Applies an affine transformation specified by the parameters given. # Arguments x: 2D numpy array, single image. theta: Rotat

Reads and saves all brain volumes into a single table file. Parameters ---------- dataset_dir : The path to all

Generates data containing batch_size samples

Generalized Dice Score https://arxiv.org/pdf/1707.03237

Getting the brain indexes that will be used by the generator. if mode=='train' => the original indexes will be used (because we bui

Performs normalization on each modalities of input

Removes 1% of the top and bottom intensities and perform normalization on the input 2D slice.

Removes 1% of the top and bottom intensities and perform normalization on the input 2D slice.

Perform a slice-based normalization on each modalities of input volume.

Updates indexes after each epoch

A function that reads and crops a brain modalities (nii.gz format) Parameters ---------- brain_dir : string The path to

A function that reads and crops a brain modalities (nii.gz format) Parameters ---------- brain_dir : string The pa

Reads data from table with respect to the 'view'

Save the segmented results into a .nii.gz file, so that it can be uploaded to the BraTS server. Note that to correctly save the segmented bra

creates a squeeze and excitation block https://arxiv.org/abs/1709.01507 Parameters ---------- x : tensor Input keras

The function that builds/loads UNet model, initializes the data generators for training and validation, and finally trains the model.

Builds UNet model Parameters ---------- input_shape : tuple Shape of the input data (height, width, channel) modifie

Custom data generator based on Keras Sequance class. This implementation enables multiprocessing and on-the-fly data augmentation

The final loss function consists of the summation of two losses "GDL" and "CE" with a regularization term.