Ultrafast radiographic imaging and tracking: An overview of instruments, methods, data, and applications

‪Zhehui Wang, Andrew F. T. Leong, A. Dragone, A. E. Gleason, R. Ballabriga, Christopher Campbell, M. Campbell, Samuel J. Clark, Cinzia Da Vià, Dana M. Dattelbaum, M. Demarteau, L. Fabris, Kamel Fezzaa, Eric R. Fossum, Sol M. Grüner, Todd C. Hufnagel, Xiaolu Ju, Ke Li, X. Llopart, Bratislav Lukić, Alexander Rack, J. Strehlow, Audrey Corbeil Therrien, Julia Thom-Levy, Feixiang Wang, Tiqiao Xiao, Mingwei Xu, Xin Yue

Abstract

Ultrafast radiographic imaging and tracking (U-RadIT) use state-of-the-art ionizing particle and light sources to experimentally study sub-nanosecond dynamic processes in physics, chemistry, biology, geology, materials science and other fields. These processes, fundamental to nuclear fusion energy, advanced manufacturing, green transportation and others, often involve one mole or more atoms, and thus are challenging to compute by using the first principles of quantum physics or other forward models. One of the central problems in U-RadIT is to optimize information yield through, e.g. high-luminosity X-ray and particle sources, efficient imaging and tracking detectors, novel methods to collect data, and large-bandwidth online and offline data processing, regulated by the underlying physics, statistics, and computing power. We review and highlight recent progress in: a.) Detectors; b.) U-RadIT modalities; c.) Data and algorithms; and d.) Applications. Hardware-centric approaches to U-RadIT optimization are constrained by detector material properties, low signal-to-noise ratio, high cost and long development cycles of critical hardware components such as ASICs. Interpretation of experimental data, including comparisons with forward models, is frequently hindered by sparse measurements, model and measurement uncertainties, and noise. Alternatively, U-RadIT make increasing use of data science and machine learning algorithms, including experimental implementations of compressed sensing. Machine learning and artificial intelligence approaches, refined by physics and materials information, may also contribute significantly to data interpretation, uncertainty quantification, and U-RadIT optimization.

Introduction: The Challenge of Extreme Dynamics

Understanding how materials behave under extreme conditions—such as high pressure, intense heat, or high-velocity impact—requires observing atomic and molecular events that occur on sub-nanosecond timescales deep within opaque materials. While the fundamental quantum physics of these interactions is understood, simulating the collective behavior of millions of atoms remains computationally prohibitive. Consequently, experimental observation via Ultrafast Radiographic Imaging and Tracking (U-RadIT) is essential.

U-RadIT faces a massive "information-yield optimization" problem. Researchers are constrained by hardware limits, including detector speed, radiation damage, and data transfer bottlenecks, as well as the immense cost of operating accelerator or laser facilities. As a result, experiments often yield sparse, noisy data that is difficult to interpret using traditional methods.

The Spatiotemporal Landscape of U-RadIT

The field of U-RadIT spans a vast range of length and time scales, requiring a diverse array of radiation sources. From synchrotrons and X-ray Free Electron Lasers (XFELs) to high-energy particle beams, each source is optimized for specific physical regimes, such as gas phase dynamics or solid-state lattice deformations.

Spatiotemporal regimes of various radiation sources and physical processes, mapping length scales against time scales.

A typical U-RadIT experiment involves a complex chain of components: a high-energy radiation source, a target driven into an extreme state by external stimuli (drivers), and a high-speed detector system. The drivers can range from gas guns and explosives to pulsed lasers and high-voltage electricity, each designed to induce specific dynamic phenomena in the target material.

Schematic of a typical U-RadIT experimental setup showing the source, target, and detector components alongside various target drivers.

Technological Pillars: Detectors and Modalities

To overcome physical bottlenecks, the field is shifting toward hybrid pixelated array detectors (PADs) and Timepix ASICs, which enable data-driven readout and sub-nanosecond timestamping. To bypass data transmission limits, burst-mode CMOS cameras are employed; these store frames directly within the pixel's memory to achieve frame rates in the millions or billions per second.

Beyond hardware, novel imaging modalities are expanding what can be seen:

  • X-ray Phase Contrast Imaging (XPCI): Utilizes wavefield phase shifts rather than simple absorption to image low-contrast materials like polymers or biological tissues.
  • Dynamic Diffractive Imaging: Allows researchers to watch crystal lattices deform in real-time.
  • Move Contrast X-ray Imaging (MCXI): Isolates moving components from static backgrounds to enhance clarity in complex dynamic scenes.

Data Science and Machine Learning Integration

As hardware hits physical limits, software has become the primary bridge to higher information yield. The field is heavily adopting Compressed Sensing (CS), which allows for high-fidelity image reconstruction from highly undersampled data—a critical capability when experimental time is limited to a single line-of-sight.

Furthermore, Neural Networks (such as U-Nets and deep image priors) are increasingly replacing slow, iterative physics equations for rapid phase retrieval and noise reduction. These AI approaches, when refined by physics and materials information, contribute significantly to data interpretation and uncertainty quantification.

Key Breakthroughs and Applications

Modern U-RadIT has enabled several high-impact discoveries:

  • Inertial Fusion Energy: Researchers at the European Synchrotron (ESRF) have imaged shock-induced cavity collapse at 3.8 million frames per second to study hydrodynamic instabilities critical to fusion.
  • Additive Manufacturing: At the Advanced Photon Source (APS), simultaneous X-ray and thermal imaging combined with CNNs has been used to detect stochastic "keyhole" porosity defects in 3D-printed titanium with sub-millisecond resolution.
  • Extreme Material Dynamics: At the Linac Coherent Light Source (LCLS), researchers captured ultrafast lattice dynamics of silicon transitioning to a molten state using 10-picosecond X-ray diffraction intervals.

Future Directions and Limitations

Despite these advances, challenges remain. Standard silicon sensors are often transparent to high-energy X-rays, and high-Z alternatives like CdTe still suffer from manufacturing defects and non-linear responses. Additionally, sharp density jumps in shockwaves can cause "phase wrapping" issues in XPCI that confuse current algorithms.

Future work will focus on integrating 4D particle tracking into radiography and developing robust Uncertainty Quantification (UQ) algorithms. These tools will be essential for bounding errors in neural network reconstructions and guiding the design of the next generation of high-cost accelerator experiments.

Summary updated

Abstract

Ultrafast radiographic imaging and tracking (U-RadIT) use state-of-the-art ionizing particle and light sources to experimentally study sub-nanosecond transients or dynamic processes in physics, chemistry, biology, geology, materials science and other fields. These processes are fundamental to modern technologies and applications, such as nuclear fusion energy, advanced manufacturing, communication, and green transportation, which often involve one mole or more atoms and elementary particles, and thus are challenging to compute by using the first principles of quantum physics or other forward models. One of the central problems in U-RadIT is to optimize information yield through, e.g. high-luminosity X-ray and particle sources, efficient imaging and tracking detectors, novel methods to collect data, and large-bandwidth online and offline data processing, regulated by the underlying physics, statistics, and computing power. We review and highlight recent progress in: (a.) Detectors such as high-speed complementary metal-oxide semiconductor (CMOS) cameras, hybrid pixelated array detectors integrated with Timepix4 and other application-specific integrated circuits (ASICs), and digital photon detectors; (b.) U-RadIT modalities such as dynamic phase contrast imaging, dynamic diffractive imaging, and four-dimensional (4D) particle tracking; (c.) U-RadIT data and algorithms such as neural networks and machine learning, and (d.) Applications in ultrafast dynamic material science using XFELs, synchrotrons and laser-driven sources. Hardware-centric approaches to U-RadIT optimization are constrained by detector material properties, low signal-to-noise ratio, high cost and long development cycles of critical hardware components such as ASICs. Interpretation of experimental data, including comparisons with forward models, is frequently hindered by sparse measurements, model and measurement uncertainties, and noise. Alternatively, U-RadIT make increasing use of data science and machine learning algorithms, including experimental implementations of compressed sensing. Machine learning and artificial intelligence approaches, refined by physics and materials information, may also contribute significantly to data interpretation, uncertainty quantification and U-RadIT optimization.
BibTeX
@inproceedings{wang2023ultrafast,
  author = {‪Zhehui Wang and Andrew F. T. Leong and A. Dragone and A. E. Gleason and R. Ballabriga and Christopher Campbell and M. Campbell and Samuel J. Clark and Cinzia Da Vià and Dana M. Dattelbaum and M. Demarteau and L. Fabris and Kamel Fezzaa and Eric R. Fossum and Sol M. Grüner and Todd C. Hufnagel and Xiaolu Ju and Ke Li and X. Llopart and Bratislav Lukić and Alexander Rack and J. Strehlow and Audrey Corbeil Therrien and Julia Thom-Levy and Feixiang Wang and Tiqiao Xiao and Mingwei Xu and Xin Yue},
  title = {{Ultrafast radiographic imaging and tracking: An overview of instruments, methods, data, and applications}},
  booktitle = {Nuclear},
  year = {2023},
  doi = {10.1016/j.nima.2023.168690},
}
Figure