A Parameter Refinement Method for Ptychography Based on Deep Learning Concepts

Abstract

X-ray ptychography is an advanced computational microscopy technique, which is delivering exceptionally detailed quantitative imaging of biological and nanotechnology specimens, which can be used for high-precision X-ray measurements. However, coarse parametrisation in propagation distance, position errors and partial coherence frequently threaten the experimental viability. In this work, we formally introduce these actors, solving the whole reconstruction as an optimisation problem. A modern deep learning framework was used to autonomously correct the setup incoherences, thus improving the quality of a ptychography reconstruction. Automatic procedures are indeed crucial to reduce the time for a reliable analysis, which has a significant impact on all the fields that use this kind of microscopy. We implemented our algorithm in our software framework, SciComPty, releasing it as open-source. We tested our system on both synthetic datasets, as well as on real data acquired at the TwinMic beamline of the Elettra synchrotron facility.

A modular software framework for the design and implementation of ptychography algorithms

“…Recently, the rPIE algorithm ( Maiden, Johnson & Li, 2017 ) has been proposed and studied: while being reasonably simple, it provides a fast convergence to a good object estimate ( Maiden, Johnson & Li, 2017 ). Compared to ePIE, we noticed that the rPIE algorithm, at least in our implementation, also provides a large computational FOV, which is comparable to the one seen in more advanced, but also computational eager, optimisation algorithms ( Guizar-Sicairos & Fienup, 2008 ; Thibault & Guizar-Sicairos, 2012 ; Guzzi et al, 2021b ). Indeed, these latter methods are used to refine a previous reconstruction, as they are more prone to stagnation ( Thibault & Guizar-Sicairos, 2012 ).…”

“…A correction method must then be introduced a posteriori , increasing the reconstruction time. In Guizar-Sicairos & Fienup (2008) ; Guzzi et al (2021b) positions are corrected through an optimisation method, within the gradient-based reconstruction; in Mandula et al (2016) , instead, the authors propose to use a gradient-less method based on Powell (1964) to guide the position refinement procedure. In this work, we consider a fast and reliable way to correct translations.…”

“…Writing such programs is typically done by an expert software engineer, and this step may discourage the prototyping or the study of new methods. Ptychography algorithms are complex not only in their implementation, but also per se : simple computational systems suffers from setup inconsistencies, which usually takes the form of bad parameter modelling of positions ( Zhang et al, 2013 ), distances ( Guzzi et al, 2021b ), illumination conditions ( Thibault & Menzel, 2013 ), just to enumerate a few. Such problems produce very noticeable artefacts.…”

Automatic Differentiation for Inverse Problems in X-ray Imaging and Microscopy

“…Following our framework, it is possible to rapidly prototype state-of-the-art solutions for computational microscopy methods. Possible examples are: a compressive sensing reconstruction algorithm for sparse datasets, a single image super-resolution algorithm for recovering missing information from a low-resolution image, a 2D/3D tomography reconstruction algorithm that can deal with the missing wedge problem, sparse acquisition and axial misalignment, and a ptychography algorithm, which can retrieve many geometry parameters [ 13 ]. Even if the aforementioned solutions are purely “conventional”, the use of such an AI component as AD blurs out the distinction between classical and AI methods [ 12 , 14 , 15 ].…”

Compressive Ptychography using Deep Image and Generative Priors

“…Its effectiveness is shown also for ptychographic inverse problem in [33]. In addition, supervised approaches such as the deep generative priors (DGPs) have also been applied for reducing imaging artifacts associated with undersampling [34], [35], as well as to overcome issues related to partialcoherence of the source [36]. Successful reconstructions using generative networks are demonstrated in [37] for Fourier ptychography and [38] for x-ray ptychography.…”