Fast and flexible reconstruction algorithms and workflows for high-throughput tomographic imaging
Over the past few years, at the TOMCAT beamline at the Swiss Light Source a new state-of-the-art endstation devoted to tomographic microscopy with sub-second temporal resolution has been established, with new scientific results being published in different fields. The new GigaFRoST detector can stream the data with rates up to multiple GB/s in a continuous manner and therefore enables experiments not possible so far. To fully exploit these recent technological achievements, the IT infrastructure needs to be matched to these high and sustained data rates. The computational tools necessary for the post-processing of raw tomographic projections have however generally not experienced the same efficiency increase as the experimental facilities, hindering optimal exploitation of this new potential.
In the TOMCAT group, we developed a fast, flexible and user-friendly post-processing pipeline overcoming this efficiency mismatch and delivering reconstructed tomographic datasets just few seconds after the data have been acquired, enabling efficient post-processing of TBs of tomographic data. If the acquired datasets are though strongly underconstrained, consisting of only few noisy projections as it is typically the case when the time resolution is pushed to the limit, routines based on iterative algorithms and a priori information are required to satisfactorily reconstruct 3D volumes. In our experience, dedicated procedures for each different problem are though necessary. Taking advantage of the recent development of efficient projectors, we are therefore aiming at a highly flexible parallel iterative reconstruction framework, where the different algorithm components (e.g. projectors, regularizers and solvers) can be easily interchanged and therefore the reconstruction schemes easily optimized for each single experiment. In addition to enabling the reconstruction of a large variety of datasets, this flexible solution will provide invaluable insight and hands-on experience important for defining future directions.