def train_nodule_segmentation_no_augmentation_normalization_dice(
dataset_file,
output_weights_file,
batch_size=5,
num_epochs=10,
last_epoch=0,
initial_weights=None,
):
"""Train the network from scratch or from a preexisting set of weights on the dataset"""
# Loaders
dataset = h5py.File(dataset_file, "r")
df = loader.dataset_metadata_as_dataframe(dataset,
key='nodule_masks_spherical')
df_training = df[df.subset.isin([0, 1, 2, 3, 4, 5, 6, 7]) & df.has_mask]
dataset.close()
training_loader = loader.NoduleSegmentationSequence(dataset_file,
batch_size,
dataframe=df_training,
epoch_frac=1.0,
epoch_shuffle=False)
df_validation = df[df.subset.isin([8]) & df.has_mask]
validation_loader = loader.NoduleSegmentationSequence(
dataset_file,
batch_size,
dataframe=df_validation,
epoch_frac=1.0,
epoch_shuffle=False)
# Callbacks
model_checkpoint = ModelCheckpoint(output_weights_file,
monitor='val_loss',
verbose=1,
save_best_only=True)
early_stopping = EarlyStopping(monitor='val_loss', patience=10)
history_log = HistoryLog(output_weights_file + ".history")
# Setup network
network_size = [*DEFAULT_UNET_SIZE, 1, dice_coef_loss]
model = Unet(*network_size)
if initial_weights:
model.load_weights(initial_weights)
# Train
model.fit_generator(
generator=training_loader,
epochs=num_epochs,
initial_epoch=last_epoch,
verbose=1,
validation_data=validation_loader,
use_multiprocessing=True,
workers=4,
max_queue_size=20,
shuffle=True,
callbacks=[model_checkpoint, early_stopping, history_log])
def train_lung_segmentation(
dataset_file,
output_weights_file,
batch_size=5,
num_epochs=10,
last_epoch=0,
initial_weights=None,
):
"""Train the network from scratch or from a preexisting set of weights on the dataset"""
# Loaders
training_loader = loader.LungSegmentationSequence(dataset_file,
batch_size,
epoch_frac=0.1)
validation_loader = loader.LungSegmentationSequence(dataset_file,
batch_size,
subsets={8},
epoch_frac=0.3,
epoch_shuffle=False)
# Callbacks
model_checkpoint = ModelCheckpoint(output_weights_file,
monitor='val_loss',
verbose=1,
save_best_only=True)
early_stopping = EarlyStopping(monitor='val_loss', patience=10)
history_log = HistoryLog(output_weights_file + ".history")
# Setup network
network_size = [*DEFAULT_UNET_SIZE, 1]
model = Unet(*network_size)
if initial_weights:
model.load_weights(initial_weights)
# Train
model.fit_generator(
generator=training_loader,
epochs=num_epochs,
initial_epoch=last_epoch,
verbose=1,
validation_data=validation_loader,
use_multiprocessing=True,
workers=4,
max_queue_size=20,
shuffle=True,
callbacks=[model_checkpoint, early_stopping, history_log])
def train_fp_reduction_resnet(
model_builder,
dataset_file,
candidates_file,
output_weights_file,
batch_size=64,
num_epochs=30,
last_epoch=0,
initial_weights=None,
):
df = pd.read_csv(candidates_file)
df = df[~df.subset.isin([9])]
df = df[df.seriesuid.isin(
pd.Series(
df.seriesuid.unique()).sample(frac=0.35))].sort_values("seriesuid")
with h5py.File(dataset_file, 'r') as dataset:
cubes = preprocessing.load_cubes(df, dataset)
df = pd.DataFrame({
"cube": cubes,
"class": df["class"].tolist(),
"subset": df["subset"].tolist()
})
# Rebalance dataset to 2:1
no_nodule_len = len(df[df["class"] == 0])
df = pd.concat([
df[df["class"] == 0], df[df["class"] == 1].sample(n=no_nodule_len // 2,
replace=True)
],
ignore_index=True)
df_training = df[df.subset.isin([0, 1, 2, 3, 4, 5, 6, 7])]
df_validation = df[df.subset.isin([8])]
training_loader = loader.NoduleClassificationSequence(
batch_size, dataframe=df_training, do_augmentation=True)
validation_loader = loader.NoduleClassificationSequence(
batch_size, dataframe=df_validation, do_augmentation=False)
# Callbacks
model_checkpoint = ModelCheckpoint(output_weights_file,
monitor='val_loss',
verbose=1,
save_best_only=True)
early_stopping = EarlyStopping(monitor='val_loss', patience=10)
history_log = HistoryLog(output_weights_file + ".history")
# Setup network
model = model_builder((32, 32, 32, 1), 1)
model.compile(
optimizer=Adam(lr=1e-3),
loss='binary_crossentropy',
metrics=['accuracy'],
)
if initial_weights:
model.load_weights(initial_weights)
# Train
model.fit_generator(
generator=training_loader,
epochs=num_epochs,
initial_epoch=last_epoch,
verbose=1,
validation_data=validation_loader,
use_multiprocessing=False,
workers=8,
max_queue_size=20,
shuffle=True,
callbacks=[model_checkpoint, early_stopping, history_log])