def test_correct_results(self, _, keys, mean, std):
gaussian_fn = RandGaussianNoised(keys=keys, prob=1.0, mean=mean, std=std)
gaussian_fn.set_random_state(seed)
noised = gaussian_fn({"img": self.imt})
np.random.seed(seed)
np.random.random()
expected = self.imt + np.random.normal(mean, np.random.uniform(0, std), size=self.imt.shape)
np.testing.assert_allclose(expected, noised["img"], atol=1e-5, rtol=1e-5)
def test_numpy_or_torch(keys, mean, std, imt):
gaussian_fn = RandGaussianNoised(keys=keys, prob=1.0, mean=mean, std=std)
gaussian_fn.set_random_state(seed)
noised = gaussian_fn({k: imt for k in keys})
np.random.seed(seed)
np.random.random()
for k in keys:
expected = imt + np.random.normal(
mean, np.random.uniform(0, std), size=imt.shape)
np.testing.assert_allclose(expected, noised[k], atol=1e-5, rtol=1e-5)
def test_correct_results(self, _, im_type, keys, mean, std):
gaussian_fn = RandGaussianNoised(keys=keys,
prob=1.0,
mean=mean,
std=std,
dtype=np.float64)
gaussian_fn.set_random_state(seed)
im = im_type(self.imt)
noised = gaussian_fn({k: im for k in keys})
np.random.seed(seed)
# simulate the randomize() of transform
np.random.random()
noise = np.random.normal(mean,
np.random.uniform(0, std),
size=self.imt.shape)
for k in keys:
expected = self.imt + noise
if isinstance(noised[k], torch.Tensor):
noised[k] = noised[k].cpu()
np.testing.assert_allclose(expected,
noised[k],
atol=1e-5,
rtol=1e-5)
def get_xforms(args, mode="train", keys=("image", "label")):
"""returns a composed transform for train/val/infer."""
xforms = [
LoadNiftid(keys),
AddChanneld(keys),
Orientationd(keys, axcodes="LPS"),
Spacingd(keys,
pixdim=(1.25, 1.25, 5.0),
mode=("bilinear", "nearest")[:len(keys)]),
ScaleIntensityRanged(keys[0],
a_min=-1000.0,
a_max=500.0,
b_min=0.0,
b_max=1.0,
clip=True),
]
if mode == "train":
xforms.extend([
SpatialPadd(keys,
spatial_size=(args.patch_size, args.patch_size, -1),
mode="reflect"), # ensure at least 192x192
RandAffined(
keys,
prob=0.15,
rotate_range=(-0.05, 0.05),
scale_range=(-0.1, 0.1),
mode=("bilinear", "nearest"),
as_tensor_output=False,
),
RandCropByPosNegLabeld(keys,
label_key=keys[1],
spatial_size=(args.patch_size,
args.patch_size,
args.n_slice),
num_samples=3),
RandGaussianNoised(keys[0], prob=0.15, std=0.01),
RandFlipd(keys, spatial_axis=0, prob=0.5),
RandFlipd(keys, spatial_axis=1, prob=0.5),
RandFlipd(keys, spatial_axis=2, prob=0.5),
])
dtype = (np.float32, np.uint8)
if mode == "val":
dtype = (np.float32, np.uint8)
if mode == "infer":
dtype = (np.float32, )
xforms.extend([CastToTyped(keys, dtype=dtype), ToTensord(keys)])
return monai.transforms.Compose(xforms)
def get_xforms_with_synthesis(mode="synthesis", keys=("image", "label"), keys2=("image", "label", "synthetic_lesion")):
"""returns a composed transform for train/val/infer."""
xforms = [
LoadImaged(keys),
AddChanneld(keys),
Orientationd(keys, axcodes="LPS"),
Spacingd(keys, pixdim=(1.25, 1.25, 5.0), mode=("bilinear", "nearest")[: len(keys)]),
ScaleIntensityRanged(keys[0], a_min=-1000.0, a_max=500.0, b_min=0.0, b_max=1.0, clip=True),
CopyItemsd(keys,1, names=['image_1', 'label_1']),
]
if mode == "synthesis":
xforms.extend([
SpatialPadd(keys, spatial_size=(192, 192, -1), mode="reflect"), # ensure at least 192x192
RandCropByPosNegLabeld(keys, label_key=keys[1], spatial_size=(192, 192, 16), num_samples=3),
TransCustom(keys, path_synthesis, read_cea_aug_slice2,
pseudo_healthy_with_texture, scans_syns, decreasing_sequence, GEN=15,
POST_PROCESS=True, mask_outer_ring=True, new_value=.5),
RandAffined(
# keys,
keys2,
prob=0.15,
rotate_range=(0.05, 0.05, None), # 3 parameters control the transform on 3 dimensions
scale_range=(0.1, 0.1, None),
mode=("bilinear", "nearest", "bilinear"),
# mode=("bilinear", "nearest"),
as_tensor_output=False
),
RandGaussianNoised((keys2[0],keys2[2]), prob=0.15, std=0.01),
# RandGaussianNoised(keys[0], prob=0.15, std=0.01),
RandFlipd(keys, spatial_axis=0, prob=0.5),
RandFlipd(keys, spatial_axis=1, prob=0.5),
RandFlipd(keys, spatial_axis=2, prob=0.5),
TransCustom2(0.333)
])
dtype = (np.float32, np.uint8)
# dtype = (np.float32, np.uint8, np.float32)
xforms.extend([CastToTyped(keys, dtype=dtype)])
return monai.transforms.Compose(xforms)
def run_training(train_file_list, valid_file_list, config_info):
"""
Pipeline to train a dynUNet segmentation model in MONAI. It is composed of the following main blocks:
* Data Preparation: Extract the filenames and prepare the training/validation processing transforms
* Load Data: Load training and validation data to PyTorch DataLoader
* Network Preparation: Define the network, loss function, optimiser and learning rate scheduler
* MONAI Evaluator: Initialise the dynUNet evaluator, i.e. the class providing utilities to perform validation
during training. Attach handlers to save the best model on the validation set. A 2D sliding window approach
on the 3D volume is used at evaluation. The mean 3D Dice is used as validation metric.
* MONAI Trainer: Initialise the dynUNet trainer, i.e. the class providing utilities to perform the training loop.
* Run training: The MONAI trainer is run, performing training and validation during training.
Args:
train_file_list: .txt or .csv file (with no header) storing two-columns filenames for training:
image filename in the first column and segmentation filename in the second column.
The two columns should be separated by a comma.
See monaifbs/config/mock_train_file_list_for_dynUnet_training.txt for an example of the expected format.
valid_file_list: .txt or .csv file (with no header) storing two-columns filenames for validation:
image filename in the first column and segmentation filename in the second column.
The two columns should be separated by a comma.
See monaifbs/config/mock_valid_file_list_for_dynUnet_training.txt for an example of the expected format.
config_info: dict, contains configuration parameters for sampling, network and training.
See monaifbs/config/monai_dynUnet_training_config.yml for an example of the expected fields.
"""
"""
Read input and configuration parameters
"""
# print MONAI config information
logging.basicConfig(stream=sys.stdout, level=logging.INFO)
print_config()
# print to log the parameter setups
print(yaml.dump(config_info))
# extract network parameters, perform checks/set defaults if not present and print them to log
if 'seg_labels' in config_info['training'].keys():
seg_labels = config_info['training']['seg_labels']
else:
seg_labels = [1]
nr_out_channels = len(seg_labels)
print("Considering the following {} labels in the segmentation: {}".format(nr_out_channels, seg_labels))
patch_size = config_info["training"]["inplane_size"] + [1]
print("Considering patch size = {}".format(patch_size))
spacing = config_info["training"]["spacing"]
print("Bringing all images to spacing = {}".format(spacing))
if 'model_to_load' in config_info['training'].keys() and config_info['training']['model_to_load'] is not None:
model_to_load = config_info['training']['model_to_load']
if not os.path.exists(model_to_load):
raise FileNotFoundError("Cannot find model: {}".format(model_to_load))
else:
print("Loading model from {}".format(model_to_load))
else:
model_to_load = None
# set up either GPU or CPU usage
if torch.cuda.is_available():
print("\n#### GPU INFORMATION ###")
print("Using device number: {}, name: {}\n".format(torch.cuda.current_device(), torch.cuda.get_device_name()))
current_device = torch.device("cuda:0")
else:
current_device = torch.device("cpu")
print("Using device: {}".format(current_device))
# set determinism if required
if 'manual_seed' in config_info['training'].keys() and config_info['training']['manual_seed'] is not None:
seed = config_info['training']['manual_seed']
else:
seed = None
if seed is not None:
print("Using determinism with seed = {}\n".format(seed))
set_determinism(seed=seed)
"""
Setup data output directory
"""
out_model_dir = os.path.join(config_info['output']['out_dir'],
datetime.now().strftime('%Y-%m-%d_%H-%M-%S') + '_' +
config_info['output']['out_postfix'])
print("Saving to directory {}\n".format(out_model_dir))
# create cache directory to store results for Persistent Dataset
if 'cache_dir' in config_info['output'].keys():
out_cache_dir = config_info['output']['cache_dir']
else:
out_cache_dir = os.path.join(out_model_dir, 'persistent_cache')
persistent_cache: Path = Path(out_cache_dir)
persistent_cache.mkdir(parents=True, exist_ok=True)
"""
Data preparation
"""
# Read the input files for training and validation
print("*** Loading input data for training...")
train_files = create_data_list_of_dictionaries(train_file_list)
print("Number of inputs for training = {}".format(len(train_files)))
val_files = create_data_list_of_dictionaries(valid_file_list)
print("Number of inputs for validation = {}".format(len(val_files)))
# Define MONAI processing transforms for the training data. This includes:
# - Load Nifti files and convert to format Batch x Channel x Dim1 x Dim2 x Dim3
# - CropForegroundd: Reduce the background from the MR image
# - InPlaneSpacingd: Perform in-plane resampling to the desired spacing, but preserve the resolution along the
# last direction (lowest resolution) to avoid introducing motion artefact resampling errors
# - SpatialPadd: Pad the in-plane size to the defined network input patch size [N, M] if needed
# - NormalizeIntensityd: Apply whitening
# - RandSpatialCropd: Crop a random patch from the input with size [B, C, N, M, 1]
# - SqueezeDimd: Convert the 3D patch to a 2D one as input to the network (i.e. bring it to size [B, C, N, M])
# - Apply data augmentation (RandZoomd, RandRotated, RandGaussianNoised, RandGaussianSmoothd, RandScaleIntensityd,
# RandFlipd)
# - ToTensor: convert to pytorch tensor
train_transforms = Compose(
[
LoadNiftid(keys=["image", "label"]),
AddChanneld(keys=["image", "label"]),
CropForegroundd(keys=["image", "label"], source_key="image"),
InPlaneSpacingd(
keys=["image", "label"],
pixdim=spacing,
mode=("bilinear", "nearest"),
),
SpatialPadd(keys=["image", "label"], spatial_size=patch_size,
mode=["constant", "edge"]),
NormalizeIntensityd(keys=["image"], nonzero=False, channel_wise=True),
RandSpatialCropd(keys=["image", "label"], roi_size=patch_size, random_size=False),
SqueezeDimd(keys=["image", "label"], dim=-1),
RandZoomd(
keys=["image", "label"],
min_zoom=0.9,
max_zoom=1.2,
mode=("bilinear", "nearest"),
align_corners=(True, None),
prob=0.16,
),
RandRotated(keys=["image", "label"], range_x=90, range_y=90, prob=0.2,
keep_size=True, mode=["bilinear", "nearest"],
padding_mode=["zeros", "border"]),
RandGaussianNoised(keys=["image"], std=0.01, prob=0.15),
RandGaussianSmoothd(
keys=["image"],
sigma_x=(0.5, 1.15),
sigma_y=(0.5, 1.15),
sigma_z=(0.5, 1.15),
prob=0.15,
),
RandScaleIntensityd(keys=["image"], factors=0.3, prob=0.15),
RandFlipd(["image", "label"], spatial_axis=[0, 1], prob=0.5),
ToTensord(keys=["image", "label"]),
]
)
# Define MONAI processing transforms for the validation data
# - Load Nifti files and convert to format Batch x Channel x Dim1 x Dim2 x Dim3
# - CropForegroundd: Reduce the background from the MR image
# - InPlaneSpacingd: Perform in-plane resampling to the desired spacing, but preserve the resolution along the
# last direction (lowest resolution) to avoid introducing motion artefact resampling errors
# - SpatialPadd: Pad the in-plane size to the defined network input patch size [N, M] if needed
# - NormalizeIntensityd: Apply whitening
# - ToTensor: convert to pytorch tensor
# NOTE: The validation data is kept 3D as a 2D sliding window approach is used throughout the volume at inference
val_transforms = Compose(
[
LoadNiftid(keys=["image", "label"]),
AddChanneld(keys=["image", "label"]),
CropForegroundd(keys=["image", "label"], source_key="image"),
InPlaneSpacingd(
keys=["image", "label"],
pixdim=spacing,
mode=("bilinear", "nearest"),
),
SpatialPadd(keys=["image", "label"], spatial_size=patch_size, mode=["constant", "edge"]),
NormalizeIntensityd(keys=["image"], nonzero=False, channel_wise=True),
ToTensord(keys=["image", "label"]),
]
)
"""
Load data
"""
# create training data loader
train_ds = PersistentDataset(data=train_files, transform=train_transforms,
cache_dir=persistent_cache)
train_loader = DataLoader(train_ds,
batch_size=config_info['training']['batch_size_train'],
shuffle=True,
num_workers=config_info['device']['num_workers'])
check_train_data = misc.first(train_loader)
print("Training data tensor shapes:")
print("Image = {}; Label = {}".format(check_train_data["image"].shape, check_train_data["label"].shape))
# create validation data loader
if config_info['training']['batch_size_valid'] != 1:
raise Exception("Batch size different from 1 at validation ar currently not supported")
val_ds = PersistentDataset(data=val_files, transform=val_transforms, cache_dir=persistent_cache)
val_loader = DataLoader(val_ds,
batch_size=1,
shuffle=False,
num_workers=config_info['device']['num_workers'])
check_valid_data = misc.first(val_loader)
print("Validation data tensor shapes (Example):")
print("Image = {}; Label = {}\n".format(check_valid_data["image"].shape, check_valid_data["label"].shape))
"""
Network preparation
"""
print("*** Preparing the network ...")
# automatically extracts the strides and kernels based on nnU-Net empirical rules
spacings = spacing[:2]
sizes = patch_size[:2]
strides, kernels = [], []
while True:
spacing_ratio = [sp / min(spacings) for sp in spacings]
stride = [2 if ratio <= 2 and size >= 8 else 1 for (ratio, size) in zip(spacing_ratio, sizes)]
kernel = [3 if ratio <= 2 else 1 for ratio in spacing_ratio]
if all(s == 1 for s in stride):
break
sizes = [i / j for i, j in zip(sizes, stride)]
spacings = [i * j for i, j in zip(spacings, stride)]
kernels.append(kernel)
strides.append(stride)
strides.insert(0, len(spacings) * [1])
kernels.append(len(spacings) * [3])
# initialise the network
net = DynUNet(
spatial_dims=2,
in_channels=1,
out_channels=nr_out_channels,
kernel_size=kernels,
strides=strides,
upsample_kernel_size=strides[1:],
norm_name="instance",
deep_supervision=True,
deep_supr_num=2,
res_block=False,
).to(current_device)
print(net)
# define the loss function
loss_function = choose_loss_function(nr_out_channels, config_info)
# define the optimiser and the learning rate scheduler
opt = torch.optim.SGD(net.parameters(), lr=float(config_info['training']['lr']), momentum=0.95)
scheduler = torch.optim.lr_scheduler.LambdaLR(
opt, lr_lambda=lambda epoch: (1 - epoch / config_info['training']['nr_train_epochs']) ** 0.9
)
"""
MONAI evaluator
"""
print("*** Preparing the dynUNet evaluator engine...\n")
# val_post_transforms = Compose(
# [
# Activationsd(keys="pred", sigmoid=True),
# ]
# )
val_handlers = [
StatsHandler(output_transform=lambda x: None),
TensorBoardStatsHandler(log_dir=os.path.join(out_model_dir, "valid"),
output_transform=lambda x: None,
global_epoch_transform=lambda x: trainer.state.iteration),
CheckpointSaver(save_dir=out_model_dir, save_dict={"net": net, "opt": opt}, save_key_metric=True,
file_prefix='best_valid'),
]
if config_info['output']['val_image_to_tensorboad']:
val_handlers.append(TensorBoardImageHandler(log_dir=os.path.join(out_model_dir, "valid"),
batch_transform=lambda x: (x["image"], x["label"]),
output_transform=lambda x: x["pred"], interval=2))
# Define customized evaluator
class DynUNetEvaluator(SupervisedEvaluator):
def _iteration(self, engine, batchdata):
inputs, targets = self.prepare_batch(batchdata)
inputs, targets = inputs.to(engine.state.device), targets.to(engine.state.device)
flip_inputs_1 = torch.flip(inputs, dims=(2,))
flip_inputs_2 = torch.flip(inputs, dims=(3,))
flip_inputs_3 = torch.flip(inputs, dims=(2, 3))
def _compute_pred():
pred = self.inferer(inputs, self.network)
# use random flipping as data augmentation at inference
flip_pred_1 = torch.flip(self.inferer(flip_inputs_1, self.network), dims=(2,))
flip_pred_2 = torch.flip(self.inferer(flip_inputs_2, self.network), dims=(3,))
flip_pred_3 = torch.flip(self.inferer(flip_inputs_3, self.network), dims=(2, 3))
return (pred + flip_pred_1 + flip_pred_2 + flip_pred_3) / 4
# execute forward computation
self.network.eval()
with torch.no_grad():
if self.amp:
with torch.cuda.amp.autocast():
predictions = _compute_pred()
else:
predictions = _compute_pred()
return {"image": inputs, "label": targets, "pred": predictions}
evaluator = DynUNetEvaluator(
device=current_device,
val_data_loader=val_loader,
network=net,
inferer=SlidingWindowInferer2D(roi_size=patch_size, sw_batch_size=4, overlap=0.0),
post_transform=None,
key_val_metric={
"Mean_dice": MeanDice(
include_background=False,
to_onehot_y=True,
mutually_exclusive=True,
output_transform=lambda x: (x["pred"], x["label"]),
)
},
val_handlers=val_handlers,
amp=False,
)
"""
MONAI trainer
"""
print("*** Preparing the dynUNet trainer engine...\n")
# train_post_transforms = Compose(
# [
# Activationsd(keys="pred", sigmoid=True),
# ]
# )
validation_every_n_epochs = config_info['training']['validation_every_n_epochs']
epoch_len = len(train_ds) // train_loader.batch_size
validation_every_n_iters = validation_every_n_epochs * epoch_len
# define event handlers for the trainer
writer_train = SummaryWriter(log_dir=os.path.join(out_model_dir, "train"))
train_handlers = [
LrScheduleHandler(lr_scheduler=scheduler, print_lr=True),
ValidationHandler(validator=evaluator, interval=validation_every_n_iters, epoch_level=False),
StatsHandler(tag_name="train_loss", output_transform=lambda x: x["loss"]),
TensorBoardStatsHandler(summary_writer=writer_train,
log_dir=os.path.join(out_model_dir, "train"), tag_name="Loss",
output_transform=lambda x: x["loss"],
global_epoch_transform=lambda x: trainer.state.iteration),
CheckpointSaver(save_dir=out_model_dir, save_dict={"net": net, "opt": opt},
save_final=True,
save_interval=2, epoch_level=True,
n_saved=config_info['output']['max_nr_models_saved']),
]
if model_to_load is not None:
train_handlers.append(CheckpointLoader(load_path=model_to_load, load_dict={"net": net, "opt": opt}))
# define customized trainer
class DynUNetTrainer(SupervisedTrainer):
def _iteration(self, engine, batchdata):
inputs, targets = self.prepare_batch(batchdata)
inputs, targets = inputs.to(engine.state.device), targets.to(engine.state.device)
def _compute_loss(preds, label):
labels = [label] + [interpolate(label, pred.shape[2:]) for pred in preds[1:]]
return sum([0.5 ** i * self.loss_function(p, l) for i, (p, l) in enumerate(zip(preds, labels))])
self.network.train()
self.optimizer.zero_grad()
if self.amp and self.scaler is not None:
with torch.cuda.amp.autocast():
predictions = self.inferer(inputs, self.network)
loss = _compute_loss(predictions, targets)
self.scaler.scale(loss).backward()
self.scaler.step(self.optimizer)
self.scaler.update()
else:
predictions = self.inferer(inputs, self.network)
loss = _compute_loss(predictions, targets).mean()
loss.backward()
self.optimizer.step()
return {"image": inputs, "label": targets, "pred": predictions, "loss": loss.item()}
trainer = DynUNetTrainer(
device=current_device,
max_epochs=config_info['training']['nr_train_epochs'],
train_data_loader=train_loader,
network=net,
optimizer=opt,
loss_function=loss_function,
inferer=SimpleInferer(),
post_transform=None,
key_train_metric=None,
train_handlers=train_handlers,
amp=False,
)
"""
Run training
"""
print("*** Run training...")
trainer.run()
print("Done!")
keys=["image", "label"],
sigma_range=(0, 1),
magnitude_range=(0, 1),
spatial_size=None,
prob=0.5,
rotate_range=(0, -math.pi / 36, math.pi / 36, 0), # -15, 15 / -5, 5
shear_range=None,
translate_range=None,
scale_range=None,
mode=("bilinear", "nearest"),
padding_mode="zeros",
# as_tensor_output=False
),
RandGaussianNoised(keys=["image"],
prob=0.5,
mean=0.0,
std=0.1
# allow_missing_keys=False
),
#RandScaleIntensityd(
# keys=["image"],
# factors=0.05, # this is 10%, try 5%
# prob=0.5
#),
#RandGaussianSmoothd(
# keys=["image"],
# sigma_x=(0.25, 1.5),
# sigma_y=(0.25, 1.5),
# sigma_z=(0.25, 1.5),
# prob=0.5,
# approx='erf'
# allow_missing_keys=False
def infer(data_folder=".", model_folder="runs", prediction_folder="output"):
"""
run inference, the output folder will be "./output"
"""
ckpts = sorted(glob.glob(os.path.join(model_folder, "*.pt")))
ckpt = ckpts[-1]
for x in ckpts:
logging.info(f"available model file: {x}.")
logging.info("----")
logging.info(f"using {ckpt}.")
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
net = get_net().to(device)
net.load_state_dict(torch.load(ckpt, map_location=device))
net.eval()
image_folder = os.path.abspath(data_folder)
images = sorted(glob.glob(os.path.join(image_folder, "*_ct.nii.gz")))
logging.info(f"infer: image ({len(images)}) folder: {data_folder}")
infer_files = [{"image": img} for img in images]
keys = ("image", )
infer_transforms = get_xforms("infer", keys)
infer_ds = monai.data.Dataset(data=infer_files, transform=infer_transforms)
infer_loader = monai.data.DataLoader(
infer_ds,
batch_size=
1, # image-level batch to the sliding window method, not the window-level batch
num_workers=2,
pin_memory=torch.cuda.is_available(),
)
inferer = get_inferer()
saver = monai.data.NiftiSaver(output_dir=prediction_folder, mode="nearest")
with torch.no_grad():
for infer_data in infer_loader:
logging.info(
f"segmenting {infer_data['image_meta_dict']['filename_or_obj']}"
)
preds = inferer(infer_data[keys[0]].to(device), net)
n = 1.0
for _ in range(4):
# test time augmentations
_img = RandGaussianNoised(keys[0], prob=1.0,
std=0.01)(infer_data)[keys[0]]
pred = inferer(_img.to(device), net)
preds = preds + pred
n = n + 1.0
for dims in [[2], [3]]:
flip_pred = inferer(torch.flip(_img.to(device), dims=dims),
net)
pred = torch.flip(flip_pred, dims=dims)
preds = preds + pred
n = n + 1.0
preds = preds / n
preds = (preds.argmax(dim=1, keepdims=True)).float()
saver.save_batch(preds, infer_data["image_meta_dict"])
# copy the saved segmentations into the required folder structure for submission
submission_dir = os.path.join(prediction_folder, "to_submit")
if not os.path.exists(submission_dir):
os.makedirs(submission_dir)
files = glob.glob(os.path.join(prediction_folder, "volume*", "*.nii.gz"))
for f in files:
new_name = os.path.basename(f)
new_name = new_name[len("volume-covid19-A-0"):]
new_name = new_name[:-len("_ct_seg.nii.gz")] + ".nii.gz"
to_name = os.path.join(submission_dir, new_name)
shutil.copy(f, to_name)
logging.info(f"predictions copied to {submission_dir}.")
def main():
opt = Options().parse()
# monai.config.print_config()
logging.basicConfig(stream=sys.stdout, level=logging.INFO)
if opt.gpu_ids != '-1':
num_gpus = len(opt.gpu_ids.split(','))
else:
num_gpus = 0
print('number of GPU:', num_gpus)
# Data loader creation
# train images
train_images = sorted(glob(os.path.join(opt.images_folder, 'train', 'image*.nii')))
train_segs = sorted(glob(os.path.join(opt.labels_folder, 'train', 'label*.nii')))
train_images_for_dice = sorted(glob(os.path.join(opt.images_folder, 'train', 'image*.nii')))
train_segs_for_dice = sorted(glob(os.path.join(opt.labels_folder, 'train', 'label*.nii')))
# validation images
val_images = sorted(glob(os.path.join(opt.images_folder, 'val', 'image*.nii')))
val_segs = sorted(glob(os.path.join(opt.labels_folder, 'val', 'label*.nii')))
# test images
test_images = sorted(glob(os.path.join(opt.images_folder, 'test', 'image*.nii')))
test_segs = sorted(glob(os.path.join(opt.labels_folder, 'test', 'label*.nii')))
# augment the data list for training
for i in range(int(opt.increase_factor_data)):
train_images.extend(train_images)
train_segs.extend(train_segs)
print('Number of training patches per epoch:', len(train_images))
print('Number of training images per epoch:', len(train_images_for_dice))
print('Number of validation images per epoch:', len(val_images))
print('Number of test images per epoch:', len(test_images))
# Creation of data directories for data_loader
train_dicts = [{'image': image_name, 'label': label_name}
for image_name, label_name in zip(train_images, train_segs)]
train_dice_dicts = [{'image': image_name, 'label': label_name}
for image_name, label_name in zip(train_images_for_dice, train_segs_for_dice)]
val_dicts = [{'image': image_name, 'label': label_name}
for image_name, label_name in zip(val_images, val_segs)]
test_dicts = [{'image': image_name, 'label': label_name}
for image_name, label_name in zip(test_images, test_segs)]
# Transforms list
# Need to concatenate multiple channels here if you want multichannel segmentation
# Check other examples on Monai webpage.
if opt.resolution is not None:
train_transforms = [
LoadImaged(keys=['image', 'label']),
AddChanneld(keys=['image', 'label']),
NormalizeIntensityd(keys=['image']),
ScaleIntensityd(keys=['image']),
Spacingd(keys=['image', 'label'], pixdim=opt.resolution, mode=('bilinear', 'nearest')),
RandFlipd(keys=['image', 'label'], prob=0.1, spatial_axis=1),
RandFlipd(keys=['image', 'label'], prob=0.1, spatial_axis=0),
RandFlipd(keys=['image', 'label'], prob=0.1, spatial_axis=2),
RandAffined(keys=['image', 'label'], mode=('bilinear', 'nearest'), prob=0.1,
rotate_range=(np.pi / 36, np.pi / 36, np.pi * 2), padding_mode="zeros"),
RandAffined(keys=['image', 'label'], mode=('bilinear', 'nearest'), prob=0.1,
rotate_range=(np.pi / 36, np.pi / 2, np.pi / 36), padding_mode="zeros"),
RandAffined(keys=['image', 'label'], mode=('bilinear', 'nearest'), prob=0.1,
rotate_range=(np.pi / 2, np.pi / 36, np.pi / 36), padding_mode="zeros"),
Rand3DElasticd(keys=['image', 'label'], mode=('bilinear', 'nearest'), prob=0.1,
sigma_range=(5, 8), magnitude_range=(100, 200), scale_range=(0.15, 0.15, 0.15),
padding_mode="zeros"),
RandAdjustContrastd(keys=['image'], gamma=(0.5, 2.5), prob=0.1),
RandGaussianNoised(keys=['image'], prob=0.1, mean=np.random.uniform(0, 0.5), std=np.random.uniform(0, 1)),
RandShiftIntensityd(keys=['image'], offsets=np.random.uniform(0,0.3), prob=0.1),
RandSpatialCropd(keys=['image', 'label'], roi_size=opt.patch_size, random_size=False),
ToTensord(keys=['image', 'label'])
]
val_transforms = [
LoadImaged(keys=['image', 'label']),
AddChanneld(keys=['image', 'label']),
NormalizeIntensityd(keys=['image']),
ScaleIntensityd(keys=['image']),
Spacingd(keys=['image', 'label'], pixdim=opt.resolution, mode=('bilinear', 'nearest')),
ToTensord(keys=['image', 'label'])
]
else:
train_transforms = [
LoadImaged(keys=['image', 'label']),
AddChanneld(keys=['image', 'label']),
NormalizeIntensityd(keys=['image']),
ScaleIntensityd(keys=['image']),
RandFlipd(keys=['image', 'label'], prob=0.1, spatial_axis=1),
RandFlipd(keys=['image', 'label'], prob=0.1, spatial_axis=0),
RandFlipd(keys=['image', 'label'], prob=0.1, spatial_axis=2),
RandAffined(keys=['image', 'label'], mode=('bilinear', 'nearest'), prob=0.1,
rotate_range=(np.pi / 36, np.pi / 36, np.pi * 2), padding_mode="zeros"),
RandAffined(keys=['image', 'label'], mode=('bilinear', 'nearest'), prob=0.1,
rotate_range=(np.pi / 36, np.pi / 2, np.pi / 36), padding_mode="zeros"),
RandAffined(keys=['image', 'label'], mode=('bilinear', 'nearest'), prob=0.1,
rotate_range=(np.pi / 2, np.pi / 36, np.pi / 36), padding_mode="zeros"),
Rand3DElasticd(keys=['image', 'label'], mode=('bilinear', 'nearest'), prob=0.1,
sigma_range=(5, 8), magnitude_range=(100, 200), scale_range=(0.15, 0.15, 0.15), padding_mode="zeros"),
RandAdjustContrastd(keys=['image'], gamma=(0.5, 2.5), prob=0.1),
RandGaussianNoised(keys=['image'], prob=0.1, mean=np.random.uniform(0, 0.5), std=np.random.uniform(0, 1)),
RandShiftIntensityd(keys=['image'], offsets=np.random.uniform(0,0.3), prob=0.1),
RandSpatialCropd(keys=['image', 'label'], roi_size=opt.patch_size, random_size=False),
ToTensord(keys=['image', 'label'])
]
val_transforms = [
LoadImaged(keys=['image', 'label']),
AddChanneld(keys=['image', 'label']),
NormalizeIntensityd(keys=['image']),
ScaleIntensityd(keys=['image']),
ToTensord(keys=['image', 'label'])
]
train_transforms = Compose(train_transforms)
val_transforms = Compose(val_transforms)
# create a training data loader
check_train = monai.data.Dataset(data=train_dicts, transform=train_transforms)
train_loader = DataLoader(check_train, batch_size=opt.batch_size, shuffle=True, num_workers=opt.workers, pin_memory=torch.cuda.is_available())
# create a training_dice data loader
check_val = monai.data.Dataset(data=train_dice_dicts, transform=val_transforms)
train_dice_loader = DataLoader(check_val, batch_size=1, num_workers=opt.workers, pin_memory=torch.cuda.is_available())
# create a validation data loader
check_val = monai.data.Dataset(data=val_dicts, transform=val_transforms)
val_loader = DataLoader(check_val, batch_size=1, num_workers=opt.workers, pin_memory=torch.cuda.is_available())
# create a validation data loader
check_val = monai.data.Dataset(data=test_dicts, transform=val_transforms)
test_loader = DataLoader(check_val, batch_size=1, num_workers=opt.workers, pin_memory=torch.cuda.is_available())
# # try to use all the available GPUs
# devices = get_devices_spec(None)
# build the network
net = build_net()
net.cuda()
if num_gpus > 1:
net = torch.nn.DataParallel(net)
if opt.preload is not None:
net.load_state_dict(torch.load(opt.preload))
dice_metric = DiceMetric(include_background=True, reduction="mean")
post_trans = Compose([Activations(sigmoid=True), AsDiscrete(threshold_values=True)])
# loss_function = monai.losses.DiceLoss(sigmoid=True)
# loss_function = monai.losses.TverskyLoss(sigmoid=True, alpha=0.3, beta=0.7)
loss_function = monai.losses.DiceCELoss(sigmoid=True)
optim = torch.optim.Adam(net.parameters(), lr=opt.lr)
net_scheduler = get_scheduler(optim, opt)
# start a typical PyTorch training
val_interval = 1
best_metric = -1
best_metric_epoch = -1
epoch_loss_values = list()
metric_values = list()
writer = SummaryWriter()
for epoch in range(opt.epochs):
print("-" * 10)
print(f"epoch {epoch + 1}/{opt.epochs}")
net.train()
epoch_loss = 0
step = 0
for batch_data in train_loader:
step += 1
inputs, labels = batch_data["image"].cuda(), batch_data["label"].cuda()
optim.zero_grad()
outputs = net(inputs)
loss = loss_function(outputs, labels)
loss.backward()
optim.step()
epoch_loss += loss.item()
epoch_len = len(check_train) // train_loader.batch_size
print(f"{step}/{epoch_len}, train_loss: {loss.item():.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:.4f}")
update_learning_rate(net_scheduler, optim)
if (epoch + 1) % val_interval == 0:
net.eval()
with torch.no_grad():
def plot_dice(images_loader):
metric_sum = 0.0
metric_count = 0
val_images = None
val_labels = None
val_outputs = None
for data in images_loader:
val_images, val_labels = data["image"].cuda(), data["label"].cuda()
roi_size = opt.patch_size
sw_batch_size = 4
val_outputs = sliding_window_inference(val_images, roi_size, sw_batch_size, net)
val_outputs = post_trans(val_outputs)
value, _ = dice_metric(y_pred=val_outputs, y=val_labels)
metric_count += len(value)
metric_sum += value.item() * len(value)
metric = metric_sum / metric_count
metric_values.append(metric)
return metric, val_images, val_labels, val_outputs
metric, val_images, val_labels, val_outputs = plot_dice(val_loader)
# Save best model
if metric > best_metric:
best_metric = metric
best_metric_epoch = epoch + 1
torch.save(net.state_dict(), "best_metric_model.pth")
print("saved new best metric model")
metric_train, train_images, train_labels, train_outputs = plot_dice(train_dice_loader)
metric_test, test_images, test_labels, test_outputs = plot_dice(test_loader)
# Logger bar
print(
"current epoch: {} Training dice: {:.4f} Validation dice: {:.4f} Testing dice: {:.4f} Best Validation dice: {:.4f} at epoch {}".format(
epoch + 1, metric_train, metric, metric_test, best_metric, best_metric_epoch
)
)
writer.add_scalar("Mean_epoch_loss", epoch_loss, epoch + 1)
writer.add_scalar("Testing_dice", metric_test, epoch + 1)
writer.add_scalar("Training_dice", metric_train, epoch + 1)
writer.add_scalar("Validation_dice", metric, epoch + 1)
# plot the last model output as GIF image in TensorBoard with the corresponding image and label
val_outputs = (val_outputs.sigmoid() >= 0.5).float()
plot_2d_or_3d_image(val_images, epoch + 1, writer, index=0, tag="validation image")
plot_2d_or_3d_image(val_labels, epoch + 1, writer, index=0, tag="validation label")
plot_2d_or_3d_image(val_outputs, epoch + 1, writer, index=0, tag="validation inference")
print(f"train completed, best_metric: {best_metric:.4f} at epoch: {best_metric_epoch}")
writer.close()
def get_task_transforms(mode, task_id, pos_sample_num, neg_sample_num,
num_samples):
if mode != "test":
keys = ["image", "label"]
else:
keys = ["image"]
load_transforms = [
LoadImaged(keys=keys),
EnsureChannelFirstd(keys=keys),
]
# 2. sampling
sample_transforms = [
PreprocessAnisotropic(
keys=keys,
clip_values=clip_values[task_id],
pixdim=spacing[task_id],
normalize_values=normalize_values[task_id],
model_mode=mode,
),
]
# 3. spatial transforms
if mode == "train":
other_transforms = [
SpatialPadd(keys=["image", "label"],
spatial_size=patch_size[task_id]),
RandCropByPosNegLabeld(
keys=["image", "label"],
label_key="label",
spatial_size=patch_size[task_id],
pos=pos_sample_num,
neg=neg_sample_num,
num_samples=num_samples,
image_key="image",
image_threshold=0,
),
RandZoomd(
keys=["image", "label"],
min_zoom=0.9,
max_zoom=1.2,
mode=("trilinear", "nearest"),
align_corners=(True, None),
prob=0.15,
),
RandGaussianNoised(keys=["image"], std=0.01, prob=0.15),
RandGaussianSmoothd(
keys=["image"],
sigma_x=(0.5, 1.15),
sigma_y=(0.5, 1.15),
sigma_z=(0.5, 1.15),
prob=0.15,
),
RandScaleIntensityd(keys=["image"], factors=0.3, prob=0.15),
RandFlipd(["image", "label"], spatial_axis=[0], prob=0.5),
RandFlipd(["image", "label"], spatial_axis=[1], prob=0.5),
RandFlipd(["image", "label"], spatial_axis=[2], prob=0.5),
CastToTyped(keys=["image", "label"], dtype=(np.float32, np.uint8)),
EnsureTyped(keys=["image", "label"]),
]
elif mode == "validation":
other_transforms = [
CastToTyped(keys=["image", "label"], dtype=(np.float32, np.uint8)),
EnsureTyped(keys=["image", "label"]),
]
else:
other_transforms = [
CastToTyped(keys=["image"], dtype=(np.float32)),
EnsureTyped(keys=["image"]),
]
all_transforms = load_transforms + sample_transforms + other_transforms
return Compose(all_transforms)
def test_train_timing(self):
images = sorted(glob(os.path.join(self.data_dir, "img*.nii.gz")))
segs = sorted(glob(os.path.join(self.data_dir, "seg*.nii.gz")))
train_files = [{
"image": img,
"label": seg
} for img, seg in zip(images[:32], segs[:32])]
val_files = [{
"image": img,
"label": seg
} for img, seg in zip(images[-9:], segs[-9:])]
device = torch.device("cuda:0")
# define transforms for train and validation
train_transforms = Compose([
LoadImaged(keys=["image", "label"]),
EnsureChannelFirstd(keys=["image", "label"]),
Spacingd(keys=["image", "label"],
pixdim=(1.0, 1.0, 1.0),
mode=("bilinear", "nearest")),
ScaleIntensityd(keys="image"),
CropForegroundd(keys=["image", "label"], source_key="image"),
# pre-compute foreground and background indexes
# and cache them to accelerate training
FgBgToIndicesd(keys="label", fg_postfix="_fg", bg_postfix="_bg"),
# change to execute transforms with Tensor data
EnsureTyped(keys=["image", "label"]),
# move the data to GPU and cache to avoid CPU -> GPU sync in every epoch
ToDeviced(keys=["image", "label"], device=device),
# randomly crop out patch samples from big
# image based on pos / neg ratio
# the image centers of negative samples
# must be in valid image area
RandCropByPosNegLabeld(
keys=["image", "label"],
label_key="label",
spatial_size=(64, 64, 64),
pos=1,
neg=1,
num_samples=4,
fg_indices_key="label_fg",
bg_indices_key="label_bg",
),
RandFlipd(keys=["image", "label"], prob=0.5, spatial_axis=[1, 2]),
RandAxisFlipd(keys=["image", "label"], prob=0.5),
RandRotate90d(keys=["image", "label"],
prob=0.5,
spatial_axes=(1, 2)),
RandZoomd(keys=["image", "label"],
prob=0.5,
min_zoom=0.8,
max_zoom=1.2,
keep_size=True),
RandRotated(
keys=["image", "label"],
prob=0.5,
range_x=np.pi / 4,
mode=("bilinear", "nearest"),
align_corners=True,
dtype=np.float64,
),
RandAffined(keys=["image", "label"],
prob=0.5,
rotate_range=np.pi / 2,
mode=("bilinear", "nearest")),
RandGaussianNoised(keys="image", prob=0.5),
RandStdShiftIntensityd(keys="image",
prob=0.5,
factors=0.05,
nonzero=True),
])
val_transforms = Compose([
LoadImaged(keys=["image", "label"]),
EnsureChannelFirstd(keys=["image", "label"]),
Spacingd(keys=["image", "label"],
pixdim=(1.0, 1.0, 1.0),
mode=("bilinear", "nearest")),
ScaleIntensityd(keys="image"),
CropForegroundd(keys=["image", "label"], source_key="image"),
EnsureTyped(keys=["image", "label"]),
# move the data to GPU and cache to avoid CPU -> GPU sync in every epoch
ToDeviced(keys=["image", "label"], device=device),
])
max_epochs = 5
learning_rate = 2e-4
val_interval = 1 # do validation for every epoch
# set CacheDataset, ThreadDataLoader and DiceCE loss for MONAI fast training
train_ds = CacheDataset(data=train_files,
transform=train_transforms,
cache_rate=1.0,
num_workers=8)
val_ds = CacheDataset(data=val_files,
transform=val_transforms,
cache_rate=1.0,
num_workers=5)
# disable multi-workers because `ThreadDataLoader` works with multi-threads
train_loader = ThreadDataLoader(train_ds,
num_workers=0,
batch_size=4,
shuffle=True)
val_loader = ThreadDataLoader(val_ds, num_workers=0, batch_size=1)
loss_function = DiceCELoss(to_onehot_y=True,
softmax=True,
squared_pred=True,
batch=True)
model = UNet(
spatial_dims=3,
in_channels=1,
out_channels=2,
channels=(16, 32, 64, 128, 256),
strides=(2, 2, 2, 2),
num_res_units=2,
norm=Norm.BATCH,
).to(device)
# Novograd paper suggests to use a bigger LR than Adam,
# because Adam does normalization by element-wise second moments
optimizer = Novograd(model.parameters(), learning_rate * 10)
scaler = torch.cuda.amp.GradScaler()
post_pred = Compose(
[EnsureType(), AsDiscrete(argmax=True, to_onehot=2)])
post_label = Compose([EnsureType(), AsDiscrete(to_onehot=2)])
dice_metric = DiceMetric(include_background=True,
reduction="mean",
get_not_nans=False)
best_metric = -1
total_start = time.time()
for epoch in range(max_epochs):
epoch_start = time.time()
print("-" * 10)
print(f"epoch {epoch + 1}/{max_epochs}")
model.train()
epoch_loss = 0
step = 0
for batch_data in train_loader:
step_start = time.time()
step += 1
optimizer.zero_grad()
# set AMP for training
with torch.cuda.amp.autocast():
outputs = model(batch_data["image"])
loss = loss_function(outputs, batch_data["label"])
scaler.scale(loss).backward()
scaler.step(optimizer)
scaler.update()
epoch_loss += loss.item()
epoch_len = math.ceil(len(train_ds) / train_loader.batch_size)
print(f"{step}/{epoch_len}, train_loss: {loss.item():.4f}"
f" step time: {(time.time() - step_start):.4f}")
epoch_loss /= step
print(f"epoch {epoch + 1} average loss: {epoch_loss:.4f}")
if (epoch + 1) % val_interval == 0:
model.eval()
with torch.no_grad():
for val_data in val_loader:
roi_size = (96, 96, 96)
sw_batch_size = 4
# set AMP for validation
with torch.cuda.amp.autocast():
val_outputs = sliding_window_inference(
val_data["image"], roi_size, sw_batch_size,
model)
val_outputs = [
post_pred(i) for i in decollate_batch(val_outputs)
]
val_labels = [
post_label(i)
for i in decollate_batch(val_data["label"])
]
dice_metric(y_pred=val_outputs, y=val_labels)
metric = dice_metric.aggregate().item()
dice_metric.reset()
if metric > best_metric:
best_metric = metric
print(
f"epoch: {epoch + 1} current mean dice: {metric:.4f}, best mean dice: {best_metric:.4f}"
)
print(
f"time consuming of epoch {epoch + 1} is: {(time.time() - epoch_start):.4f}"
)
total_time = time.time() - total_start
print(
f"train completed, best_metric: {best_metric:.4f} total time: {total_time:.4f}"
)
# test expected metrics
self.assertGreater(best_metric, 0.95)
def main():
opt = Options().parse()
# monai.config.print_config()
logging.basicConfig(stream=sys.stdout, level=logging.INFO)
set_determinism(seed=0)
logging.basicConfig(stream=sys.stdout, level=logging.INFO)
device = torch.device(opt.gpu_id)
# ------- Data loader creation ----------
# images
images = sorted(glob(os.path.join(opt.images_folder, 'image*.nii')))
segs = sorted(glob(os.path.join(opt.labels_folder, 'label*.nii')))
train_files = []
val_files = []
for i in range(opt.models_ensemble):
train_files.append([{
"image": img,
"label": seg
} for img, seg in zip(
images[:(opt.split_val * i)] +
images[(opt.split_val *
(i + 1)):(len(images) -
opt.split_val)], segs[:(opt.split_val * i)] +
segs[(opt.split_val * (i + 1)):(len(images) - opt.split_val)])])
val_files.append([{
"image": img,
"label": seg
} for img, seg in zip(
images[(opt.split_val * i):(opt.split_val *
(i + 1))], segs[(opt.split_val *
i):(opt.split_val *
(i + 1))])])
test_files = [{
"image": img,
"label": seg
} for img, seg in zip(images[(len(images) -
opt.split_test):len(images)], segs[(
len(images) -
opt.split_test):len(images)])]
# ----------- Transforms list --------------
if opt.resolution is not None:
train_transforms = [
LoadImaged(keys=['image', 'label']),
AddChanneld(keys=['image', 'label']),
NormalizeIntensityd(keys=['image']),
ScaleIntensityd(keys=['image']),
Spacingd(keys=['image', 'label'],
pixdim=opt.resolution,
mode=('bilinear', 'nearest')),
RandFlipd(keys=['image', 'label'], prob=0.1, spatial_axis=1),
RandFlipd(keys=['image', 'label'], prob=0.1, spatial_axis=0),
RandFlipd(keys=['image', 'label'], prob=0.1, spatial_axis=2),
RandAffined(keys=['image', 'label'],
mode=('bilinear', 'nearest'),
prob=0.1,
rotate_range=(np.pi / 36, np.pi / 36, np.pi * 2),
padding_mode="zeros"),
RandAffined(keys=['image', 'label'],
mode=('bilinear', 'nearest'),
prob=0.1,
rotate_range=(np.pi / 36, np.pi / 2, np.pi / 36),
padding_mode="zeros"),
RandAffined(keys=['image', 'label'],
mode=('bilinear', 'nearest'),
prob=0.1,
rotate_range=(np.pi / 2, np.pi / 36, np.pi / 36),
padding_mode="zeros"),
Rand3DElasticd(keys=['image', 'label'],
mode=('bilinear', 'nearest'),
prob=0.1,
sigma_range=(5, 8),
magnitude_range=(100, 200),
scale_range=(0.15, 0.15, 0.15),
padding_mode="zeros"),
RandAdjustContrastd(keys=['image'], gamma=(0.5, 2.5), prob=0.1),
RandGaussianNoised(keys=['image'],
prob=0.1,
mean=np.random.uniform(0, 0.5),
std=np.random.uniform(0, 1)),
RandShiftIntensityd(keys=['image'],
offsets=np.random.uniform(0, 0.3),
prob=0.1),
RandSpatialCropd(keys=['image', 'label'],
roi_size=opt.patch_size,
random_size=False),
ToTensord(keys=['image', 'label'])
]
val_transforms = [
LoadImaged(keys=['image', 'label']),
AddChanneld(keys=['image', 'label']),
NormalizeIntensityd(keys=['image']),
ScaleIntensityd(keys=['image']),
Spacingd(keys=['image', 'label'],
pixdim=opt.resolution,
mode=('bilinear', 'nearest')),
ToTensord(keys=['image', 'label'])
]
else:
train_transforms = [
LoadImaged(keys=['image', 'label']),
AddChanneld(keys=['image', 'label']),
NormalizeIntensityd(keys=['image']),
ScaleIntensityd(keys=['image']),
RandFlipd(keys=['image', 'label'], prob=0.1, spatial_axis=1),
RandFlipd(keys=['image', 'label'], prob=0.1, spatial_axis=0),
RandFlipd(keys=['image', 'label'], prob=0.1, spatial_axis=2),
RandAffined(keys=['image', 'label'],
mode=('bilinear', 'nearest'),
prob=0.1,
rotate_range=(np.pi / 36, np.pi / 36, np.pi * 2),
padding_mode="zeros"),
RandAffined(keys=['image', 'label'],
mode=('bilinear', 'nearest'),
prob=0.1,
rotate_range=(np.pi / 36, np.pi / 2, np.pi / 36),
padding_mode="zeros"),
RandAffined(keys=['image', 'label'],
mode=('bilinear', 'nearest'),
prob=0.1,
rotate_range=(np.pi / 2, np.pi / 36, np.pi / 36),
padding_mode="zeros"),
Rand3DElasticd(keys=['image', 'label'],
mode=('bilinear', 'nearest'),
prob=0.1,
sigma_range=(5, 8),
magnitude_range=(100, 200),
scale_range=(0.15, 0.15, 0.15),
padding_mode="zeros"),
RandAdjustContrastd(keys=['image'], gamma=(0.5, 2.5), prob=0.1),
RandGaussianNoised(keys=['image'],
prob=0.1,
mean=np.random.uniform(0, 0.5),
std=np.random.uniform(0, 1)),
RandShiftIntensityd(keys=['image'],
offsets=np.random.uniform(0, 0.3),
prob=0.1),
RandSpatialCropd(keys=['image', 'label'],
roi_size=opt.patch_size,
random_size=False),
ToTensord(keys=['image', 'label'])
]
val_transforms = [
LoadImaged(keys=['image', 'label']),
AddChanneld(keys=['image', 'label']),
NormalizeIntensityd(keys=['image']),
ScaleIntensityd(keys=['image']),
ToTensord(keys=['image', 'label'])
]
train_transforms = Compose(train_transforms)
val_transforms = Compose(val_transforms)
# ---------- Creation of DataLoaders -------------
train_dss = [
CacheDataset(data=train_files[i], transform=train_transforms)
for i in range(opt.models_ensemble)
]
train_loaders = [
DataLoader(train_dss[i],
batch_size=opt.batch_size,
shuffle=True,
num_workers=opt.workers,
pin_memory=torch.cuda.is_available())
for i in range(opt.models_ensemble)
]
val_dss = [
CacheDataset(data=val_files[i], transform=val_transforms)
for i in range(opt.models_ensemble)
]
val_loaders = [
DataLoader(val_dss[i],
batch_size=1,
num_workers=opt.workers,
pin_memory=torch.cuda.is_available())
for i in range(opt.models_ensemble)
]
test_ds = CacheDataset(data=test_files, transform=val_transforms)
test_loader = DataLoader(test_ds,
batch_size=1,
num_workers=opt.workers,
pin_memory=torch.cuda.is_available())
def train(index):
# ---------- Build the nn-Unet network ------------
if opt.resolution is None:
sizes, spacings = opt.patch_size, opt.spacing
else:
sizes, spacings = opt.patch_size, opt.resolution
strides, kernels = [], []
while True:
spacing_ratio = [sp / min(spacings) for sp in spacings]
stride = [
2 if ratio <= 2 and size >= 8 else 1
for (ratio, size) in zip(spacing_ratio, sizes)
]
kernel = [3 if ratio <= 2 else 1 for ratio in spacing_ratio]
if all(s == 1 for s in stride):
break
sizes = [i / j for i, j in zip(sizes, stride)]
spacings = [i * j for i, j in zip(spacings, stride)]
kernels.append(kernel)
strides.append(stride)
strides.insert(0, len(spacings) * [1])
kernels.append(len(spacings) * [3])
net = monai.networks.nets.DynUNet(
spatial_dims=3,
in_channels=opt.in_channels,
out_channels=opt.out_channels,
kernel_size=kernels,
strides=strides,
upsample_kernel_size=strides[1:],
res_block=True,
# act=act_type,
# norm=Norm.BATCH,
).to(device)
from torch.autograd import Variable
from torchsummaryX import summary
data = Variable(
torch.randn(int(opt.batch_size), int(opt.in_channels),
int(opt.patch_size[0]), int(opt.patch_size[1]),
int(opt.patch_size[2]))).cuda()
out = net(data)
summary(net, data)
print("out size: {}".format(out.size()))
# if opt.preload is not None:
# net.load_state_dict(torch.load(opt.preload))
# ---------- ------------------------ ------------
optim = torch.optim.Adam(net.parameters(), lr=opt.lr)
lr_scheduler = torch.optim.lr_scheduler.LambdaLR(
optim, lr_lambda=lambda epoch: (1 - epoch / opt.epochs)**0.9)
loss_function = monai.losses.DiceCELoss(sigmoid=True)
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),
CheckpointSaver(save_dir="./runs/",
save_dict={"net": net},
save_key_metric=True),
]
evaluator = SupervisedEvaluator(
device=device,
val_data_loader=val_loaders[index],
network=net,
inferer=SlidingWindowInferer(roi_size=opt.patch_size,
sw_batch_size=opt.batch_size,
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"]),
)
},
val_handlers=val_handlers)
train_post_transforms = Compose([
Activationsd(keys="pred", sigmoid=True),
AsDiscreted(keys="pred", threshold_values=True),
# KeepLargestConnectedComponentd(keys="pred", applied_labels=[1]),
])
train_handlers = [
ValidationHandler(validator=evaluator,
interval=5,
epoch_level=True),
LrScheduleHandler(lr_scheduler=lr_scheduler, print_lr=True),
StatsHandler(tag_name="train_loss",
output_transform=lambda x: x["loss"]),
CheckpointSaver(save_dir="./runs/",
save_dict={
"net": net,
"opt": optim
},
save_final=True,
epoch_level=True),
]
trainer = SupervisedTrainer(
device=device,
max_epochs=opt.epochs,
train_data_loader=train_loaders[index],
network=net,
optimizer=optim,
loss_function=loss_function,
inferer=SimpleInferer(),
post_transform=train_post_transforms,
amp=False,
train_handlers=train_handlers,
)
trainer.run()
return net
models = [train(i) for i in range(opt.models_ensemble)]
# -------- Test the models ---------
def ensemble_evaluate(post_transforms, models):
evaluator = EnsembleEvaluator(
device=device,
val_data_loader=test_loader,
pred_keys=opt.pred_keys,
networks=models,
inferer=SlidingWindowInferer(roi_size=opt.patch_size,
sw_batch_size=opt.batch_size,
overlap=0.5),
post_transform=post_transforms,
key_val_metric={
"test_mean_dice":
MeanDice(
include_background=True,
output_transform=lambda x: (x["pred"], x["label"]),
)
},
)
evaluator.run()
mean_post_transforms = Compose([
MeanEnsembled(
keys=opt.pred_keys,
output_key="pred",
# in this particular example, we use validation metrics as weights
weights=opt.weights_models,
),
Activationsd(keys="pred", sigmoid=True),
AsDiscreted(keys="pred", threshold_values=True),
# KeepLargestConnectedComponentd(keys="pred", applied_labels=[1])
])
print('Results from MeanEnsembled:')
ensemble_evaluate(mean_post_transforms, models)
vote_post_transforms = Compose([
Activationsd(keys=opt.pred_keys, sigmoid=True),
# transform data into discrete before voting
AsDiscreted(keys=opt.pred_keys, threshold_values=True),
# KeepLargestConnectedComponentd(keys="pred", applied_labels=[1]),
VoteEnsembled(keys=opt.pred_keys, output_key="pred"),
])
print('Results from VoteEnsembled:')
ensemble_evaluate(vote_post_transforms, models)
def main():
parser = argparse.ArgumentParser(description="training")
parser.add_argument(
"--checkpoint",
type=str,
default=None,
help="checkpoint full path",
)
parser.add_argument(
"--factor_ram_cost",
default=0.0,
type=float,
help="factor to determine RAM cost in the searched architecture",
)
parser.add_argument(
"--fold",
action="store",
required=True,
help="fold index in N-fold cross-validation",
)
parser.add_argument(
"--json",
action="store",
required=True,
help="full path of .json file",
)
parser.add_argument(
"--json_key",
action="store",
required=True,
help="selected key in .json data list",
)
parser.add_argument(
"--local_rank",
required=int,
help="local process rank",
)
parser.add_argument(
"--num_folds",
action="store",
required=True,
help="number of folds in cross-validation",
)
parser.add_argument(
"--output_root",
action="store",
required=True,
help="output root",
)
parser.add_argument(
"--root",
action="store",
required=True,
help="data root",
)
args = parser.parse_args()
logging.basicConfig(stream=sys.stdout, level=logging.INFO)
if not os.path.exists(args.output_root):
os.makedirs(args.output_root, exist_ok=True)
amp = True
determ = True
factor_ram_cost = args.factor_ram_cost
fold = int(args.fold)
input_channels = 1
learning_rate = 0.025
learning_rate_arch = 0.001
learning_rate_milestones = np.array([0.4, 0.8])
num_images_per_batch = 1
num_epochs = 1430 # around 20k iteration
num_epochs_per_validation = 100
num_epochs_warmup = 715
num_folds = int(args.num_folds)
num_patches_per_image = 1
num_sw_batch_size = 6
output_classes = 3
overlap_ratio = 0.625
patch_size = (96, 96, 96)
patch_size_valid = (96, 96, 96)
spacing = [1.0, 1.0, 1.0]
print("factor_ram_cost", factor_ram_cost)
# deterministic training
if determ:
set_determinism(seed=0)
# initialize the distributed training process, every GPU runs in a process
dist.init_process_group(backend="nccl", init_method="env://")
# dist.barrier()
world_size = dist.get_world_size()
with open(args.json, "r") as f:
json_data = json.load(f)
split = len(json_data[args.json_key]) // num_folds
list_train = json_data[args.json_key][:(
split * fold)] + json_data[args.json_key][(split * (fold + 1)):]
list_valid = json_data[args.json_key][(split * fold):(split * (fold + 1))]
# training data
files = []
for _i in range(len(list_train)):
str_img = os.path.join(args.root, list_train[_i]["image"])
str_seg = os.path.join(args.root, list_train[_i]["label"])
if (not os.path.exists(str_img)) or (not os.path.exists(str_seg)):
continue
files.append({"image": str_img, "label": str_seg})
train_files = files
random.shuffle(train_files)
train_files_w = train_files[:len(train_files) // 2]
train_files_w = partition_dataset(data=train_files_w,
shuffle=True,
num_partitions=world_size,
even_divisible=True)[dist.get_rank()]
print("train_files_w:", len(train_files_w))
train_files_a = train_files[len(train_files) // 2:]
train_files_a = partition_dataset(data=train_files_a,
shuffle=True,
num_partitions=world_size,
even_divisible=True)[dist.get_rank()]
print("train_files_a:", len(train_files_a))
# validation data
files = []
for _i in range(len(list_valid)):
str_img = os.path.join(args.root, list_valid[_i]["image"])
str_seg = os.path.join(args.root, list_valid[_i]["label"])
if (not os.path.exists(str_img)) or (not os.path.exists(str_seg)):
continue
files.append({"image": str_img, "label": str_seg})
val_files = files
val_files = partition_dataset(data=val_files,
shuffle=False,
num_partitions=world_size,
even_divisible=False)[dist.get_rank()]
print("val_files:", len(val_files))
# network architecture
device = torch.device(f"cuda:{args.local_rank}")
torch.cuda.set_device(device)
train_transforms = Compose([
LoadImaged(keys=["image", "label"]),
EnsureChannelFirstd(keys=["image", "label"]),
Orientationd(keys=["image", "label"], axcodes="RAS"),
Spacingd(keys=["image", "label"],
pixdim=spacing,
mode=("bilinear", "nearest"),
align_corners=(True, True)),
CastToTyped(keys=["image"], dtype=(torch.float32)),
ScaleIntensityRanged(keys=["image"],
a_min=-87.0,
a_max=199.0,
b_min=0.0,
b_max=1.0,
clip=True),
CastToTyped(keys=["image", "label"], dtype=(np.float16, np.uint8)),
CopyItemsd(keys=["label"], times=1, names=["label4crop"]),
Lambdad(
keys=["label4crop"],
func=lambda x: np.concatenate(tuple([
ndimage.binary_dilation(
(x == _k).astype(x.dtype), iterations=48).astype(x.dtype)
for _k in range(output_classes)
]),
axis=0),
overwrite=True,
),
EnsureTyped(keys=["image", "label"]),
CastToTyped(keys=["image"], dtype=(torch.float32)),
SpatialPadd(keys=["image", "label", "label4crop"],
spatial_size=patch_size,
mode=["reflect", "constant", "constant"]),
RandCropByLabelClassesd(keys=["image", "label"],
label_key="label4crop",
num_classes=output_classes,
ratios=[
1,
] * output_classes,
spatial_size=patch_size,
num_samples=num_patches_per_image),
Lambdad(keys=["label4crop"], func=lambda x: 0),
RandRotated(keys=["image", "label"],
range_x=0.3,
range_y=0.3,
range_z=0.3,
mode=["bilinear", "nearest"],
prob=0.2),
RandZoomd(keys=["image", "label"],
min_zoom=0.8,
max_zoom=1.2,
mode=["trilinear", "nearest"],
align_corners=[True, None],
prob=0.16),
RandGaussianSmoothd(keys=["image"],
sigma_x=(0.5, 1.15),
sigma_y=(0.5, 1.15),
sigma_z=(0.5, 1.15),
prob=0.15),
RandScaleIntensityd(keys=["image"], factors=0.3, prob=0.5),
RandShiftIntensityd(keys=["image"], offsets=0.1, prob=0.5),
RandGaussianNoised(keys=["image"], std=0.01, prob=0.15),
RandFlipd(keys=["image", "label"], spatial_axis=0, prob=0.5),
RandFlipd(keys=["image", "label"], spatial_axis=1, prob=0.5),
RandFlipd(keys=["image", "label"], spatial_axis=2, prob=0.5),
CastToTyped(keys=["image", "label"],
dtype=(torch.float32, torch.uint8)),
ToTensord(keys=["image", "label"]),
])
val_transforms = Compose([
LoadImaged(keys=["image", "label"]),
EnsureChannelFirstd(keys=["image", "label"]),
Orientationd(keys=["image", "label"], axcodes="RAS"),
Spacingd(keys=["image", "label"],
pixdim=spacing,
mode=("bilinear", "nearest"),
align_corners=(True, True)),
CastToTyped(keys=["image"], dtype=(torch.float32)),
ScaleIntensityRanged(keys=["image"],
a_min=-87.0,
a_max=199.0,
b_min=0.0,
b_max=1.0,
clip=True),
CastToTyped(keys=["image", "label"], dtype=(np.float32, np.uint8)),
EnsureTyped(keys=["image", "label"]),
ToTensord(keys=["image", "label"])
])
train_ds_a = monai.data.CacheDataset(data=train_files_a,
transform=train_transforms,
cache_rate=1.0,
num_workers=8)
train_ds_w = monai.data.CacheDataset(data=train_files_w,
transform=train_transforms,
cache_rate=1.0,
num_workers=8)
val_ds = monai.data.CacheDataset(data=val_files,
transform=val_transforms,
cache_rate=1.0,
num_workers=2)
# monai.data.Dataset can be used as alternatives when debugging or RAM space is limited.
# train_ds_a = monai.data.Dataset(data=train_files_a, transform=train_transforms)
# train_ds_w = monai.data.Dataset(data=train_files_w, transform=train_transforms)
# val_ds = monai.data.Dataset(data=val_files, transform=val_transforms)
train_loader_a = ThreadDataLoader(train_ds_a,
num_workers=0,
batch_size=num_images_per_batch,
shuffle=True)
train_loader_w = ThreadDataLoader(train_ds_w,
num_workers=0,
batch_size=num_images_per_batch,
shuffle=True)
val_loader = ThreadDataLoader(val_ds,
num_workers=0,
batch_size=1,
shuffle=False)
# DataLoader can be used as alternatives when ThreadDataLoader is less efficient.
# train_loader_a = DataLoader(train_ds_a, batch_size=num_images_per_batch, shuffle=True, num_workers=2, pin_memory=torch.cuda.is_available())
# train_loader_w = DataLoader(train_ds_w, batch_size=num_images_per_batch, shuffle=True, num_workers=2, pin_memory=torch.cuda.is_available())
# val_loader = DataLoader(val_ds, batch_size=1, shuffle=False, num_workers=2, pin_memory=torch.cuda.is_available())
dints_space = monai.networks.nets.TopologySearch(
channel_mul=0.5,
num_blocks=12,
num_depths=4,
use_downsample=True,
device=device,
)
model = monai.networks.nets.DiNTS(
dints_space=dints_space,
in_channels=input_channels,
num_classes=output_classes,
use_downsample=True,
)
model = model.to(device)
model = torch.nn.SyncBatchNorm.convert_sync_batchnorm(model)
post_pred = Compose(
[EnsureType(),
AsDiscrete(argmax=True, to_onehot=output_classes)])
post_label = Compose([EnsureType(), AsDiscrete(to_onehot=output_classes)])
# loss function
loss_func = monai.losses.DiceCELoss(
include_background=False,
to_onehot_y=True,
softmax=True,
squared_pred=True,
batch=True,
smooth_nr=0.00001,
smooth_dr=0.00001,
)
# optimizer
optimizer = torch.optim.SGD(model.weight_parameters(),
lr=learning_rate * world_size,
momentum=0.9,
weight_decay=0.00004)
arch_optimizer_a = torch.optim.Adam([dints_space.log_alpha_a],
lr=learning_rate_arch * world_size,
betas=(0.5, 0.999),
weight_decay=0.0)
arch_optimizer_c = torch.optim.Adam([dints_space.log_alpha_c],
lr=learning_rate_arch * world_size,
betas=(0.5, 0.999),
weight_decay=0.0)
print()
if torch.cuda.device_count() > 1:
if dist.get_rank() == 0:
print("Let's use", torch.cuda.device_count(), "GPUs!")
model = DistributedDataParallel(model,
device_ids=[device],
find_unused_parameters=True)
if args.checkpoint != None and os.path.isfile(args.checkpoint):
print("[info] fine-tuning pre-trained checkpoint {0:s}".format(
args.checkpoint))
model.load_state_dict(torch.load(args.checkpoint, map_location=device))
torch.cuda.empty_cache()
else:
print("[info] training from scratch")
# amp
if amp:
from torch.cuda.amp import autocast, GradScaler
scaler = GradScaler()
if dist.get_rank() == 0:
print("[info] amp enabled")
# start a typical PyTorch training
val_interval = num_epochs_per_validation
best_metric = -1
best_metric_epoch = -1
epoch_loss_values = list()
idx_iter = 0
metric_values = list()
if dist.get_rank() == 0:
writer = SummaryWriter(
log_dir=os.path.join(args.output_root, "Events"))
with open(os.path.join(args.output_root, "accuracy_history.csv"),
"a") as f:
f.write("epoch\tmetric\tloss\tlr\ttime\titer\n")
dataloader_a_iterator = iter(train_loader_a)
start_time = time.time()
for epoch in range(num_epochs):
decay = 0.5**np.sum([
(epoch - num_epochs_warmup) /
(num_epochs - num_epochs_warmup) > learning_rate_milestones
])
lr = learning_rate * decay
for param_group in optimizer.param_groups:
param_group["lr"] = lr
if dist.get_rank() == 0:
print("-" * 10)
print(f"epoch {epoch + 1}/{num_epochs}")
print("learning rate is set to {}".format(lr))
model.train()
epoch_loss = 0
loss_torch = torch.zeros(2, dtype=torch.float, device=device)
epoch_loss_arch = 0
loss_torch_arch = torch.zeros(2, dtype=torch.float, device=device)
step = 0
for batch_data in train_loader_w:
step += 1
inputs, labels = batch_data["image"].to(
device), batch_data["label"].to(device)
if world_size == 1:
for _ in model.weight_parameters():
_.requires_grad = True
else:
for _ in model.module.weight_parameters():
_.requires_grad = True
dints_space.log_alpha_a.requires_grad = False
dints_space.log_alpha_c.requires_grad = False
optimizer.zero_grad()
if amp:
with autocast():
outputs = model(inputs)
if output_classes == 2:
loss = loss_func(torch.flip(outputs, dims=[1]),
1 - labels)
else:
loss = loss_func(outputs, labels)
scaler.scale(loss).backward()
scaler.step(optimizer)
scaler.update()
else:
outputs = model(inputs)
if output_classes == 2:
loss = loss_func(torch.flip(outputs, dims=[1]), 1 - labels)
else:
loss = loss_func(outputs, labels)
loss.backward()
optimizer.step()
epoch_loss += loss.item()
loss_torch[0] += loss.item()
loss_torch[1] += 1.0
epoch_len = len(train_loader_w)
idx_iter += 1
if dist.get_rank() == 0:
print("[{0}] ".format(str(datetime.now())[:19]) +
f"{step}/{epoch_len}, train_loss: {loss.item():.4f}")
writer.add_scalar("train_loss", loss.item(),
epoch_len * epoch + step)
if epoch < num_epochs_warmup:
continue
try:
sample_a = next(dataloader_a_iterator)
except StopIteration:
dataloader_a_iterator = iter(train_loader_a)
sample_a = next(dataloader_a_iterator)
inputs_search, labels_search = sample_a["image"].to(
device), sample_a["label"].to(device)
if world_size == 1:
for _ in model.weight_parameters():
_.requires_grad = False
else:
for _ in model.module.weight_parameters():
_.requires_grad = False
dints_space.log_alpha_a.requires_grad = True
dints_space.log_alpha_c.requires_grad = True
# linear increase topology and RAM loss
entropy_alpha_c = torch.tensor(0.).to(device)
entropy_alpha_a = torch.tensor(0.).to(device)
ram_cost_full = torch.tensor(0.).to(device)
ram_cost_usage = torch.tensor(0.).to(device)
ram_cost_loss = torch.tensor(0.).to(device)
topology_loss = torch.tensor(0.).to(device)
probs_a, arch_code_prob_a = dints_space.get_prob_a(child=True)
entropy_alpha_a = -((probs_a) * torch.log(probs_a + 1e-5)).mean()
entropy_alpha_c = -(F.softmax(dints_space.log_alpha_c, dim=-1) * \
F.log_softmax(dints_space.log_alpha_c, dim=-1)).mean()
topology_loss = dints_space.get_topology_entropy(probs_a)
ram_cost_full = dints_space.get_ram_cost_usage(inputs.shape,
full=True)
ram_cost_usage = dints_space.get_ram_cost_usage(inputs.shape)
ram_cost_loss = torch.abs(factor_ram_cost -
ram_cost_usage / ram_cost_full)
arch_optimizer_a.zero_grad()
arch_optimizer_c.zero_grad()
combination_weights = (epoch - num_epochs_warmup) / (
num_epochs - num_epochs_warmup)
if amp:
with autocast():
outputs_search = model(inputs_search)
if output_classes == 2:
loss = loss_func(torch.flip(outputs_search, dims=[1]),
1 - labels_search)
else:
loss = loss_func(outputs_search, labels_search)
loss += combination_weights * ((entropy_alpha_a + entropy_alpha_c) + ram_cost_loss \
+ 0.001 * topology_loss)
scaler.scale(loss).backward()
scaler.step(arch_optimizer_a)
scaler.step(arch_optimizer_c)
scaler.update()
else:
outputs_search = model(inputs_search)
if output_classes == 2:
loss = loss_func(torch.flip(outputs_search, dims=[1]),
1 - labels_search)
else:
loss = loss_func(outputs_search, labels_search)
loss += 1.0 * (combination_weights * (entropy_alpha_a + entropy_alpha_c) + ram_cost_loss \
+ 0.001 * topology_loss)
loss.backward()
arch_optimizer_a.step()
arch_optimizer_c.step()
epoch_loss_arch += loss.item()
loss_torch_arch[0] += loss.item()
loss_torch_arch[1] += 1.0
if dist.get_rank() == 0:
print(
"[{0}] ".format(str(datetime.now())[:19]) +
f"{step}/{epoch_len}, train_loss_arch: {loss.item():.4f}")
writer.add_scalar("train_loss_arch", loss.item(),
epoch_len * epoch + step)
# synchronizes all processes and reduce results
dist.barrier()
dist.all_reduce(loss_torch, op=torch.distributed.ReduceOp.SUM)
loss_torch = loss_torch.tolist()
loss_torch_arch = loss_torch_arch.tolist()
if dist.get_rank() == 0:
loss_torch_epoch = loss_torch[0] / loss_torch[1]
print(
f"epoch {epoch + 1} average loss: {loss_torch_epoch:.4f}, best mean dice: {best_metric:.4f} at epoch {best_metric_epoch}"
)
if epoch >= num_epochs_warmup:
loss_torch_arch_epoch = loss_torch_arch[0] / loss_torch_arch[1]
print(
f"epoch {epoch + 1} average arch loss: {loss_torch_arch_epoch:.4f}, best mean dice: {best_metric:.4f} at epoch {best_metric_epoch}"
)
if (epoch + 1) % val_interval == 0:
torch.cuda.empty_cache()
model.eval()
with torch.no_grad():
metric = torch.zeros((output_classes - 1) * 2,
dtype=torch.float,
device=device)
metric_sum = 0.0
metric_count = 0
metric_mat = []
val_images = None
val_labels = None
val_outputs = None
_index = 0
for val_data in val_loader:
val_images = val_data["image"].to(device)
val_labels = val_data["label"].to(device)
roi_size = patch_size_valid
sw_batch_size = num_sw_batch_size
if amp:
with torch.cuda.amp.autocast():
pred = sliding_window_inference(
val_images,
roi_size,
sw_batch_size,
lambda x: model(x),
mode="gaussian",
overlap=overlap_ratio,
)
else:
pred = sliding_window_inference(
val_images,
roi_size,
sw_batch_size,
lambda x: model(x),
mode="gaussian",
overlap=overlap_ratio,
)
val_outputs = pred
val_outputs = post_pred(val_outputs[0, ...])
val_outputs = val_outputs[None, ...]
val_labels = post_label(val_labels[0, ...])
val_labels = val_labels[None, ...]
value = compute_meandice(y_pred=val_outputs,
y=val_labels,
include_background=False)
print(_index + 1, "/", len(val_loader), value)
metric_count += len(value)
metric_sum += value.sum().item()
metric_vals = value.cpu().numpy()
if len(metric_mat) == 0:
metric_mat = metric_vals
else:
metric_mat = np.concatenate((metric_mat, metric_vals),
axis=0)
for _c in range(output_classes - 1):
val0 = torch.nan_to_num(value[0, _c], nan=0.0)
val1 = 1.0 - torch.isnan(value[0, 0]).float()
metric[2 * _c] += val0 * val1
metric[2 * _c + 1] += val1
_index += 1
# synchronizes all processes and reduce results
dist.barrier()
dist.all_reduce(metric, op=torch.distributed.ReduceOp.SUM)
metric = metric.tolist()
if dist.get_rank() == 0:
for _c in range(output_classes - 1):
print(
"evaluation metric - class {0:d}:".format(_c + 1),
metric[2 * _c] / metric[2 * _c + 1])
avg_metric = 0
for _c in range(output_classes - 1):
avg_metric += metric[2 * _c] / metric[2 * _c + 1]
avg_metric = avg_metric / float(output_classes - 1)
print("avg_metric", avg_metric)
if avg_metric > best_metric:
best_metric = avg_metric
best_metric_epoch = epoch + 1
best_metric_iterations = idx_iter
node_a_d, arch_code_a_d, arch_code_c_d, arch_code_a_max_d = dints_space.decode(
)
torch.save(
{
"node_a": node_a_d,
"arch_code_a": arch_code_a_d,
"arch_code_a_max": arch_code_a_max_d,
"arch_code_c": arch_code_c_d,
"iter_num": idx_iter,
"epochs": epoch + 1,
"best_dsc": best_metric,
"best_path": best_metric_iterations,
},
os.path.join(args.output_root,
"search_code_" + str(idx_iter) + ".pth"),
)
print("saved new best metric model")
dict_file = {}
dict_file["best_avg_dice_score"] = float(best_metric)
dict_file["best_avg_dice_score_epoch"] = int(
best_metric_epoch)
dict_file["best_avg_dice_score_iteration"] = int(idx_iter)
with open(os.path.join(args.output_root, "progress.yaml"),
"w") as out_file:
documents = yaml.dump(dict_file, stream=out_file)
print(
"current epoch: {} current mean dice: {:.4f} best mean dice: {:.4f} at epoch {}"
.format(epoch + 1, avg_metric, best_metric,
best_metric_epoch))
current_time = time.time()
elapsed_time = (current_time - start_time) / 60.0
with open(
os.path.join(args.output_root,
"accuracy_history.csv"), "a") as f:
f.write(
"{0:d}\t{1:.5f}\t{2:.5f}\t{3:.5f}\t{4:.1f}\t{5:d}\n"
.format(epoch + 1, avg_metric, loss_torch_epoch,
lr, elapsed_time, idx_iter))
dist.barrier()
torch.cuda.empty_cache()
print(
f"train completed, best_metric: {best_metric:.4f} at epoch: {best_metric_epoch}"
)
if dist.get_rank() == 0:
writer.close()
dist.destroy_process_group()
return
def main():
opt = Options().parse()
# monai.config.print_config()
logging.basicConfig(stream=sys.stdout, level=logging.INFO)
# check gpus
if opt.gpu_ids != '-1':
num_gpus = len(opt.gpu_ids.split(','))
else:
num_gpus = 0
print('number of GPU:', num_gpus)
# Data loader creation
# train images
train_images = sorted(glob(os.path.join(opt.images_folder, 'train', 'image*.nii')))
train_segs = sorted(glob(os.path.join(opt.labels_folder, 'train', 'label*.nii')))
train_images_for_dice = sorted(glob(os.path.join(opt.images_folder, 'train', 'image*.nii')))
train_segs_for_dice = sorted(glob(os.path.join(opt.labels_folder, 'train', 'label*.nii')))
# validation images
val_images = sorted(glob(os.path.join(opt.images_folder, 'val', 'image*.nii')))
val_segs = sorted(glob(os.path.join(opt.labels_folder, 'val', 'label*.nii')))
# test images
test_images = sorted(glob(os.path.join(opt.images_folder, 'test', 'image*.nii')))
test_segs = sorted(glob(os.path.join(opt.labels_folder, 'test', 'label*.nii')))
# augment the data list for training
for i in range(int(opt.increase_factor_data)):
train_images.extend(train_images)
train_segs.extend(train_segs)
print('Number of training patches per epoch:', len(train_images))
print('Number of training images per epoch:', len(train_images_for_dice))
print('Number of validation images per epoch:', len(val_images))
print('Number of test images per epoch:', len(test_images))
# Creation of data directories for data_loader
train_dicts = [{'image': image_name, 'label': label_name}
for image_name, label_name in zip(train_images, train_segs)]
train_dice_dicts = [{'image': image_name, 'label': label_name}
for image_name, label_name in zip(train_images_for_dice, train_segs_for_dice)]
val_dicts = [{'image': image_name, 'label': label_name}
for image_name, label_name in zip(val_images, val_segs)]
test_dicts = [{'image': image_name, 'label': label_name}
for image_name, label_name in zip(test_images, test_segs)]
# Transforms list
if opt.resolution is not None:
train_transforms = [
LoadImaged(keys=['image', 'label']),
AddChanneld(keys=['image', 'label']),
# ThresholdIntensityd(keys=['image'], threshold=-135, above=True, cval=-135), # CT HU filter
# ThresholdIntensityd(keys=['image'], threshold=215, above=False, cval=215),
CropForegroundd(keys=['image', 'label'], source_key='image'), # crop CropForeground
NormalizeIntensityd(keys=['image']), # augmentation
ScaleIntensityd(keys=['image']), # intensity
Spacingd(keys=['image', 'label'], pixdim=opt.resolution, mode=('bilinear', 'nearest')), # resolution
RandFlipd(keys=['image', 'label'], prob=0.15, spatial_axis=1),
RandFlipd(keys=['image', 'label'], prob=0.15, spatial_axis=0),
RandFlipd(keys=['image', 'label'], prob=0.15, spatial_axis=2),
RandAffined(keys=['image', 'label'], mode=('bilinear', 'nearest'), prob=0.1,
rotate_range=(np.pi / 36, np.pi / 36, np.pi * 2), padding_mode="zeros"),
RandAffined(keys=['image', 'label'], mode=('bilinear', 'nearest'), prob=0.1,
rotate_range=(np.pi / 36, np.pi / 2, np.pi / 36), padding_mode="zeros"),
RandAffined(keys=['image', 'label'], mode=('bilinear', 'nearest'), prob=0.1,
rotate_range=(np.pi / 2, np.pi / 36, np.pi / 36), padding_mode="zeros"),
Rand3DElasticd(keys=['image', 'label'], mode=('bilinear', 'nearest'), prob=0.1,
sigma_range=(5, 8), magnitude_range=(100, 200), scale_range=(0.15, 0.15, 0.15),
padding_mode="zeros"),
RandGaussianSmoothd(keys=["image"], sigma_x=(0.5, 1.15), sigma_y=(0.5, 1.15), sigma_z=(0.5, 1.15), prob=0.1,),
RandAdjustContrastd(keys=['image'], gamma=(0.5, 2.5), prob=0.1),
RandGaussianNoised(keys=['image'], prob=0.1, mean=np.random.uniform(0, 0.5), std=np.random.uniform(0, 15)),
RandShiftIntensityd(keys=['image'], offsets=np.random.uniform(0,0.3), prob=0.1),
SpatialPadd(keys=['image', 'label'], spatial_size=opt.patch_size, method= 'end'), # pad if the image is smaller than patch
RandSpatialCropd(keys=['image', 'label'], roi_size=opt.patch_size, random_size=False),
ToTensord(keys=['image', 'label'])
]
val_transforms = [
LoadImaged(keys=['image', 'label']),
AddChanneld(keys=['image', 'label']),
# ThresholdIntensityd(keys=['image'], threshold=-135, above=True, cval=-135),
# ThresholdIntensityd(keys=['image'], threshold=215, above=False, cval=215),
CropForegroundd(keys=['image', 'label'], source_key='image'), # crop CropForeground
NormalizeIntensityd(keys=['image']), # intensity
ScaleIntensityd(keys=['image']),
Spacingd(keys=['image', 'label'], pixdim=opt.resolution, mode=('bilinear', 'nearest')), # resolution
SpatialPadd(keys=['image', 'label'], spatial_size=opt.patch_size, method= 'end'), # pad if the image is smaller than patch
ToTensord(keys=['image', 'label'])
]
else:
train_transforms = [
LoadImaged(keys=['image', 'label']),
AddChanneld(keys=['image', 'label']),
# ThresholdIntensityd(keys=['image'], threshold=-135, above=True, cval=-135),
# ThresholdIntensityd(keys=['image'], threshold=215, above=False, cval=215),
CropForegroundd(keys=['image', 'label'], source_key='image'), # crop CropForeground
NormalizeIntensityd(keys=['image']), # augmentation
ScaleIntensityd(keys=['image']), # intensity
RandFlipd(keys=['image', 'label'], prob=0.15, spatial_axis=1),
RandFlipd(keys=['image', 'label'], prob=0.15, spatial_axis=0),
RandFlipd(keys=['image', 'label'], prob=0.15, spatial_axis=2),
RandAffined(keys=['image', 'label'], mode=('bilinear', 'nearest'), prob=0.1,
rotate_range=(np.pi / 36, np.pi / 36, np.pi * 2), padding_mode="zeros"),
RandAffined(keys=['image', 'label'], mode=('bilinear', 'nearest'), prob=0.1,
rotate_range=(np.pi / 36, np.pi / 2, np.pi / 36), padding_mode="zeros"),
RandAffined(keys=['image', 'label'], mode=('bilinear', 'nearest'), prob=0.1,
rotate_range=(np.pi / 2, np.pi / 36, np.pi / 36), padding_mode="zeros"),
Rand3DElasticd(keys=['image', 'label'], mode=('bilinear', 'nearest'), prob=0.1,
sigma_range=(5, 8), magnitude_range=(100, 200), scale_range=(0.15, 0.15, 0.15),
padding_mode="zeros"),
RandGaussianSmoothd(keys=["image"], sigma_x=(0.5, 1.15), sigma_y=(0.5, 1.15), sigma_z=(0.5, 1.15), prob=0.1,),
RandAdjustContrastd(keys=['image'], gamma=(0.5, 2.5), prob=0.1),
RandGaussianNoised(keys=['image'], prob=0.1, mean=np.random.uniform(0, 0.5), std=np.random.uniform(0, 1)),
RandShiftIntensityd(keys=['image'], offsets=np.random.uniform(0,0.3), prob=0.1),
SpatialPadd(keys=['image', 'label'], spatial_size=opt.patch_size, method= 'end'), # pad if the image is smaller than patch
RandSpatialCropd(keys=['image', 'label'], roi_size=opt.patch_size, random_size=False),
ToTensord(keys=['image', 'label'])
]
val_transforms = [
LoadImaged(keys=['image', 'label']),
AddChanneld(keys=['image', 'label']),
# ThresholdIntensityd(keys=['image'], threshold=-135, above=True, cval=-135),
# ThresholdIntensityd(keys=['image'], threshold=215, above=False, cval=215),
CropForegroundd(keys=['image', 'label'], source_key='image'), # crop CropForeground
NormalizeIntensityd(keys=['image']), # intensity
ScaleIntensityd(keys=['image']),
SpatialPadd(keys=['image', 'label'], spatial_size=opt.patch_size, method= 'end'), # pad if the image is smaller than patch
ToTensord(keys=['image', 'label'])
]
train_transforms = Compose(train_transforms)
val_transforms = Compose(val_transforms)
# create a training data loader
check_train = monai.data.Dataset(data=train_dicts, transform=train_transforms)
train_loader = DataLoader(check_train, batch_size=opt.batch_size, shuffle=True, collate_fn=list_data_collate, num_workers=opt.workers, pin_memory=False)
# create a training_dice data loader
check_val = monai.data.Dataset(data=train_dice_dicts, transform=val_transforms)
train_dice_loader = DataLoader(check_val, batch_size=1, num_workers=opt.workers, collate_fn=list_data_collate, pin_memory=False)
# create a validation data loader
check_val = monai.data.Dataset(data=val_dicts, transform=val_transforms)
val_loader = DataLoader(check_val, batch_size=1, num_workers=opt.workers, collate_fn=list_data_collate, pin_memory=False)
# create a validation data loader
check_val = monai.data.Dataset(data=test_dicts, transform=val_transforms)
test_loader = DataLoader(check_val, batch_size=1, num_workers=opt.workers, collate_fn=list_data_collate, pin_memory=False)
# build the network
if opt.network is 'nnunet':
net = build_net() # nn build_net
elif opt.network is 'unetr':
net = build_UNETR() # UneTR
net.cuda()
if num_gpus > 1:
net = torch.nn.DataParallel(net)
if opt.preload is not None:
net.load_state_dict(torch.load(opt.preload))
dice_metric = DiceMetric(include_background=True, reduction="mean", get_not_nans=False)
post_trans = Compose([EnsureType(), Activations(sigmoid=True), AsDiscrete(threshold_values=True)])
loss_function = monai.losses.DiceCELoss(sigmoid=True)
torch.backends.cudnn.benchmark = opt.benchmark
if opt.network is 'nnunet':
optim = torch.optim.SGD(net.parameters(), lr=opt.lr, momentum=0.99, weight_decay=3e-5, nesterov=True,)
net_scheduler = torch.optim.lr_scheduler.LambdaLR(optim, lr_lambda=lambda epoch: (1 - epoch / opt.epochs) ** 0.9)
elif opt.network is 'unetr':
optim = torch.optim.AdamW(net.parameters(), lr=1e-4, weight_decay=1e-5)
# start a typical PyTorch training
val_interval = 1
best_metric = -1
best_metric_epoch = -1
epoch_loss_values = list()
writer = SummaryWriter()
for epoch in range(opt.epochs):
print("-" * 10)
print(f"epoch {epoch + 1}/{opt.epochs}")
net.train()
epoch_loss = 0
step = 0
for batch_data in train_loader:
step += 1
inputs, labels = batch_data["image"].cuda(), batch_data["label"].cuda()
optim.zero_grad()
outputs = net(inputs)
loss = loss_function(outputs, labels)
loss.backward()
optim.step()
epoch_loss += loss.item()
epoch_len = len(check_train) // train_loader.batch_size
print(f"{step}/{epoch_len}, train_loss: {loss.item():.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:.4f}")
if opt.network is 'nnunet':
update_learning_rate(net_scheduler, optim)
if (epoch + 1) % val_interval == 0:
net.eval()
with torch.no_grad():
def plot_dice(images_loader):
val_images = None
val_labels = None
val_outputs = None
for data in images_loader:
val_images, val_labels = data["image"].cuda(), data["label"].cuda()
roi_size = opt.patch_size
sw_batch_size = 4
val_outputs = sliding_window_inference(val_images, roi_size, sw_batch_size, net)
val_outputs = [post_trans(i) for i in decollate_batch(val_outputs)]
dice_metric(y_pred=val_outputs, y=val_labels)
# aggregate the final mean dice result
metric = dice_metric.aggregate().item()
# reset the status for next validation round
dice_metric.reset()
return metric, val_images, val_labels, val_outputs
metric, val_images, val_labels, val_outputs = plot_dice(val_loader)
# Save best model
if metric > best_metric:
best_metric = metric
best_metric_epoch = epoch + 1
torch.save(net.state_dict(), "best_metric_model.pth")
print("saved new best metric model")
metric_train, train_images, train_labels, train_outputs = plot_dice(train_dice_loader)
metric_test, test_images, test_labels, test_outputs = plot_dice(test_loader)
# Logger bar
print(
"current epoch: {} Training dice: {:.4f} Validation dice: {:.4f} Testing dice: {:.4f} Best Validation dice: {:.4f} at epoch {}".format(
epoch + 1, metric_train, metric, metric_test, best_metric, best_metric_epoch
)
)
writer.add_scalar("Mean_epoch_loss", epoch_loss, epoch + 1)
writer.add_scalar("Testing_dice", metric_test, epoch + 1)
writer.add_scalar("Training_dice", metric_train, epoch + 1)
writer.add_scalar("Validation_dice", metric, epoch + 1)
# plot the last model output as GIF image in TensorBoard with the corresponding image and label
# val_outputs = (val_outputs.sigmoid() >= 0.5).float()
plot_2d_or_3d_image(val_images, epoch + 1, writer, index=0, tag="validation image")
plot_2d_or_3d_image(val_labels, epoch + 1, writer, index=0, tag="validation label")
plot_2d_or_3d_image(val_outputs, epoch + 1, writer, index=0, tag="validation inference")
plot_2d_or_3d_image(test_images, epoch + 1, writer, index=0, tag="test image")
plot_2d_or_3d_image(test_labels, epoch + 1, writer, index=0, tag="test label")
plot_2d_or_3d_image(test_outputs, epoch + 1, writer, index=0, tag="test inference")
print(f"train completed, best_metric: {best_metric:.4f} at epoch: {best_metric_epoch}")
writer.close()