Sep 23 – 27, 2024
ESRF Auditorium
Europe/Paris timezone

Frame Tripping at the EMBL P14 Beamline: from Raw Data to Tomographic Reconstruction in 30s

Sep 25, 2024, 2:40 PM
15m
Hybrid event (ESRF Auditorium)

Hybrid event

ESRF Auditorium

EPN Campus ESRF - ILL 71 Av. des Martyrs, 38000 Grenoble
Talk Data Analysis Data Analysis

Speaker

Marina Nikolova (EMBL)

Description

Since the initial proof-of-principle X-Ray imaging experiments at the EMBL beamline P14 at Petra III [1], a high throughput tomography (HiTT) setup has been established [2], and is now offered as a regular user service. The setup features a high performance data acquisition and processing software system.
Here we present a recent highlight: the achievement of significant gains in the 3D reconstruction speed. We are now automatically producing a fully reconstructed data set ~30s after the completion of the acquisition. This includes the flat-field correction and phase retrieval of 4 x 1810 frames of size 4 Megapixels followed by reconstruction using TomoPy Gridrec [3]. The implementation of the latter has been modified in order to reduce its memory usage and to limit any application of padding i.e. extraneous looping over frames to only once. Furthermore Gridrec has been incorporated in our phase-retrieval software: TOMO CTF, thereby bypassing all of the TomoPy python wrapper routines and directly calling its C routines instead.
MPI [4] has been employed for cluster inter-node communication thus removing the previously necessary step of reading the CTF-corrected data from disk before proceeding with the reconstruction. A variable number of cluster nodes can be configured for the tomographic processing. By default we are using 4 during regular beamtime data acquisition, a total of 352 cores. The data is written directly onto a 1.2 PB parallel file cluster storage, BeeGFS.
Through a further refactoring of the software more advanced acquisition strategies can also be automatically processed fast. At present the reconstruction of a 360 degree rotation with an extended field of view is being added to TOMO CTF.
A newly integrated PCO camera contributes to a more stable and error-free operation.
These recent developments have been instrumental in providing streamlined and user-friendly computational services during HiTT experiments at P14.
We thank the BMBF for funding the beamline compute resources ('Verbundforschung', 05K16GU1 and 05K19GU1).

[1] Polikarpov, M., Bourenkov, G., Snigireva, I., Snigirev, A., Zimmermann, S., Csanko, K., Brockhauser, S. & Schneider, T. R. (2019). Acta Cryst. D75, 947–958.
[2] Albers J, Nikolova M, Svetlove A, Darif N, Lawson MJ, Schneider TR, Schwab Y, Bourenkov G and Duke E. High Throughput Tomography (HiTT) on EMBL beamline P14 on PETRA III. Journal of Synchrotron Radiation (2024). DOI: 10.1107/S160057752300944X
[3] Gürsoy D, De Carlo F, Xiao X, and Jacobsen C. Tomopy: a framework for the analysis of synchrotron tomographic data. Journal of Synchrotron Radiation, 21(5):1188–1193, 2014.
[4] https://www.mpich.org/

Abstract publication I agree that the abstract will be published on the web site

Primary authors

Co-authors

Presentation materials