This was not used. I was just curious because others used this. Turns out this is not necessary for my networks

here we go one step, then use polyLR :param epoch: :return:

TN / (TN + FN)

changes best checkpoint pickle nnunettrainer class name to nnUNetTrainer :param experiments: :return:

This implementation differs from the official surface dice implementation! These two are not comparable!!!!! The normalized surface dice is

overwrite patient-based early stopping. Always run to 1000 epochs :return:

overwrite patient-based early stopping. Always run to 1000 epochs :return:

Use this function to predict a 2D image. If this is a 3D U-Net it will crash because you cannot predict a 2D image with that (you dum

We need to wrap this because we need to enforce self.network.do_ds = False for prediction

We need to wrap this because we need to enforce self.network.do_ds = False for prediction

Preprocesses a numpy array encoding a batch of images. # Arguments x: a 4D numpy array consists of RGB values within [0, 255]. # Retur

Preprocesses a numpy array encoding a batch of images. # Arguments x: a 4D numpy array consists of RGB values within [0, 255]. # Retur

input standardizing function Args: x: numpy.ndarray with shape (H, W, C) size: tuple (H_new, W_new), resized input shape Retur

input standardizing function Args: x: numpy.ndarray with shape (H, W, C) size: tuple (H_new, W_new), resized input shape Retur

Use this to predict new data :param input_files: :param output_file: :param softmax_ouput_file: :param mixed_

Use this to predict new data :param input_files: :param output_file: :param softmax_ouput_file: :param mixed_

Compute reduced focal loss between target and output logits. Source https://github.com/BloodAxe/pytorch-toolbelt See :class:`~pytorch_too

This was not used. I was just curious because others used this. Turns out this is not necessary for my networks

data and seg must already have been transposed by transpose_forward. properties are the un-transposed values (spacing etc) :p

data and seg must already have been transposed by transpose_forward. properties are the un-transposed values (spacing etc) :p

changes best checkpoint pickle nnunettrainer class name to nnUNetTrainer :param experiments: :return:

Reverse list

:param target_spacings: list of lists [[1.25, 1.25, 5]] :param input_folder_with_cropped_npz: dim: c, x, y, z | npz_file['data'] np.

gradient clipping improves training stability :param data_generator: :param do_backprop: :param run_online_evaluatio

gradient clipping improves training stability :param data_generator: :param do_backprop: :param run_online_evaluatio

due to deep supervision the return value and the reference are now lists of tensors. We only need the full resolution output because

On all datasets except 3d fullres on spleen the preprocessed data would look identical to ExperimentPlanner3D_v21 (I tested decathlon