Three-Dimensional, Multimodal Synchrotron Data for Machine Learning Applications

TL;DR

A unique, multimodal synchrotron dataset of a bespoke zinc-doped Zeolite 13X sample that can be used to develop advanced deep learning and data fusion pipelines and 3D reconstruction algorithms is presented.

Abstract

Machine learning techniques are being increasingly applied in medical and physical sciences across a variety of imaging modalities; however, an important issue when developing these tools is the availability of good quality training data. Here we present a unique, multimodal synchrotron dataset of a bespoke zinc-doped Zeolite 13X sample that can be used to develop advanced deep learning and data fusion pipelines. Multi-resolution micro X-ray computed tomography was performed on a zinc-doped Zeolite 13X fragment to characterise its pores and features, before spatially resolved X-ray diffraction computed tomography was carried out to characterise the homogeneous distribution of sodium and zinc phases. Zinc absorption was controlled to create a simple, spatially isolated, two-phase material. Both raw and processed data is available as a series of Zenodo entries. Altogether we present a spatially resolved, three-dimensional, multimodal, multi-resolution dataset that can be used for the development of machine learning techniques. Such techniques include development of super-resolution, multimodal data fusion, and 3D reconstruction algorithm development.

Figures (4)

• Figure 1: Summary of the datasets collected. (a) shows the experimental setup of a borosilicate capillary containing the Zn-doped Zeolite 13X sample, along with a 0.5mm ruby sphere. (b) is a single projection of the sample capillary from the I13-2 beamline at a 0.8125 micron pixel-size. (c) displays subslices of 3 XCT resolutions (0.8125, 1.625 and 2.6 micron pixel-size) and both XRD-CT diffraction spot-sizes ($25\mu m$ and $50\mu m$) for sodium phase of the sample here. The viridis colormap is used to identify XRD-CT against XCT. The XCT and XRD-CT datasets have been spatially aligned and registered with the help of the fiducial markers.
• Figure 2: Example of a sinogram at q=1.649 for a spot size of $25\mu m$ and the reconstructed image using a gridrec reconstruction algorithm in the TomoPy Python package. The reconstructed image shows the Sodium phase in that $25\mu m$ region of the Zn-doped Zeolite 13X sample. The data used to create this figure file number 43322 in https://zenodo.org/records/13329639.green_xrdct_diad
• Figure 3: Side-by-side images showing the same cross-section of the sample across all four different resolution on I13-2. Note that the smallest pixel-size of 0.325 microns a FOV smaller than the size of the sample and results in some exclusion of the sample.
• Figure 4: Data processing pipelines for both (a) XCT data and (b) XRD-CT data from acquisition through to reconstruction to post-processing. Softwares and file formats are also indicated