"shear": [0, 0, 0], # degrees
"translation": [16, 16, 16], # mms (from -128 to 128)
"reflection": [0, 0, 0] #Bernoulli p
}
IMAGE_SIZE = 64
"Put in here the preprocessors for your data." \
"They will be run consequently on the datadict of the dataloader in the order of your list."
preprocessors = [
AugmentOnlyPositive(tags=["luna:3d", "luna:segmentation"],
output_shape=(IMAGE_SIZE,IMAGE_SIZE,IMAGE_SIZE), # in pixels
norm_patch_size=(IMAGE_SIZE,IMAGE_SIZE,IMAGE_SIZE), # in mms
augmentation_params=AUGMENTATION_PARAMETERS
),
ZMUV("luna:3d", bias = -648.59027, std = 679.21021),
]
#####################
# training #
#####################
"This is the train dataloader. We will train until this one stops loading data."
"You can set the number of epochs, the datasets and if you want it multiprocessed"
training_data = LunaDataLoader(
only_positive=True,
sets=TRAINING,
epochs=30,
preprocessors=preprocessors,
multiprocess=True,
crash_on_exception=True,
)
mode="constant"
)
preprocessors = [
# DicomToHU(tags=[tag+"3d"]),
augment_roi(
augmentation_params={
"scale": [1, 1, 1], # factor
"uniform scale": 1, # factor
"rotation": [0, 0, 180], # degrees
"shear": [0, 0, 0], # degrees
"translation": [5, 5, 5], # mm
"reflection": [0, 0, 0]}, # Bernoulli p
),
# DefaultNormalizer(tags=[tag+"3d"])
ZMUV(tag+"3d", bias=-648.59027, std=679.21021)
]
preprocessors_valid = [
# DicomToHU(tags=[tag+"3d"]),
augment_roi(),
# DefaultNormalizer(tags=[tag+"3d"])
ZMUV(tag+"3d", bias=-648.59027, std=679.21021)
]
#####################
# training #
#####################
"This is the train dataloader. We will train until this one stops loading data."
"You can set the number of epochs, the datasets and if you want it multiprocessed"
plot = False
# extract nodules in background thread
multiprocess = True
# the tag for the new data
tag = "stage1:"
# put in the pixelspacing tag if it's not loaded yet to be able to make patches
extra_tags=[]
# for building the segmentation model, the input tag should be replaced
replace_input_tags = {"luna:3d": tag+"3d"} #{old:new}
# prep before patches
preprocessors = [DicomToHU(tags=[tag+"3d"])]
# prep on the patches
postpreprocessors = [ZMUV(tag+"3d", bias = -648.59027, std = 679.21021)]
data_loader= Stage1DataLoader(
sets=[TRAINING, VALIDATION],
preprocessors=preprocessors,
epochs=1,
multiprocess=False,
crash_on_exception=True)
batch_size = 1 # only works with 1
# function to call to extract nodules from the fully reconstructed segmentation
def extract_nodules(segmentation):
"""segmentation is a 3D array"""
rois = blob_dog(segmentation, min_sigma=1, max_sigma=15, threshold=0.1)
if rois.shape[0] > 0:
Augment3D(
tags=["bcolzall:3d"],
output_shape=nn_input_shape,
norm_patch_shape=norm_patch_shape,
augmentation_params={
"scale": [1, 1, 1], # factor
"uniform scale": 1, # factor
"rotation": [0, 0, 180], # degrees
"shear": [0, 0, 0], # degrees
"translation": [50, 50, 50], # mm
"reflection": [.5, .5, .5]
}, # Bernoulli p
interp_order=1,
mode="constant"),
# DefaultNormalizer(tags=["bcolzall:3d"])
ZMUV("bcolzall:3d", bias=-648.59027, std=679.21021)
]
preprocessors_valid = [
# DicomToHU(tags=["bcolzall:3d"]),
Augment3D(tags=["bcolzall:3d"],
output_shape=nn_input_shape,
norm_patch_shape=norm_patch_shape,
interp_order=1,
mode="constant"),
ZMUV("bcolzall:3d", bias=-648.59027, std=679.21021)
]
#####################
# training #
#####################