Computational protocol: Automatic processing of multimodal tomography datasets

Similar protocols

Protocol publication

[…] The first step in the processing chain involves correcting the datasets by an I 0 measurement. In the process list used for this work we do this in one of two ways. The first involves using a bespoke plugin, which takes the I 0 dataset and another dataset and outputs a single, corrected dataset normalized to the input beam flux. Another way to do this would be to use a plugin that performs basic operations. This plugin can take in, and output, any number of datasets, and subjects them to basic linear operations (available from NumPy) by parsing an input string of the command. Here the former of these choices are used because it standardizes the procedure. [...] The process to be applied to the data in the chain is to reduce the XRD images via azimuthal integration to representative patterns (Fig. 2). This is an example of a one-to-one I/O mapping, but where the data changes shape. Here, we have written a plugin using the CPU implementation of the ESRF package pyFAI (Kieffer & Karkoulis, 2013). This choice demonstrates one of the central tenets of Savu; that we should not reinvent the wheel for processes that already have heavily optimized solutions. The CPU version of pyFAI reduces each image into line profiles taking 300 ms per 2083 × 4150 frame. We rely on input from the user in the form of a NeXus calibration file for the detector geometry. In this instance, this is generated using DAWN (Basham et al., 2015) via a LaB6 calibrant. […]

Pipeline specifications

Software tools Numpy, PyFAI
Applications Miscellaneous, Small-angle scattering