channels=(16, 32, 64, 128, 256),
strides=(2, 2, 2, 2),
num_res_units=2,
).to(device)
model_filename = os.path.join(root_dir, "best_metric_model.pth")
model.load_state_dict(torch.load(model_filename))
model.eval()
with torch.no_grad():
metric_sum = 0.0
metric_count = 0
saver = NiftiSaver(output_dir=os.path.join(root_dir, "output"),
dtype=int)
for val_data in val_loader:
val_images, val_labels = val_data["img"].to(
device), val_data["seg"].to(device)
# define sliding window size and batch size for windows inference
sw_batch_size, roi_size = 4, (96, 96, 96)
val_outputs = sliding_window_inference(val_images, roi_size,
sw_batch_size, model)
value = compute_meandice(y_pred=val_outputs,
y=val_labels,
include_background=True,
to_onehot_y=False,
add_sigmoid=True)
metric_count += len(value)
metric_sum += value.sum().item()
val_outputs = (val_outputs.sigmoid() >= 0.5).float()
saver.save_batch(
val_outputs, {
"filename_or_obj": val_data["img.filename_or_obj"],
"affine": val_data["img.affine"]
})
metric = metric_sum / metric_count
return metric
def run_inference_test(root_dir, device="cuda:0"):
images = sorted(glob(os.path.join(root_dir, "im*.nii.gz")))
segs = sorted(glob(os.path.join(root_dir, "seg*.nii.gz")))
val_files = [{"img": img, "seg": seg} for img, seg in zip(images, segs)]
# define transforms for image and segmentation
val_transforms = Compose(
[
LoadImaged(keys=["img", "seg"]),
EnsureChannelFirstd(keys=["img", "seg"]),
# resampling with align_corners=True or dtype=float64 will generate
# slight different results between PyTorch 1.5 an 1.6
Spacingd(keys=["img", "seg"], pixdim=[1.2, 0.8, 0.7], mode=["bilinear", "nearest"], dtype=np.float32),
ScaleIntensityd(keys="img"),
ToTensord(keys=["img", "seg"]),
]
)
val_ds = monai.data.Dataset(data=val_files, transform=val_transforms)
# sliding window inference need to input 1 image in every iteration
val_loader = monai.data.DataLoader(val_ds, batch_size=1, num_workers=4)
val_post_tran = Compose([ToTensor(), Activations(sigmoid=True), AsDiscrete(threshold=0.5)])
dice_metric = DiceMetric(include_background=True, reduction="mean", get_not_nans=False)
model = UNet(
spatial_dims=3,
in_channels=1,
out_channels=1,
channels=(16, 32, 64, 128, 256),
strides=(2, 2, 2, 2),
num_res_units=2,
).to(device)
model_filename = os.path.join(root_dir, "best_metric_model.pth")
model.load_state_dict(torch.load(model_filename))
with eval_mode(model):
# resampling with align_corners=True or dtype=float64 will generate
# slight different results between PyTorch 1.5 an 1.6
saver = SaveImage(
output_dir=os.path.join(root_dir, "output"),
dtype=np.float32,
output_ext=".nii.gz",
output_postfix="seg",
mode="bilinear",
)
for val_data in val_loader:
val_images, val_labels = val_data["img"].to(device), val_data["seg"].to(device)
# define sliding window size and batch size for windows inference
sw_batch_size, roi_size = 4, (96, 96, 96)
val_outputs = sliding_window_inference(val_images, roi_size, sw_batch_size, model)
# decollate prediction into a list
val_outputs = [val_post_tran(i) for i in decollate_batch(val_outputs)]
val_meta = decollate_batch(val_data[PostFix.meta("img")])
# compute metrics
dice_metric(y_pred=val_outputs, y=val_labels)
for img, meta in zip(val_outputs, val_meta): # save a decollated batch of files
saver(img, meta)
return dice_metric.aggregate().item()
def main(tempdir):
monai.config.print_config()
logging.basicConfig(stream=sys.stdout, level=logging.INFO)
print(f"generating synthetic data to {tempdir} (this may take a while)")
for i in range(5):
im, seg = create_test_image_2d(128, 128, num_seg_classes=1)
Image.fromarray(im.astype("uint8")).save(os.path.join(tempdir, f"img{i:d}.png"))
Image.fromarray(seg.astype("uint8")).save(os.path.join(tempdir, f"seg{i:d}.png"))
images = sorted(glob(os.path.join(tempdir, "img*.png")))
segs = sorted(glob(os.path.join(tempdir, "seg*.png")))
val_files = [{"img": img, "seg": seg} for img, seg in zip(images, segs)]
# define transforms for image and segmentation
val_transforms = Compose(
[
LoadImaged(keys=["img", "seg"]),
AddChanneld(keys=["img", "seg"]),
ScaleIntensityd(keys="img"),
ToTensord(keys=["img", "seg"]),
]
)
val_ds = monai.data.Dataset(data=val_files, transform=val_transforms)
# sliding window inference need to input 1 image in every iteration
val_loader = DataLoader(val_ds, batch_size=1, num_workers=4, collate_fn=list_data_collate)
dice_metric = DiceMetric(include_background=True, to_onehot_y=False, sigmoid=True, reduction="mean")
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model = UNet(
dimensions=2,
in_channels=1,
out_channels=1,
channels=(16, 32, 64, 128, 256),
strides=(2, 2, 2, 2),
num_res_units=2,
).to(device)
model.load_state_dict(torch.load("best_metric_model_segmentation2d_dict.pth"))
model.eval()
with torch.no_grad():
metric_sum = 0.0
metric_count = 0
saver = PNGSaver(output_dir="./output")
for val_data in val_loader:
val_images, val_labels = val_data["img"].to(device), val_data["seg"].to(device)
# define sliding window size and batch size for windows inference
roi_size = (96, 96)
sw_batch_size = 4
val_outputs = sliding_window_inference(val_images, roi_size, sw_batch_size, model)
value = dice_metric(y_pred=val_outputs, y=val_labels)
metric_count += len(value)
metric_sum += value.item() * len(value)
val_outputs = val_outputs.sigmoid() >= 0.5
saver.save_batch(val_outputs)
metric = metric_sum / metric_count
print("evaluation metric:", metric)
def main():
monai.config.print_config()
logging.basicConfig(stream=sys.stdout, level=logging.INFO)
# IXI dataset as a demo, downloadable from https://brain-development.org/ixi-dataset/
images = [
os.sep.join(["workspace", "data", "medical", "ixi", "IXI-T1", "IXI607-Guys-1097-T1.nii.gz"]),
os.sep.join(["workspace", "data", "medical", "ixi", "IXI-T1", "IXI175-HH-1570-T1.nii.gz"]),
os.sep.join(["workspace", "data", "medical", "ixi", "IXI-T1", "IXI385-HH-2078-T1.nii.gz"]),
os.sep.join(["workspace", "data", "medical", "ixi", "IXI-T1", "IXI344-Guys-0905-T1.nii.gz"]),
os.sep.join(["workspace", "data", "medical", "ixi", "IXI-T1", "IXI409-Guys-0960-T1.nii.gz"]),
os.sep.join(["workspace", "data", "medical", "ixi", "IXI-T1", "IXI584-Guys-1129-T1.nii.gz"]),
os.sep.join(["workspace", "data", "medical", "ixi", "IXI-T1", "IXI253-HH-1694-T1.nii.gz"]),
os.sep.join(["workspace", "data", "medical", "ixi", "IXI-T1", "IXI092-HH-1436-T1.nii.gz"]),
os.sep.join(["workspace", "data", "medical", "ixi", "IXI-T1", "IXI574-IOP-1156-T1.nii.gz"]),
os.sep.join(["workspace", "data", "medical", "ixi", "IXI-T1", "IXI585-Guys-1130-T1.nii.gz"]),
]
# 2 binary labels for gender classification: man and woman
labels = np.array([0, 0, 1, 0, 1, 0, 1, 0, 1, 0], dtype=np.int64)
val_files = [{"img": img, "label": label} for img, label in zip(images, labels)]
# Define transforms for image
val_transforms = Compose(
[
LoadNiftid(keys=["img"]),
AddChanneld(keys=["img"]),
ScaleIntensityd(keys=["img"]),
Resized(keys=["img"], spatial_size=(96, 96, 96)),
ToTensord(keys=["img"]),
]
)
# create a validation data loader
val_ds = monai.data.Dataset(data=val_files, transform=val_transforms)
val_loader = DataLoader(val_ds, batch_size=2, num_workers=4, pin_memory=torch.cuda.is_available())
# Create DenseNet121
device = torch.device("cuda:0")
model = monai.networks.nets.densenet.densenet121(spatial_dims=3, in_channels=1, out_channels=2).to(device)
model.load_state_dict(torch.load("best_metric_model.pth"))
model.eval()
with torch.no_grad():
num_correct = 0.0
metric_count = 0
saver = CSVSaver(output_dir="./output")
for val_data in val_loader:
val_images, val_labels = val_data["img"].to(device), val_data["label"].to(device)
val_outputs = model(val_images).argmax(dim=1)
value = torch.eq(val_outputs, val_labels)
metric_count += len(value)
num_correct += value.sum().item()
saver.save_batch(val_outputs, val_data["img_meta_dict"])
metric = num_correct / metric_count
print("evaluation metric:", metric)
saver.finalize()
def test_values(self):
testing_dir = os.path.join(os.path.dirname(os.path.realpath(__file__)),
"testing_data")
transform = Compose([
LoadImaged(keys="image"),
AddChanneld(keys="image"),
ScaleIntensityd(keys="image"),
ToTensord(keys=["image", "label"]),
])
def _test_dataset(dataset):
self.assertEqual(
len(dataset),
int(MEDNIST_FULL_DATASET_LENGTH * dataset.test_frac))
self.assertTrue("image" in dataset[0])
self.assertTrue("label" in dataset[0])
self.assertTrue(PostFix.meta("image") in dataset[0])
self.assertTupleEqual(dataset[0]["image"].shape, (1, 64, 64))
with skip_if_downloading_fails():
data = MedNISTDataset(root_dir=testing_dir,
transform=transform,
section="test",
download=True,
copy_cache=False)
_test_dataset(data)
# testing from
data = MedNISTDataset(root_dir=Path(testing_dir),
transform=transform,
section="test",
download=False)
self.assertEqual(data.get_num_classes(), 6)
_test_dataset(data)
data = MedNISTDataset(root_dir=testing_dir,
section="test",
download=False)
self.assertTupleEqual(data[0]["image"].shape, (64, 64))
# test same dataset length with different random seed
data = MedNISTDataset(root_dir=testing_dir,
transform=transform,
section="test",
download=False,
seed=42)
_test_dataset(data)
self.assertEqual(data[0]["class_name"], "AbdomenCT")
self.assertEqual(data[0]["label"].cpu().item(), 0)
shutil.rmtree(os.path.join(testing_dir, "MedNIST"))
try:
MedNISTDataset(root_dir=testing_dir,
transform=transform,
section="test",
download=False)
except RuntimeError as e:
print(str(e))
self.assertTrue(str(e).startswith("Cannot find dataset directory"))
def test_range_scale(self):
key = 'img'
scaler = ScaleIntensityd(keys=[key], minv=1.0, maxv=2.0)
result = scaler({key: self.imt})
mina = np.min(self.imt)
maxa = np.max(self.imt)
norm = (self.imt - mina) / (maxa - mina)
expected = (norm * (2.0 - 1.0)) + 1.0
np.testing.assert_allclose(result[key], expected)
def test_values(self):
testing_dir = os.path.join(os.path.dirname(os.path.realpath(__file__)),
"testing_data")
transform = Compose([
LoadNiftid(keys=["image", "label"]),
AddChanneld(keys=["image", "label"]),
ScaleIntensityd(keys="image"),
ToTensord(keys=["image", "label"]),
])
def _test_dataset(dataset):
self.assertEqual(len(dataset), 52)
self.assertTrue("image" in dataset[0])
self.assertTrue("label" in dataset[0])
self.assertTrue("image_meta_dict" in dataset[0])
self.assertTupleEqual(dataset[0]["image"].shape, (1, 33, 47, 34))
try: # will start downloading if testing_dir doesn't have the Decathlon files
data = DecathlonDataset(
root_dir=testing_dir,
task="Task04_Hippocampus",
transform=transform,
section="validation",
download=True,
)
except (ContentTooShortError, HTTPError, RuntimeError) as e:
print(str(e))
if isinstance(e, RuntimeError):
# FIXME: skip MD5 check as current downloading method may fail
self.assertTrue(str(e).startswith("MD5 check"))
return # skipping this test due the network connection errors
_test_dataset(data)
data = DecathlonDataset(root_dir=testing_dir,
task="Task04_Hippocampus",
transform=transform,
section="validation",
download=False)
_test_dataset(data)
data = DecathlonDataset(root_dir=testing_dir,
task="Task04_Hippocampus",
section="validation",
download=False)
self.assertTupleEqual(data[0]["image"].shape, (33, 47, 34))
shutil.rmtree(os.path.join(testing_dir, "Task04_Hippocampus"))
try:
data = DecathlonDataset(
root_dir=testing_dir,
task="Task04_Hippocampus",
transform=transform,
section="validation",
download=False,
)
except RuntimeError as e:
print(str(e))
self.assertTrue(str(e).startswith("Cannot find dataset directory"))
def test_range_scale(self):
key = "img"
for p in TEST_NDARRAYS:
scaler = ScaleIntensityd(keys=[key], minv=1.0, maxv=2.0)
result = scaler({key: p(self.imt)})
mina = np.min(self.imt)
maxa = np.max(self.imt)
norm = (self.imt - mina) / (maxa - mina)
expected = (norm * (2.0 - 1.0)) + 1.0
assert_allclose(result[key], p(expected), type_test="tensor")
def get_image_transforms():
itk_reader = monai.data.ITKReader()
# Define transforms for image
image_transforms = Compose([
LoadImaged(keys=['img'], reader=itk_reader),
EnsureChannelFirstd(keys=['img']),
ScaleIntensityd(keys=['img']),
ToTensord(keys=['img']),
])
return image_transforms
def test_values(self):
tempdir = tempfile.mkdtemp()
transform = Compose([
LoadNiftid(keys=["image", "label"]),
AddChanneld(keys=["image", "label"]),
ScaleIntensityd(keys="image"),
ToTensord(keys=["image", "label"]),
])
def _test_dataset(dataset):
self.assertEqual(len(dataset), 52)
self.assertTrue("image" in dataset[0])
self.assertTrue("label" in dataset[0])
self.assertTrue("image_meta_dict" in dataset[0])
self.assertTupleEqual(dataset[0]["image"].shape, (1, 33, 47, 34))
try:
data = DecathlonDataset(root_dir=tempdir,
task="Task04_Hippocampus",
transform=transform,
section="validation",
download=True)
except RuntimeError as e:
if str(e).startswith(
"download failed due to network issue or permission denied."
):
shutil.rmtree(tempdir)
return
_test_dataset(data)
data = DecathlonDataset(root_dir=tempdir,
task="Task04_Hippocampus",
transform=transform,
section="validation",
download=False)
_test_dataset(data)
data = DecathlonDataset(root_dir=tempdir,
task="Task04_Hippocampus",
section="validation",
download=False)
self.assertTupleEqual(data[0]["image"].shape, (33, 47, 34))
shutil.rmtree(os.path.join(tempdir, "Task04_Hippocampus"))
try:
data = DecathlonDataset(root_dir=tempdir,
task="Task04_Hippocampus",
transform=transform,
section="validation",
download=False)
except RuntimeError as e:
print(str(e))
self.assertTrue(
str(e).startswith("can not find dataset directory"))
shutil.rmtree(tempdir)
def test_values(self):
testing_dir = os.path.join(os.path.dirname(os.path.realpath(__file__)),
"testing_data")
train_transform = Compose([
LoadImaged(keys=["image", "label"]),
AddChanneld(keys=["image", "label"]),
ScaleIntensityd(keys="image"),
ToTensord(keys=["image", "label"]),
])
val_transform = LoadImaged(keys=["image", "label"])
def _test_dataset(dataset):
self.assertEqual(len(dataset), 52)
self.assertTrue("image" in dataset[0])
self.assertTrue("label" in dataset[0])
self.assertTrue("image_meta_dict" in dataset[0])
self.assertTupleEqual(dataset[0]["image"].shape, (1, 34, 49, 41))
cvdataset = CrossValidation(
dataset_cls=DecathlonDataset,
nfolds=5,
seed=12345,
root_dir=testing_dir,
task="Task04_Hippocampus",
section="validation",
transform=train_transform,
download=True,
)
try: # will start downloading if testing_dir doesn't have the Decathlon files
data = cvdataset.get_dataset(folds=0)
except (ContentTooShortError, HTTPError, RuntimeError) as e:
print(str(e))
if isinstance(e, RuntimeError):
# FIXME: skip MD5 check as current downloading method may fail
self.assertTrue(str(e).startswith("md5 check"))
return # skipping this test due the network connection errors
_test_dataset(data)
# test training data for fold [1, 2, 3, 4] of 5 splits
data = cvdataset.get_dataset(folds=[1, 2, 3, 4])
self.assertTupleEqual(data[0]["image"].shape, (1, 35, 52, 33))
self.assertEqual(len(data), 208)
# test train / validation for fold 4 of 5 splits
data = cvdataset.get_dataset(folds=[4],
transform=val_transform,
download=False)
# val_transform doesn't add the channel dim to shape
self.assertTupleEqual(data[0]["image"].shape, (38, 53, 30))
self.assertEqual(len(data), 52)
data = cvdataset.get_dataset(folds=[0, 1, 2, 3])
self.assertTupleEqual(data[0]["image"].shape, (1, 34, 49, 41))
self.assertEqual(len(data), 208)
def get_test_transforms(end_image_shape):
test_transforms = Compose([
LoadNiftid(keys=["image"]),
AddChanneld(keys=["image"]),
Orientationd(keys=["image"], axcodes="RAS"),
Winsorized(keys=["image"]),
NormalizeIntensityd(keys=["image"]),
ScaleIntensityd(keys=["image"]),
SpatialPadd(keys=["image"], spatial_size=end_image_shape),
ToTensord(keys=["image"]),
])
return test_transforms
def test_channel_wise(self):
key = "img"
for p in TEST_NDARRAYS:
scaler = ScaleIntensityd(keys=[key], minv=1.0, maxv=2.0, channel_wise=True)
data = p(self.imt)
result = scaler({key: data})
mina = self.imt.min()
maxa = self.imt.max()
for i, c in enumerate(data):
norm = (c - mina) / (maxa - mina)
expected = p((norm * (2.0 - 1.0)) + 1.0)
assert_allclose(result[key][i], expected, type_test="tensor", rtol=1e-7, atol=0)
def setUp(self):
self.root_dir = os.environ.get("MONAI_DATA_DIRECTORY")
if not self.root_dir:
self.root_dir = os.path.join(
os.path.dirname(os.path.realpath(__file__)), "testing_data")
self.transforms = Compose([
LoadImaged(keys="image"),
AddChanneld(keys="image"),
ScaleIntensityd(keys="image"),
ToTensord(keys="image"),
])
def test_values(self):
testing_dir = os.path.join(os.path.dirname(os.path.realpath(__file__)),
"testing_data")
transform = Compose([
LoadImaged(keys="image"),
AddChanneld(keys="image"),
ScaleIntensityd(keys="image"),
ToTensord(keys=["image", "label"]),
])
def _test_dataset(dataset):
self.assertEqual(len(dataset), 5986)
self.assertTrue("image" in dataset[0])
self.assertTrue("label" in dataset[0])
self.assertTrue("image_meta_dict" in dataset[0])
self.assertTupleEqual(dataset[0]["image"].shape, (1, 64, 64))
try: # will start downloading if testing_dir doesn't have the MedNIST files
data = MedNISTDataset(root_dir=testing_dir,
transform=transform,
section="test",
download=True)
except (ContentTooShortError, HTTPError, RuntimeError) as e:
print(str(e))
if isinstance(e, RuntimeError):
# FIXME: skip MD5 check as current downloading method may fail
self.assertTrue(str(e).startswith("md5 check"))
return # skipping this test due the network connection errors
_test_dataset(data)
# testing from
data = MedNISTDataset(root_dir=testing_dir,
transform=transform,
section="test",
download=False)
data.get_num_classes()
_test_dataset(data)
data = MedNISTDataset(root_dir=testing_dir,
section="test",
download=False)
self.assertTupleEqual(data[0]["image"].shape, (64, 64))
shutil.rmtree(os.path.join(testing_dir, "MedNIST"))
try:
data = MedNISTDataset(root_dir=testing_dir,
transform=transform,
section="test",
download=False)
except RuntimeError as e:
print(str(e))
self.assertTrue(str(e).startswith("Cannot find dataset directory"))
def test_values(self):
testing_dir = os.path.join(os.path.dirname(os.path.realpath(__file__)),
"testing_data")
train_transform = Compose([
LoadImaged(keys=["image", "label"]),
AddChanneld(keys=["image", "label"]),
ScaleIntensityd(keys="image"),
ToTensord(keys=["image", "label"]),
])
val_transform = LoadImaged(keys=["image", "label"])
def _test_dataset(dataset):
self.assertEqual(len(dataset), 52)
self.assertTrue("image" in dataset[0])
self.assertTrue("label" in dataset[0])
self.assertTrue(PostFix.meta("image") in dataset[0])
self.assertTupleEqual(dataset[0]["image"].shape, (1, 34, 49, 41))
cvdataset = CrossValidation(
dataset_cls=DecathlonDataset,
nfolds=5,
seed=12345,
root_dir=testing_dir,
task="Task04_Hippocampus",
section="validation",
transform=train_transform,
download=True,
)
with skip_if_downloading_fails():
data = cvdataset.get_dataset(folds=0)
_test_dataset(data)
# test training data for fold [1, 2, 3, 4] of 5 splits
data = cvdataset.get_dataset(folds=[1, 2, 3, 4])
self.assertTupleEqual(data[0]["image"].shape, (1, 35, 52, 33))
self.assertEqual(len(data), 208)
# test train / validation for fold 4 of 5 splits
data = cvdataset.get_dataset(folds=[4],
transform=val_transform,
download=False)
# val_transform doesn't add the channel dim to shape
self.assertTupleEqual(data[0]["image"].shape, (38, 53, 30))
self.assertEqual(len(data), 52)
data = cvdataset.get_dataset(folds=[0, 1, 2, 3])
self.assertTupleEqual(data[0]["image"].shape, (1, 34, 49, 41))
self.assertEqual(len(data), 208)
def test_values(self):
tempdir = tempfile.mkdtemp()
transform = Compose([
LoadPNGd(keys="image"),
AddChanneld(keys="image"),
ScaleIntensityd(keys="image"),
ToTensord(keys=["image", "label"]),
])
def _test_dataset(dataset):
self.assertEqual(len(dataset), 5986)
self.assertTrue("image" in dataset[0])
self.assertTrue("label" in dataset[0])
self.assertTrue("image_meta_dict" in dataset[0])
self.assertTupleEqual(dataset[0]["image"].shape, (1, 64, 64))
data = MedNISTDataset(root_dir=tempdir,
transform=transform,
section="test",
download=True)
_test_dataset(data)
data = MedNISTDataset(root_dir=tempdir,
transform=transform,
section="test",
download=False)
_test_dataset(data)
data = MedNISTDataset(root_dir=tempdir, section="test", download=False)
self.assertTupleEqual(data[0]["image"].shape, (64, 64))
shutil.rmtree(os.path.join(tempdir, "MedNIST"))
try:
data = MedNISTDataset(root_dir=tempdir,
transform=transform,
section="test",
download=False)
except RuntimeError as e:
print(str(e))
self.assertTrue(
str(e).startswith("can not find dataset directory"))
shutil.rmtree(tempdir)
def _get_loader(self, folders):
images = []
segs = []
for folder in folders:
images += glob(os.path.join(folder, "*_im.nii.gz"))
segs += glob(os.path.join(folder, "*_seg.nii.gz"))
images = sorted(images, key=os.path.basename)
segs = sorted(segs, key=os.path.basename)
files = [{"img": img, "seg": seg} for img, seg in zip(images, segs)]
transforms = Compose([
LoadImaged(keys=["img", "seg"]),
AsChannelFirstd(keys=["img", "seg"], channel_dim=-1),
ScaleIntensityd(keys="img"),
RandCropByPosNegLabeld(keys=["img", "seg"],
label_key="seg",
spatial_size=[96, 96, 96],
pos=1,
neg=1,
num_samples=4),
RandRotate90d(keys=["img", "seg"], prob=0.5, spatial_axes=[0, 2]),
ToTensord(keys=["img", "seg"]),
])
ds = CacheDataset(data=files, transform=transforms)
loader = DataLoader(
ds,
batch_size=2,
shuffle=True,
num_workers=4,
collate_fn=list_data_collate,
pin_memory=torch.cuda.is_available(),
)
return loader
def main(tempdir):
monai.config.print_config()
logging.basicConfig(stream=sys.stdout, level=logging.INFO)
# create a temporary directory and 40 random image, mask pairs
print(f"generating synthetic data to {tempdir} (this may take a while)")
for i in range(5):
im, seg = create_test_image_3d(128, 128, 128, num_seg_classes=1, channel_dim=-1)
n = nib.Nifti1Image(im, np.eye(4))
nib.save(n, os.path.join(tempdir, f"im{i:d}.nii.gz"))
n = nib.Nifti1Image(seg, np.eye(4))
nib.save(n, os.path.join(tempdir, f"seg{i:d}.nii.gz"))
images = sorted(glob(os.path.join(tempdir, "im*.nii.gz")))
segs = sorted(glob(os.path.join(tempdir, "seg*.nii.gz")))
val_files = [{"image": img, "label": seg} for img, seg in zip(images, segs)]
# model file path
model_file = glob("./runs/net_key_metric*")[0]
# define transforms for image and segmentation
val_transforms = Compose(
[
LoadNiftid(keys=["image", "label"]),
AsChannelFirstd(keys=["image", "label"], channel_dim=-1),
ScaleIntensityd(keys="image"),
ToTensord(keys=["image", "label"]),
]
)
# create a validation data loader
val_ds = monai.data.Dataset(data=val_files, transform=val_transforms)
val_loader = monai.data.DataLoader(val_ds, batch_size=1, num_workers=4)
# create UNet, DiceLoss and Adam optimizer
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
net = monai.networks.nets.UNet(
dimensions=3,
in_channels=1,
out_channels=1,
channels=(16, 32, 64, 128, 256),
strides=(2, 2, 2, 2),
num_res_units=2,
).to(device)
val_post_transforms = Compose(
[
Activationsd(keys="pred", sigmoid=True),
AsDiscreted(keys="pred", threshold_values=True),
KeepLargestConnectedComponentd(keys="pred", applied_labels=[1]),
]
)
val_handlers = [
StatsHandler(output_transform=lambda x: None),
CheckpointLoader(load_path=model_file, load_dict={"net": net}),
SegmentationSaver(
output_dir="./runs/",
batch_transform=lambda batch: batch["image_meta_dict"],
output_transform=lambda output: output["pred"],
),
]
evaluator = SupervisedEvaluator(
device=device,
val_data_loader=val_loader,
network=net,
inferer=SlidingWindowInferer(roi_size=(96, 96, 96), sw_batch_size=4, overlap=0.5),
post_transform=val_post_transforms,
key_val_metric={
"val_mean_dice": MeanDice(include_background=True, output_transform=lambda x: (x["pred"], x["label"]))
},
additional_metrics={"val_acc": Accuracy(output_transform=lambda x: (x["pred"], x["label"]))},
val_handlers=val_handlers,
# if no FP16 support in GPU or PyTorch version < 1.6, will not enable AMP evaluation
amp=True if monai.config.get_torch_version_tuple() >= (1, 6) else False,
)
evaluator.run()
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),
AddChanneld(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, 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),
# test EnsureTensor for complicated dict data and invert it
CopyItemsd(PostFix.meta("image"), times=1, names="test_dict"),
# test to support Tensor, Numpy array and dictionary when inverting
EnsureTyped(keys=["image", "test_dict"]),
ToTensord("image"),
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 = CacheDataset(data, transform=transform, progress=False)
loader = DataLoader(dataset, num_workers=num_workers, batch_size=5)
inverter = Invertd(
# `image` was not copied, invert the original value directly
keys=["image_inverted", "label_inverted", "test_dict"],
transform=transform,
orig_keys=["label", "label", "test_dict"],
meta_keys=[
PostFix.meta("image_inverted"),
PostFix.meta("label_inverted"), None
],
orig_meta_keys=[
PostFix.meta("label"),
PostFix.meta("label"), None
],
nearest_interp=True,
to_tensor=[True, False, False],
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"],
meta_keys=[
PostFix.meta("image_inverted1"),
PostFix.meta("label_inverted1")
],
orig_meta_keys=[PostFix.meta("image"),
PostFix.meta("image")],
nearest_interp=[True, False],
to_tensor=[True, True],
device="cpu",
)
expected_keys = [
"image",
"image_inverted",
"image_inverted1",
PostFix.meta("image_inverted1"),
PostFix.meta("image_inverted"),
PostFix.meta("image"),
"image_transforms",
"label",
"label_inverted",
"label_inverted1",
PostFix.meta("label_inverted1"),
PostFix.meta("label_inverted"),
PostFix.meta("label"),
"label_transforms",
"test_dict",
"test_dict_transforms",
]
# 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))
# test inverted test_dict
self.assertTrue(
isinstance(item["test_dict"]["affine"], np.ndarray))
self.assertTrue(
isinstance(item["test_dict"]["filename_or_obj"], str))
# 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)(data[-1])["label"]
n_good = np.sum(np.isclose(reverted, original, atol=1e-3))
reverted_name = item[PostFix.meta("label_inverted")]["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) in (34007, 1812, 1821),
f"diff. {reverted.size - n_good}")
set_determinism(seed=None)
n = nib.Nifti1Image(im, np.eye(4))
nib.save(n, os.path.join(tempdir, 'im%i.nii.gz' % i))
n = nib.Nifti1Image(seg, np.eye(4))
nib.save(n, os.path.join(tempdir, 'seg%i.nii.gz' % i))
images = sorted(glob(os.path.join(tempdir, 'im*.nii.gz')))
segs = sorted(glob(os.path.join(tempdir, 'seg*.nii.gz')))
val_files = [{'img': img, 'seg': seg} for img, seg in zip(images, segs)]
# define transforms for image and segmentation
val_transforms = Compose([
LoadNiftid(keys=['img', 'seg']),
AsChannelFirstd(keys=['img', 'seg'], channel_dim=-1),
ScaleIntensityd(keys=['img', 'seg']),
ToTensord(keys=['img', 'seg'])
])
val_ds = monai.data.Dataset(data=val_files, transform=val_transforms)
device = torch.device("cuda:0")
net = UNet(
dimensions=3,
in_channels=1,
out_channels=1,
channels=(16, 32, 64, 128, 256),
strides=(2, 2, 2, 2),
num_res_units=2,
)
net.to(device)
def run_training_test(root_dir, device="cuda:0"):
real_images = sorted(glob(os.path.join(root_dir, "img*.nii.gz")))
train_files = [{"reals": img} for img in zip(real_images)]
# prepare real data
train_transforms = Compose([
LoadNiftid(keys=["reals"]),
AsChannelFirstd(keys=["reals"]),
ScaleIntensityd(keys=["reals"]),
RandFlipd(keys=["reals"], prob=0.5),
ToTensord(keys=["reals"]),
])
train_ds = monai.data.CacheDataset(data=train_files,
transform=train_transforms,
cache_rate=0.5)
train_loader = monai.data.DataLoader(train_ds,
batch_size=2,
shuffle=True,
num_workers=4)
learning_rate = 2e-4
betas = (0.5, 0.999)
real_label = 1
fake_label = 0
# create discriminator
disc_net = Discriminator(in_shape=(1, 64, 64),
channels=(8, 16, 32, 64, 1),
strides=(2, 2, 2, 2, 1),
num_res_units=1,
kernel_size=5).to(device)
disc_net.apply(normal_init)
disc_opt = torch.optim.Adam(disc_net.parameters(),
learning_rate,
betas=betas)
disc_loss_criterion = torch.nn.BCELoss()
def discriminator_loss(gen_images, real_images):
real = real_images.new_full((real_images.shape[0], 1), real_label)
gen = gen_images.new_full((gen_images.shape[0], 1), fake_label)
realloss = disc_loss_criterion(disc_net(real_images), real)
genloss = disc_loss_criterion(disc_net(gen_images.detach()), gen)
return torch.div(torch.add(realloss, genloss), 2)
# create generator
latent_size = 64
gen_net = Generator(latent_shape=latent_size,
start_shape=(latent_size, 8, 8),
channels=[32, 16, 8, 1],
strides=[2, 2, 2, 1])
gen_net.apply(normal_init)
gen_net.conv.add_module("activation", torch.nn.Sigmoid())
gen_net = gen_net.to(device)
gen_opt = torch.optim.Adam(gen_net.parameters(),
learning_rate,
betas=betas)
gen_loss_criterion = torch.nn.BCELoss()
def generator_loss(gen_images):
output = disc_net(gen_images)
cats = output.new_full(output.shape, real_label)
return gen_loss_criterion(output, cats)
key_train_metric = None
train_handlers = [
StatsHandler(
name="training_loss",
output_transform=lambda x: {
Keys.GLOSS: x[Keys.GLOSS],
Keys.DLOSS: x[Keys.DLOSS]
},
),
TensorBoardStatsHandler(
log_dir=root_dir,
tag_name="training_loss",
output_transform=lambda x: {
Keys.GLOSS: x[Keys.GLOSS],
Keys.DLOSS: x[Keys.DLOSS]
},
),
CheckpointSaver(save_dir=root_dir,
save_dict={
"g_net": gen_net,
"d_net": disc_net
},
save_interval=2,
epoch_level=True),
]
disc_train_steps = 2
num_epochs = 5
trainer = GanTrainer(
device,
num_epochs,
train_loader,
gen_net,
gen_opt,
generator_loss,
disc_net,
disc_opt,
discriminator_loss,
d_train_steps=disc_train_steps,
latent_shape=latent_size,
key_train_metric=key_train_metric,
train_handlers=train_handlers,
)
trainer.run()
return trainer.state
def run_training_test(root_dir, device="cuda:0", cachedataset=0):
monai.config.print_config()
images = sorted(glob(os.path.join(root_dir, "img*.nii.gz")))
segs = sorted(glob(os.path.join(root_dir, "seg*.nii.gz")))
train_files = [{"img": img, "seg": seg} for img, seg in zip(images[:20], segs[:20])]
val_files = [{"img": img, "seg": seg} for img, seg in zip(images[-20:], segs[-20:])]
# define transforms for image and segmentation
train_transforms = Compose(
[
LoadImaged(keys=["img", "seg"]),
AsChannelFirstd(keys=["img", "seg"], channel_dim=-1),
# resampling with align_corners=True or dtype=float64 will generate
# slight different results between PyTorch 1.5 an 1.6
Spacingd(keys=["img", "seg"], pixdim=[1.2, 0.8, 0.7], mode=["bilinear", "nearest"], dtype=np.float32),
ScaleIntensityd(keys="img"),
RandCropByPosNegLabeld(
keys=["img", "seg"], label_key="seg", spatial_size=[96, 96, 96], pos=1, neg=1, num_samples=4
),
RandRotate90d(keys=["img", "seg"], prob=0.8, spatial_axes=[0, 2]),
ToTensord(keys=["img", "seg"]),
]
)
train_transforms.set_random_state(1234)
val_transforms = Compose(
[
LoadImaged(keys=["img", "seg"]),
AsChannelFirstd(keys=["img", "seg"], channel_dim=-1),
# resampling with align_corners=True or dtype=float64 will generate
# slight different results between PyTorch 1.5 an 1.6
Spacingd(keys=["img", "seg"], pixdim=[1.2, 0.8, 0.7], mode=["bilinear", "nearest"], dtype=np.float32),
ScaleIntensityd(keys="img"),
ToTensord(keys=["img", "seg"]),
]
)
# create a training data loader
if cachedataset == 2:
train_ds = monai.data.CacheDataset(data=train_files, transform=train_transforms, cache_rate=0.8)
elif cachedataset == 3:
train_ds = monai.data.LMDBDataset(data=train_files, transform=train_transforms)
else:
train_ds = monai.data.Dataset(data=train_files, transform=train_transforms)
# use batch_size=2 to load images and use RandCropByPosNegLabeld to generate 2 x 4 images for network training
train_loader = monai.data.DataLoader(train_ds, batch_size=2, shuffle=True, num_workers=4)
# create a validation data loader
val_ds = monai.data.Dataset(data=val_files, transform=val_transforms)
val_loader = monai.data.DataLoader(val_ds, batch_size=1, num_workers=4)
val_post_tran = Compose([Activations(sigmoid=True), AsDiscrete(threshold_values=True)])
dice_metric = DiceMetric(include_background=True, reduction="mean")
# create UNet, DiceLoss and Adam optimizer
model = monai.networks.nets.UNet(
dimensions=3,
in_channels=1,
out_channels=1,
channels=(16, 32, 64, 128, 256),
strides=(2, 2, 2, 2),
num_res_units=2,
).to(device)
loss_function = monai.losses.DiceLoss(sigmoid=True)
optimizer = torch.optim.Adam(model.parameters(), 5e-4)
# start a typical PyTorch training
val_interval = 2
best_metric, best_metric_epoch = -1, -1
epoch_loss_values = list()
metric_values = list()
writer = SummaryWriter(log_dir=os.path.join(root_dir, "runs"))
model_filename = os.path.join(root_dir, "best_metric_model.pth")
for epoch in range(6):
print("-" * 10)
print(f"Epoch {epoch + 1}/{6}")
model.train()
epoch_loss = 0
step = 0
for batch_data in train_loader:
step += 1
inputs, labels = batch_data["img"].to(device), batch_data["seg"].to(device)
optimizer.zero_grad()
outputs = model(inputs)
loss = loss_function(outputs, labels)
loss.backward()
optimizer.step()
epoch_loss += loss.item()
epoch_len = len(train_ds) // train_loader.batch_size
print(f"{step}/{epoch_len}, train_loss:{loss.item():0.4f}")
writer.add_scalar("train_loss", loss.item(), epoch_len * epoch + step)
epoch_loss /= step
epoch_loss_values.append(epoch_loss)
print(f"epoch {epoch +1} average loss:{epoch_loss:0.4f}")
if (epoch + 1) % val_interval == 0:
model.eval()
with torch.no_grad():
metric_sum = 0.0
metric_count = 0
val_images = None
val_labels = None
val_outputs = None
for val_data in val_loader:
val_images, val_labels = val_data["img"].to(device), val_data["seg"].to(device)
sw_batch_size, roi_size = 4, (96, 96, 96)
val_outputs = val_post_tran(sliding_window_inference(val_images, roi_size, sw_batch_size, model))
value, not_nans = dice_metric(y_pred=val_outputs, y=val_labels)
metric_count += not_nans.item()
metric_sum += value.item() * not_nans.item()
metric = metric_sum / metric_count
metric_values.append(metric)
if metric > best_metric:
best_metric = metric
best_metric_epoch = epoch + 1
torch.save(model.state_dict(), model_filename)
print("saved new best metric model")
print(
f"current epoch {epoch +1} current mean dice: {metric:0.4f} "
f"best mean dice: {best_metric:0.4f} at epoch {best_metric_epoch}"
)
writer.add_scalar("val_mean_dice", metric, epoch + 1)
# plot the last model output as GIF image in TensorBoard with the corresponding image and label
plot_2d_or_3d_image(val_images, epoch + 1, writer, index=0, tag="image")
plot_2d_or_3d_image(val_labels, epoch + 1, writer, index=0, tag="label")
plot_2d_or_3d_image(val_outputs, epoch + 1, writer, index=0, tag="output")
print(f"train completed, best_metric: {best_metric:0.4f} at epoch: {best_metric_epoch}")
writer.close()
return epoch_loss_values, best_metric, best_metric_epoch
def main():
monai.config.print_config()
logging.basicConfig(stream=sys.stdout, level=logging.INFO)
# IXI dataset as a demo, downloadable from https://brain-development.org/ixi-dataset/
images = [
"/workspace/data/medical/ixi/IXI-T1/IXI314-IOP-0889-T1.nii.gz",
"/workspace/data/medical/ixi/IXI-T1/IXI249-Guys-1072-T1.nii.gz",
"/workspace/data/medical/ixi/IXI-T1/IXI609-HH-2600-T1.nii.gz",
"/workspace/data/medical/ixi/IXI-T1/IXI173-HH-1590-T1.nii.gz",
"/workspace/data/medical/ixi/IXI-T1/IXI020-Guys-0700-T1.nii.gz",
"/workspace/data/medical/ixi/IXI-T1/IXI342-Guys-0909-T1.nii.gz",
"/workspace/data/medical/ixi/IXI-T1/IXI134-Guys-0780-T1.nii.gz",
"/workspace/data/medical/ixi/IXI-T1/IXI577-HH-2661-T1.nii.gz",
"/workspace/data/medical/ixi/IXI-T1/IXI066-Guys-0731-T1.nii.gz",
"/workspace/data/medical/ixi/IXI-T1/IXI130-HH-1528-T1.nii.gz",
"/workspace/data/medical/ixi/IXI-T1/IXI607-Guys-1097-T1.nii.gz",
"/workspace/data/medical/ixi/IXI-T1/IXI175-HH-1570-T1.nii.gz",
"/workspace/data/medical/ixi/IXI-T1/IXI385-HH-2078-T1.nii.gz",
"/workspace/data/medical/ixi/IXI-T1/IXI344-Guys-0905-T1.nii.gz",
"/workspace/data/medical/ixi/IXI-T1/IXI409-Guys-0960-T1.nii.gz",
"/workspace/data/medical/ixi/IXI-T1/IXI584-Guys-1129-T1.nii.gz",
"/workspace/data/medical/ixi/IXI-T1/IXI253-HH-1694-T1.nii.gz",
"/workspace/data/medical/ixi/IXI-T1/IXI092-HH-1436-T1.nii.gz",
"/workspace/data/medical/ixi/IXI-T1/IXI574-IOP-1156-T1.nii.gz",
"/workspace/data/medical/ixi/IXI-T1/IXI585-Guys-1130-T1.nii.gz",
]
# 2 binary labels for gender classification: man and woman
labels = np.array(
[0, 0, 0, 1, 0, 0, 0, 1, 1, 0, 0, 0, 1, 0, 1, 0, 1, 0, 1, 0])
train_files = [{
"img": img,
"label": label
} for img, label in zip(images[:10], labels[:10])]
val_files = [{
"img": img,
"label": label
} for img, label in zip(images[-10:], labels[-10:])]
# define transforms for image
train_transforms = Compose([
LoadNiftid(keys=["img"]),
AddChanneld(keys=["img"]),
ScaleIntensityd(keys=["img"]),
Resized(keys=["img"], spatial_size=(96, 96, 96)),
RandRotate90d(keys=["img"], prob=0.8, spatial_axes=[0, 2]),
ToTensord(keys=["img"]),
])
val_transforms = Compose([
LoadNiftid(keys=["img"]),
AddChanneld(keys=["img"]),
ScaleIntensityd(keys=["img"]),
Resized(keys=["img"], spatial_size=(96, 96, 96)),
ToTensord(keys=["img"]),
])
# define dataset, data loader
check_ds = monai.data.Dataset(data=train_files, transform=train_transforms)
check_loader = DataLoader(check_ds,
batch_size=2,
num_workers=4,
pin_memory=torch.cuda.is_available())
check_data = monai.utils.misc.first(check_loader)
print(check_data["img"].shape, check_data["label"])
# create DenseNet121, CrossEntropyLoss and Adam optimizer
net = monai.networks.nets.densenet.densenet121(
spatial_dims=3,
in_channels=1,
out_channels=2,
)
loss = torch.nn.CrossEntropyLoss()
lr = 1e-5
opt = torch.optim.Adam(net.parameters(), lr)
device = torch.device("cuda:0")
# Ignite trainer expects batch=(img, label) and returns output=loss at every iteration,
# user can add output_transform to return other values, like: y_pred, y, etc.
def prepare_batch(batch, device=None, non_blocking=False):
return _prepare_batch((batch["img"], batch["label"]), device,
non_blocking)
trainer = create_supervised_trainer(net,
opt,
loss,
device,
False,
prepare_batch=prepare_batch)
# adding checkpoint handler to save models (network params and optimizer stats) during training
checkpoint_handler = ModelCheckpoint("./runs/",
"net",
n_saved=10,
require_empty=False)
trainer.add_event_handler(event_name=Events.EPOCH_COMPLETED,
handler=checkpoint_handler,
to_save={
"net": net,
"opt": opt
})
# StatsHandler prints loss at every iteration and print metrics at every epoch,
# we don't set metrics for trainer here, so just print loss, user can also customize print functions
# and can use output_transform to convert engine.state.output if it's not loss value
train_stats_handler = StatsHandler(name="trainer")
train_stats_handler.attach(trainer)
# TensorBoardStatsHandler plots loss at every iteration and plots metrics at every epoch, same as StatsHandler
train_tensorboard_stats_handler = TensorBoardStatsHandler()
train_tensorboard_stats_handler.attach(trainer)
# set parameters for validation
validation_every_n_epochs = 1
metric_name = "Accuracy"
# add evaluation metric to the evaluator engine
val_metrics = {
metric_name: Accuracy(),
"AUC": ROCAUC(to_onehot_y=True, add_softmax=True)
}
# Ignite evaluator expects batch=(img, label) and returns output=(y_pred, y) at every iteration,
# user can add output_transform to return other values
evaluator = create_supervised_evaluator(net,
val_metrics,
device,
True,
prepare_batch=prepare_batch)
# add stats event handler to print validation stats via evaluator
val_stats_handler = StatsHandler(
name="evaluator",
output_transform=lambda x:
None, # no need to print loss value, so disable per iteration output
global_epoch_transform=lambda x: trainer.state.epoch,
) # fetch global epoch number from trainer
val_stats_handler.attach(evaluator)
# add handler to record metrics to TensorBoard at every epoch
val_tensorboard_stats_handler = TensorBoardStatsHandler(
output_transform=lambda x:
None, # no need to plot loss value, so disable per iteration output
global_epoch_transform=lambda x: trainer.state.epoch,
) # fetch global epoch number from trainer
val_tensorboard_stats_handler.attach(evaluator)
# add early stopping handler to evaluator
early_stopper = EarlyStopping(
patience=4,
score_function=stopping_fn_from_metric(metric_name),
trainer=trainer)
evaluator.add_event_handler(event_name=Events.EPOCH_COMPLETED,
handler=early_stopper)
# create a validation data loader
val_ds = monai.data.Dataset(data=val_files, transform=val_transforms)
val_loader = DataLoader(val_ds,
batch_size=2,
num_workers=4,
pin_memory=torch.cuda.is_available())
@trainer.on(Events.EPOCH_COMPLETED(every=validation_every_n_epochs))
def run_validation(engine):
evaluator.run(val_loader)
# create a training data loader
train_ds = monai.data.Dataset(data=train_files, transform=train_transforms)
train_loader = DataLoader(train_ds,
batch_size=2,
shuffle=True,
num_workers=4,
pin_memory=torch.cuda.is_available())
train_epochs = 30
state = trainer.run(train_loader, train_epochs)
