Electron cryo-microscopy (cryo-EM) is the fastest growing technique to explore the structure of biological macromolecules. To limit radiation damage, images are recorded under low-dose conditions, which leads to high levels of experimental noise. To reduce the noise, one averages over many images, but this requires alignment and classification algorithms that are robust to the high levels of noise. When signal-to-noise ratios drop, cryo-EM 3D reconstruction algorithms become susceptible to overfitting, ultimately limiting their applicability. The algorithms can be improved by incorporating prior knowledge. The most widely used approaches in the field to date incorporate the prior knowledge that cryo-EM reconstructions are smooth in a Bayesian approach. However, in terms of information content, the smoothness prior reflects poorly compared to the vast amount of prior knowledge that structural biology has gathered in the past 50 years. I aim to develop a computational pipeline that can exploit much more of the existing knowledge about biological structures in the cryo-EM structure determination process. I will express this prior knowledge through convolutional neural networks that have been trained on many reconstructions, and use these networks in novel algorithms that optimise a regularised likelihood function. Similar approaches have excelled in image denoising and reconstruction in related areas. Preliminary results with simulated data suggest that significant improvements beyond the existing methods are possible, both in computational speed and in signal recovery capabilities. The proposed methods will enable faster computations with less user involvement, but most importantly, they will extend the applicability of cryo-EM structure determination to many more samples, alleviating the existing experimental requirements of particle size, ice thickness and sample purity.