Over the last decade Deep Neural Networks have produced unprecedented performance on a number of tasks, given sufficient data. However, in reality the collection of big datasets is extremely challenging for many fields and goals have to be achieved with limited amount of data. A way to address those issues is to perform Data Augmentation by using Deep Generative Models. In this Master Thesis we explore the field of Deep Generative Modelling of Microscopy Image Data. We develop a new, ResNet inspired, architecture for Variational Autoencoder and explore a novel training methodology inspired by ProGAN. The results are encouraging and motivate further research in the field.