def generate_train_batch(self):
data = self._data[0]
seg = self._data[1]
if self.Config.SLICE_DIRECTION == "x":
end = data.shape[0]
elif self.Config.SLICE_DIRECTION == "y":
end = data.shape[1]
elif self.Config.SLICE_DIRECTION == "z":
end = data.shape[2]
# Stop iterating if we reached end of data
if self.global_idx >= end:
# print("Stopped because end of file")
self.global_idx = 0
raise StopIteration
new_global_idx = self.global_idx + self.batch_size
# If we reach end, make last batch smaller, so it fits exactly into rest
if new_global_idx >= end:
new_global_idx = end # not end-1, because this goes into range, and there automatically -1
slice_idxs = list(range(self.global_idx, new_global_idx))
x, y = dataset_utils.sample_slices(data, seg, slice_idxs,
training_slice_direction=self.Config.SLICE_DIRECTION,
labels_type=self.Config.LABELS_TYPE)
data_dict = {"data": x, # (batch_size, channels, x, y, [z])
"seg": y} # (batch_size, channels, x, y, [z])
self.global_idx = new_global_idx
return data_dict
def generate_train_batch(self):
subjects = self._data[0]
subject_idx = int(random.uniform(0, len(subjects))) # len(subjects)-1 not needed because int always rounds to floor
if self.Config.TYPE == "combined":
if np.random.random() < 0.5:
data = np.load(join(C.DATA_PATH, "HCP_fusion_npy_270g_125mm", subjects[subject_idx], "270g_125mm_xyz.npy"), mmap_mode="r")
else:
data = np.load(join(C.DATA_PATH, "HCP_fusion_npy_32g_25mm", subjects[subject_idx], "32g_25mm_xyz.npy"), mmap_mode="r")
data = np.reshape(data, (data.shape[0], data.shape[1], data.shape[2], data.shape[3] * data.shape[4]))
seg = np.load(join(C.DATA_PATH, self.Config.DATASET_FOLDER, subjects[subject_idx], self.Config.LABELS_FILENAME + ".npy"), mmap_mode="r")
else:
data = np.load(join(C.DATA_PATH, self.Config.DATASET_FOLDER, subjects[subject_idx], self.Config.FEATURES_FILENAME + ".npy"), mmap_mode="r")
seg = np.load(join(C.DATA_PATH, self.Config.DATASET_FOLDER, subjects[subject_idx], self.Config.LABELS_FILENAME + ".npy"), mmap_mode="r")
data = np.nan_to_num(data)
seg = np.nan_to_num(seg)
slice_idxs = np.random.choice(data.shape[0], self.batch_size, False, None)
slice_direction = dataset_utils.slice_dir_to_int(self.Config.TRAINING_SLICE_DIRECTION)
x, y = dataset_utils.sample_slices(data, seg, slice_idxs,
slice_direction=slice_direction,
labels_type=self.Config.LABELS_TYPE)
data_dict = {"data": x, # (batch_size, channels, x, y, [z])
"seg": y} # (batch_size, channels, x, y, [z])
return data_dict
def generate_train_batch(self):
subjects = self._data[0]
subject_idx = int(random.uniform(0, len(subjects))) # len(subjects)-1 not needed because int always rounds to floor
data, seg = load_training_data(self.Config, subjects[subject_idx])
#Convert peaks to tensors if tensor model
if self.Config.NR_OF_GRADIENTS == 18:
data = img_utils.peak_image_to_tensor_image(data)
slice_direction = dataset_utils.slice_dir_to_int(self.Config.TRAINING_SLICE_DIRECTION)
slice_idxs = np.random.choice(data.shape[slice_direction], self.batch_size, False, None)
x, y = dataset_utils.sample_slices(data, seg, slice_idxs,
slice_direction=slice_direction,
labels_type=self.Config.LABELS_TYPE)
# Can be replaced by crop
# x = pad_nd_image(x, self.Config.INPUT_DIM, mode='constant', kwargs={'constant_values': 0})
# y = pad_nd_image(y, self.Config.INPUT_DIM, mode='constant', kwargs={'constant_values': 0})
# x = center_crop_2D_image_batched(x, self.Config.INPUT_DIM)
# y = center_crop_2D_image_batched(y, self.Config.INPUT_DIM)
#Crop and pad to input size
x, y = crop(x, y, crop_size=self.Config.INPUT_DIM) # does not work with img with batches and channels
# Works -> results as good? -> todo: make the same way for inference!
# This is needed for Schizo dataset
# x = pad_nd_image(x, shape_must_be_divisible_by=(16, 16), mode='constant', kwargs={'constant_values': 0})
# y = pad_nd_image(y, shape_must_be_divisible_by=(16, 16), mode='constant', kwargs={'constant_values': 0})
# Does not make it slower
x = x.astype(np.float32)
y = y.astype(np.float32) # if not doing this: during validation: ConnectionResetError: [Errno 104] Connection
# reset by peer
#possible optimization: sample slices from different patients and pad all to same size (size of biggest)
data_dict = {"data": x, # (batch_size, channels, x, y, [z])
"seg": y} # (batch_size, channels, x, y, [z])
return data_dict
def generate_train_batch(self):
subjects = self._data[0]
subject_idx = int(
random.uniform(0, len(subjects))
) # len(subjects)-1 not needed because int always rounds to floor
data, seg = load_training_data(self.Config, subjects[subject_idx])
slice_idxs = np.random.choice(data.shape[0], self.batch_size, False,
None)
x, y = dataset_utils.sample_slices(
data,
seg,
slice_idxs,
training_slice_direction=self.Config.TRAINING_SLICE_DIRECTION,
labels_type=self.Config.LABELS_TYPE)
data_dict = {
"data": x, # (batch_size, channels, x, y, [z])
"seg": y
} # (batch_size, channels, x, y, [z])
return data_dict