Speaker
Description
Tofu image processing toolkit has been used with a great success for nearly a decade to reconstruct microCT data acquired at imaging beamlines of the KIT Light Source in Karlsruhe, Canadian Light Source, and P23 beamline of DESY as well as laminography data acquired at ESRF ID19 and neutron-based CT scans done at Institut Laue-Langevin. The architecture, functionality, and performance of tofu have been described in J. of Synchrotron Radiation by T. Faragó, S. Gasilov, et al. in vol. 29, pp. 916-927, 2022. Here, we present in greater detail the features of tofu ez - a PyQt-based graphical user interface for generation of ufo-launch/tofu data reconstruction pipelines. Tofu ez permits one to interactively create image processing pipelines composed of up to 10 steps that encompass all essential operations encountered when dealing with microCT data acquired at synchrotrons, such as the removal of hot pixels, suppression of artifacts stemming from scintillator defects, phase retrieval, ring removal, and denoising. All of that is applicable to the half acquisition mode (360° tomography with the off-centered rotation axis to effectively double the horizontally field of view) combined with multiple vertical scans per sample. In such data sets, the horizontal and vertical overlaps between CT projections can be estimated automatically so that one obtains fully stitched data cubes in the end. Tofu ez produces a formatted bash script for as many CT data sets as it finds in the input directory. These scripts can be executed locally or submitted to a cluster as a slurm job. Recently, we have installed tofu on Maxwell cluster and added a parser of h5 files to enable reconstruction of data acquired at Hereon imaging stations at P05 and P07 beamlines.
