Data augmentation:

Artificial Intelligence is currently being used in the medical field for both diagnosis and research. Uses range from classifying tumor types to creating images similar to those obtained with contrast media to facilitate diagnosis and reduce invasive substances in patients. In this presentation three different generative neural network architectures were investigated to augment contrast-enhanced brain MRI data and evaluate a possible improvement in tumor type classification: Generative Adversarial Network, Conditional Variational Autoencoder and Transfer Learning with cVAE.