(1 + b^2) * TP / ((1 + b^2) * TP + b^2 * FN + FP)

:param spacing: :param patch_size: :param min_feature_map_size: min edge length of feature maps in bottleneck :return:

ExperimentPlanner configures pooling so that we pool late. Meaning that if the number of pooling per axis is (2, 3, 3), then the firs

We use FabiansUNet instead of Generic_UNet

We need to adapt ref

We need to adapt ref

Computation of input patch size starts out with the new median shape (in voxels) of a dataset. This is opposed to prior experiments w

ExperimentPlanner configures pooling so that we pool late. Meaning that if the number of pooling per axis is (2, 3, 3), then the firs

per default we use the 50th percentile=median for the target spacing. Higher spacing results in smaller data and thus faster and easi

per default we use the 50th percentile=median for the target spacing. Higher spacing results in smaller data and thus faster and easi

For prediction of test cases just set training=False, this will prevent loading of training data and training batchgenerator initiali

- replaced get_default_augmentation with get_moreDA_augmentation - only run this code once - loss function wrapper for deep s

For prediction of test cases just set training=False, this will prevent loading of training data and training batchgenerator initiali

- replaced get_default_augmentation with get_moreDA_augmentation - enforce to only run this code once - loss function wrapper

- replaced get_default_augmentation with get_moreDA_augmentation - enforce to only run this code once - loss function wrapper

For prediction of test cases just set training=False, this will prevent loading of training data and training batchgenerator initiali

- replaced get_default_augmentation with get_moreDA_augmentation - only run this code once - loss function wrapper for deep s

removed deep supervision :return:

this is a copy of nnUNetTrainerV2's initialize. We only add the regions to the data augmentation :param training: :param forc

this is a copy of nnUNetTrainerV2's initialize. We only add the regions to the data augmentation :param training: :param forc

initialize self.network here :return:

changed deep supervision to False :return:

- momentum 0.99 - SGD instead of Adam - self.lr_scheduler = None because we do poly_lr - deep supervision = True

- momentum 0.99 - SGD instead of Adam - self.lr_scheduler = None because we do poly_lr - deep supervision = True

changed deep supervision to False :return:

initialize self.optimizer and self.lr_scheduler (if applicable) here :return:

TP / (TP + FP + FN)

used for if the checkpoint is already in ram :param checkpoint: :param train: :return: