Deep learning based classification of dynamic processes in time-resolved X-ray tomographic microscopy.

Time-resolved X-ray tomographic microscopy is an invaluable technique to investigate dynamic processes in 3D for extended time periods. Because of the limited signal-to-noise ratio caused by the short exposure times and sparse angular sampling frequency, obtaining quantitative information through post-processing remains challenging and requires intensive manual labor. This severely limits the accessible experimental parameter space and so, prevents fully exploiting the capabilities of the dedicated time-resolved X-ray tomographic stations. Though automatic approaches, often exploiting iterative reconstruction methods, are currently being developed, the required computational costs typically remain high. Here, we propose a highly efficient reconstruction and classification pipeline (SIRT-FBP-MS-D-DIFF) that combines an algebraic filter approximation and machine learning to significantly reduce the computational time. The dynamic features are reconstructed by standard filtered back-projection with an algebraic filter to approximate iterative reconstruction quality in a computationally efficient manner. The raw reconstructions are post-processed with a trained convolutional neural network to extract the dynamic features from the low signal-to-noise ratio reconstructions in a fully automatic manner. The capabilities of the proposed pipeline are demonstrated on three different dynamic fuel cell datasets, one exploited for training and two for testing without network retraining. The proposed approach enables automatic processing of several hundreds of datasets in a single day on a single GPU node readily available at most institutions, so extending the possibilities in future dynamic X-ray tomographic investigations.

Deep denoising for multi-dimensional synchrotron X-ray tomography without high-quality reference data.

Synchrotron X-ray tomography enables the examination of the internal structure of materials at submicron spatial resolution and subsecond temporal resolution. Unavoidable experimental constraints can impose dose and time limits on the measurements, introducing noise in the reconstructed images. Convolutional neural networks (CNNs) have emerged as a powerful tool to remove noise from reconstructed images. However, their training typically requires collecting a dataset of paired noisy and high-quality measurements, which is a major obstacle to their use in practice. To circumvent this problem, methods for CNN-based denoising have recently been proposed that require no separate training data beyond the already available noisy reconstructions. Among these, the Noise2Inverse method is specifically designed for tomography and related inverse problems. To date, applications of Noise2Inverse have only taken into account 2D spatial information. In this paper, we expand the application of Noise2Inverse in space, time, and spectrum-like domains. This development enhances applications to static and dynamic micro-tomography as well as X-ray diffraction tomography. Results on real-world datasets establish that Noise2Inverse is capable of accurate denoising and enables a substantial reduction in acquisition time while maintaining image quality.

Unveiling water dynamics in fuel cells from time-resolved tomographic microscopy data.

X-ray dynamic tomographic microscopy offers new opportunities in the volumetric investigation of dynamic processes. Due to data complexity and their sheer amount, extraction of comprehensive quantitative information remains challenging due to the intensive manual interaction required. Particularly for dynamic investigations, these intensive manual requirements significantly extend the total data post-processing time, limiting possible dynamic analysis realistically to a few samples and time steps, hindering full exploitation of the new capabilities offered at dedicated time-resolved X-ray tomographic stations. In this paper, a fully automatized iterative tomographic reconstruction pipeline (rSIRT-PWC-DIFF) designed to reconstruct and segment dynamic processes within a static matrix is presented. The proposed algorithm includes automatic dynamic feature separation through difference sinograms, a virtual sinogram step for interior tomography datasets, time-regularization extended to small sub-regions for increased robustness and an automatic stopping criterion. We demonstrate the advantages of our approach on dynamic fuel cell data, for which the current data post-processing pipeline heavily relies on manual labor. The proposed approach reduces the post-processing time by at least a factor of 4 on limited computational resources. Full independence from manual interaction additionally allows straightforward up-scaling to efficiently process larger data, extensively boosting the possibilities in future dynamic X-ray tomographic investigations.

High-numerical-aperture macroscope optics for time-resolved experiments.

A novel high-quality custom-made macroscope optics, dedicated to high-resolution time-resolved X-ray tomographic microscopy at the TOMCAT beamline at the Swiss Light Source (Paul Scherrer Institut, Switzerland), is introduced. The macroscope offers 4× magnification, has a very high numerical aperture of 0.35 and it is modular and highly flexible. It can be mounted both in a horizontal and vertical configuration, enabling imaging of tall samples close to the scintillator, to avoid edge-enhancement artefacts. The macroscope performance was characterized and compared with two existing in-house imaging setups, one dedicated to high spatial and one to high temporal resolution. The novel macroscope shows superior performance for both imaging settings compared with the previous systems. For the time-resolved setup, the macroscope is 4 times more efficient than the previous system and, at the same time, the spatial resolution is also increased by a factor of 6. For the high-spatial-resolution setup, the macroscope is up to 8.5 times more efficient with a moderate spatial resolution improvement (factor of 1.5). This high efficiency, increased spatial resolution and very high image quality offered by the novel macroscope optics will make 10-20 Hz high-resolution tomographic studies routinely possible, unlocking unprecedented possibilities for the tomographic investigations of dynamic processes and radiation-sensitive samples.