Towards full-stack deep learning-empowered data processing pipeline for synchrotron tomography experiments

Abstract

Synchrotron tomography experiments are transitioning into multifunctional, cross-scale, and dynamic characterizations, enabled by new-generation synchrotron light sources and fast developments in beamline instrumentation. However, with the spatial and temporal resolving power entering a new era, this transition generates vast amounts of data, which imposes a significant burden on the data processing end. Today, as a highly accurate and efficient data processing method, deep learning shows great potential to address the big data challenge being encountered at future synchrotron beamlines. In this review, we discuss recent advances employing deep learning at different stages of the synchrotron tomography data processing pipeline. We also highlight how applications in other data-intensive fields, such as medical imaging and electron tomography, can be migrated to synchrotron tomography. Finally, we provide our thoughts on possible challenges and opportunities as well as the outlook, envisioning selected deep learning methods, curated big models, and customized learning strategies, all through an intelligent scheduling solution. Public summary: • Tomography experiments at future synchrotron beamlines face exascale big data challenges. • Deep learning is employed for various data processing tasks at current synchrotron tomography beamlines. • Future comprehensive and elongated data analysis processes require a full-stack deep learning pipeline. • The data-driven full-stack deep learning pipeline based on an intelligent scheduling center (ISC) holds the key to solve the big data challenges.