Fast reconstruction of single-shot wide-angle diffraction images through deep learning

Abstract

Single-shot x-ray imaging of short-lived nanostructures such as clusters and nanoparticles near a phase transition or non-crystalizing objects such as large proteins and viruses is currently the most elegant method for characterizing their structure. Using hard x-ray radiation provides scattering images that encode two-dimensional projections, which can be combined to identify the full three-dimensional object structure from multiple identical samples. Wide-angle scattering using XUV or soft x-rays, despite yielding lower resolution, provides three-dimensional structural information in a single shot and has opened routes towards the characterization of non-reproducible objects in the gas phase. The retrieval of the structural information contained in wide-angle scattering images is highly non-trivial, and currently no efficient rigorous algorithm is known. Here we show that deep learning networks, trained with simulated scattering data, allow for fast and accurate reconstruction of shape and orientation of nanoparticles from experimental images. The gain in speed compared to conventional retrieval techniques opens the route for automated structure reconstruction algorithms capable of real-time discrimination and pre-identification of nanostructures in scattering experiments with high repetition rate—thus representing the enabling technology for fast femtosecond nanocrystallography.

Reconstruction of nanoscale particles from single-shot wide-angle FEL diffractions patterns with physics-informed neural networks

“…Moreover, each individual experiment is non-reproducible as the Coulomb explosion prevents multiple illumination. It is also known that the particles emerging from the source have not yet relaxed to an equilibrium state at the time of illumination, hence geometric structures such as icosahedra have been found [12,34] that are not expected to be stable for large particle sizes.…”

“…This in turn implies a trade-off between reconstruction accuracy and numerical efficiency. Already in the case of only few parameters, neural networks outperform conventional forward fitting based on Monte Carlo simplex methods [34], which is expected to become even more prominent with increasing number of degrees of freedom. The limiting case is to represent the object on a discrete three-dimensional grid; such representations are commonly used for the reconstruction of real-space objects from a series of images using deep neural networks [37].…”

“…In Ref. [34] it was demonstrated that data augmentation techniques can be used to simulate these measurement artifacts and to train a neural network that is robust against such effects. We extend this augmentation approach by introducing additional filters and on-the-fly augmentation.…”

“…On the experimental side, the pre-selection of automatically recorded scattering patterns into various categories has been implemented as a classification task [10], and generative learning helped to reveal common features in patterns connected to object classes and imaging artifacts [33]. Recently, shape and orientation of icosahedral silver nanoclusters were reconstructed from experimental wide-angle scattering patterns using a neural network trained solely on simulated training data [34]. This was achieved by utilizing a convolutional neural network that, combined with data augmentation techniques, is capable of processing experimental images that suffer from a variety of physically relevant artifacts and defects.…”

Noise reduction and mask removal neural network for X-ray single-particle imaging

“…X-ray detectors can generate up to 15 Gbyte s À1 of raw data (Muennich et al, 2016) and machine learning solutions may be helpful to improve and speed up data analysis. Up to now, only a few studies have addressed this potential directly for X-ray SPI experiments, employing neural networks for image classification (Langbehn et al, 2018;Shi et al, 2019;Zimmermann et al, 2019;Ignatenko et al, 2021), for defect identification and phase retrieval (Cherukara et al, 2018;Lim et al, 2021;Wu, Juhas et al, 2021;, for shape and orientation recovery of silver nanoclusters (Stielow et al, 2020), and for the reconstruction of electron densities of metallic nanoparticles from experimental data of the 3D Fourier space (Chan et al, 2020).…”