This paper proposes a ptychographic phase-retrieval algorithm combined with a deep neural network (DNN). The proposed method allows the measurement model to be explicitly incorporated into the DNN-based approach, improving imaging capability and robustness to changes in experimental conditions.

The incorporation of prior knowledge about the structures of biological macromolecules into the reconstruction process of cryo-EM structure determination is proposed. Using a novel algorithm inspired by regularization by denoising, it is shown how convolutional neural networks can be used within this framework to improve reconstructions from simulated data.

Dissimilar hardware and software conventions at various synchrotrons lead to quantitative differences in experimental results. This paper proposes a method to improve reproducibility of tomographic reconstructions by optimizing the filtering step in commonly used reconstruction algorithms.