Real-time coherent diffraction inversion using deep generative networks

Abstract

Phase retrieval, or the process of recovering phase information in reciprocal space to reconstruct images from measured intensity alone, is the underlying basis to a variety of imaging applications including coherent diffraction imaging (CDI). Typical phase retrieval algorithms are iterative in nature, and hence, are time-consuming and computationally expensive, making real-time imaging a challenge. Furthermore, iterative phase retrieval algorithms struggle to converge to the correct solution especially in the presence of strong phase structures. In this work, we demonstrate the training and testing of CDI NN, a pair of deep deconvolutional networks trained to predict structure and phase in real space of a 2D object from its corresponding far-field diffraction intensities alone. Once trained, CDI NN can invert a diffraction pattern to an image within a few milliseconds of compute time on a standard desktop machine, opening the door to real-time imaging.

Advanced denoising for X-ray ptychography

“…Experimental results analyze a variety of datasets with different experimental noise sources and demonstrate how the proposed algorithm achieves superior reconstruction results and denoising than state-of-the-art solutions. As a future work, we will consider partial coherence [37] and position retrieval [26] in the model, and also explore additional sparsity techniques [38], deep-learning [39], and dictionary-learning methods [29] to further enhance the reconstruction results.…”

A data ecosystem to support machine learning in materials science

“…X-Ray coherent diffraction imaging (CDI) is an experimental technique that allows for determination of material structure and phase, with the phase encoding many interesting material properties, such as strain state [30]. To enable rapid phase and structure predictions from CDI data, Cherukara et al [30] built a deep convolutional neural network to predict phase and structure, using a set of simulated CDI images of varying structures with varying strains for model training. Given the importance of CDI in materials science, especially at the Advanced Photon Source at Argonne National Laboratory, the widespread availability and deployment of such a model would be of great value, enhancing the ability to gather information quickly on samples and to assess the state of an experiment in a control loop.…”

Analysis of non-iterative phase retrieval based on machine learning

“…where H[•] is the inverse function for phase retrieval. The inverse problem has been recently solved non-iteratively using machine-learning-based approaches [26][27][28][29][30][31]33]. In the present research, we use a convolutional residual network called ResNet [27,31,39] for calculating the inverse function in (2) non-iteratively.…”

“…In the second approach, a DNN is used to calculate the inverse function in the phase retrieval problem [27,28]. The second approach enables faster non-iterative phase retrieval than the conventional methods, and it has been used for real-time diffraction imaging, imaging through scattering media, computer-generated holograms, wavefront sensing, and pulse measurement [27][28][29][30][31][32][33][34]. Also, such DNN-based inversion has been introduced to optical sensing methods other than phase retrieval [35][36][37][38].…”