X-ray white beam based 26.7 Hz dynamic tomography

Abstract

Synchrotron radiation micro-computed tomography (SR-µCT) is a vital technique for the quantitative characterization of three-dimensional internal structures across diverse fields, including energy, integrated circuits, materials science, biomedicine, archaeology etc. While SR-µCT provides high spatial resolution and high image contrast, it typically offers only moderate temporal resolution, with acquisition times ranging from minutes to hours. Recently, dynamic SR-µCT has attracted significant interest for its capacity to capture real-time three-dimensional structural evolution. Here, we demonstrate a dynamic SR-µCT system operating at 26.7 Hz, developed at the BL09B test beamline of the Shanghai Synchrotron Radiation Facility using a filtered white beam. The key components of this system include an air-cooling millisecond fast shutter, an air-bearing rotation stage, a high-efficiency detector integrated with a Photron FASTCAM SA-Z camera and a custom-designed optical system, and a synchronization clock to ensure precise temporal alignment of all devices. Experimental results confirm the feasibility of this approach for in vivo four-dimensional studies, making it particularly promising for applications in biomedical research and related disciplines.

Keywords

dynamic CT white beam x-ray synchrotron radiation

Introduction

X-ray micro-computed tomography (µCT) is a non-destructive inspection technique widely used to visualize sample morphology and evaluate quantitative information on three-dimensional (3D) geometry and properties.^1,2 Since the advent of third-generation synchrotron radiation (SR) facilities in the 1990s, SR-µCT has gained prominence due to its superior advantages, including a large field of view (FOV), high spatial resolution, and high contrast. This high contrast is largely attributable to phase-contrast imaging, which offers exceptional sensitivity, particularly for low-Z materials such as those found in biomedical and carbon-based samples.

Consequently, SR-µCT has evolved into a widely utilized technique for the quantitative characterization of 3D internal structures across diverse research fields, including integrated circuits, energy, materials, and environmental science.³ Representative applications include imaging of a variety of electrode components before, during, and after battery operation,⁴ investigating an entire 3D stacked chip package at high resolutions in minutes,⁵ studying ancient and historical materials,⁶ investigating the underlying mechanisms responsible for nutrient and contaminant mobility, bioavailability, and behavior.⁷ These cases demonstrate that SR-µCT is a powerful technique capable of providing high-resolution and high-contrast 3D images. However, its main limitation has traditionally been a moderate temporal resolution, typically ranging from several minutes to hours per scan.

In recent years, monochromatic-beam-based dynamic SR-µCT with a FOV on the millimeter-to-centimeter scale has attracted increasing attention and has been applied across numerous research fields.^8,9 Applications include analyzing imaging the time evolution of the pore and micro-fracture networks during oil shale pyrolysis,¹⁰ studying the fragmentation and growth dynamics of dendritic microstructures,¹¹ unraveling dynamic processes in geologic systems,¹² studying the tissue motion for regional lung function,¹³ and investigating the dynamics of enhanced gas trapping applied to CO₂ storage in the presence of oil.¹⁴ The high energy resolution of the synchrotron monochromatic beam enables this technique to provide quantitative information.¹⁵ However, a fundamental trade-off exists between temporal resolution and FOV. Using a bending magnet or wiggler source can provide a centimeter-scale FOV, but the resulting temporal resolution is only moderate due to low flux density. Conversely, an undulator source delivers high flux density suitable for high temporal resolution, but the usable beam size is typically restricted to a few millimeters. Consequently, achieving high spatiotemporal resolution with a monochromatic beam and a large FOV (e.g., centimeters) remains highly challenging. To realize high spatiotemporal SR-µCT with a large FOV, the white beam from a bending magnet or wiggler source is often employed. This approach leverages the significantly higher flux compared to a monochromatic beam, enabling faster data acquisition across a larger FOV.¹⁶

Here, we present the development of a filtered white beam-based dynamic SR-µCT system operating at 26.7 Hz, implemented at the BL09B test beamline of the Shanghai Synchrotron Radiation Facility (SSRF). The key components of the system include an air-cooling millisecond fast shutter, an air-bearing rotation stage, and a high-efficiency detector coupled with a Photron FASTCAM SA-Z camera equipped with an in-house-developed optical system. A clock synchronization system ensures precise temporal alignment among all instruments. The capability of the system is demonstrated through dynamic SR-µCT experiments involving PMMA spheres, dry pittosporum tobira stem and a live ant.

Materials and methods

Beamline

The dynamic SR-µCT experiment was conducted at the BL09B test beamline at SSRF. The test beamline was designed for at-wavelength metrology to evaluate the performance of beamline optical components and experimental instrumentation.¹⁷ As illustrated in Figure 1, the beamline employs a 1.27 T bending magnet source with a maximum beam acceptance angle of 2.5 × 0.3 mrad², delivering a maximum available beam size of approximately 100 mm × 12 mm at the sample position, located about 41 meters from the source, under unfocused white-beam mode. The beamline is equipped with a suite of configurable optical components, including a water-cooled double-crystal monochromator (DCM), a compound refractive lenses (CRL), a toroidal mirror, and a KB mirror system. Each of these elements can be independently inserted into or removed from the beam path as needed, enabling flexible operation in various modes such as white beam, monochromatic beam, and both focused and unfocused beam configurations.

Figure 1.

Schematic of the BL09B test beamline.

During the white-beam dynamic SR-µCT experiment, all beamline optical components at BL09B were switched out, allowing the X-ray beam to pass solely through the beryllium (Be) window located approximately 38 meters from the source. The white-beam spectrum after the Be window is shown in Figure 2 (solid blue line), exhibiting a peak energy around 7 keV. This low-energy radiation is typically absorbed by biomedical samples due to its limited penetration capability. To reduce the radiation dose delivered to the sample, it is essential to minimize exposure to these low-energy photons that cannot penetrate the sample. Given that our samples are on the order of a few millimeters in diameter, a 1 mm aluminum (Al) filter was inserted into the beam path to attenuate the low-energy components of the white beam. The filtered spectrum is illustrated in Figure 2 (dashed orange line), showing a shift in the peak energy to approximately 22 keV. Both spectra were calculated using XOP 2.4 software.¹⁸

Figure 2.

The spectrum of the BL09B test beamline at the SSRF.

Experimental setup

Figure 3 shows the experimental setup for dynamic SR-µCT installed in the experimental hutch at BL09B of SSRF. The system consists of several key components: an in-house-developed air-cooled millisecond fast shutter capable of providing a minimum beam window of 30 milliseconds; a white-beam filter system; an air-bearing rotation stage (Aerotech) with a maximum rotation speed of 4800°/s; a high-speed X-ray detector; and a clock synchronization system to coordinate all instruments.

Figure 3.

The experimental step up of white beam dynamic SR-µCT (a), the Aerotech stage and alignment stages (b), and the L-type optics-coupled detector (c).

The white beam, generated by the 1.27 T bending magnet source at BL09B, delivers a heat load of approximately 73 W, which is substantial and poses risks of damage to both instrumentation and samples. For instance, the scintillator can be damaged within minutes if exposed directly to this intense radiation. To mitigate this, it is essential to limit beam exposure to only those intervals during which data acquisition occurs. This is achieved using a specialized white-beam shutter. A critical design consideration for such a shutter is the minimization of beam-through time. As illustrated in Figure 4, we developed an air-cooled double-blade shutter capable of achieving millisecond-level exposure control. Each blade is independently actuated by an electromagnet: when powered, the blade retracts; when depowered, it falls under the combined force of gravity and a spring. The blades are fabricated from tungsten carbide, with dimensions of 51 mm (width) × 10 mm (height) × 8 mm (thickness along the beam direction). By controlling the timing of each blade's descent, a precisely timed opening for X-ray transmission is created. To manage the thermal load, an air-cooling system is integrated to dissipate heat accumulated during exposure. Unlike crystal-based monochromators, which require extreme thermal stability, the shutter functions primarily as a beam-stop device. Thermal simulations performed using ANSYS software confirm that air cooling is sufficient to handle the 73 W heat load under experimental conditions. Additionally, a primary photon shutter upstream is used during sample alignment or changes to block the beam entirely, further reducing the thermal burden on the white-beam shutter. The shutter offers an adjustable opening time ranging from 30 ms to 10 s, operates effectively up to 100 keV, and is controlled via RS232 communication.

Figure 4.

The air-cooling white beam shutter.

The 1 mm Al white-beam filter, procured from Goodfellow, features a surface finish of Ra 0.1 on both sides and is mounted in a static holder without active cooling or switching mechanisms. The air-bearing rotation stage (Aerotech ABRT-200-AS) enables continuous 360° rotation at speeds of up to 4800°/s. It exhibits a positional accuracy of ±2 arc seconds, a tilt error motion below 0.6 arc seconds, and axial and radial error motions of 100 nm and 150 nm, respectively. These specifications exceed the requirements for SR-µCT experiments, ensuring high performance in dynamic imaging applications. As shown in Figure 5, measured performance at the maximum rotation speed of 4800°/s demonstrated an accuracy of 1.0 arc seconds and a tilt error of 0.49 arc seconds, surpassing the manufacturer's specifications.

Figure 5.

The accuracy (left) and tilt error (right) test results of the air-bearing rotation stage.

The detector features an L-shaped optics-coupled system, as depicted in Figure 6 and summarized in Table 1. The optical coupling system employs a long-working-distance microscope objective arranged in an L-shaped configuration, with the objective positioned behind a reflecting mirror. This design keeps the microscope objective out of the direct X-ray path, thereby shielding it from radiation damage caused by direct exposure. The optical system is optimally designed to accommodate a large numerical aperture (NA) microscope objective, significantly enhancing light collection efficiency. Additionally, the scintillator thickness is carefully matched to the NA of the microscope objective to achieve optimal imaging resolution and contrast.

Figure 6.

The sketch of the L-type optics-coupled detector.

Table 1.

The parameters of optics-coupled detector.

Item	Magnification	NA	Working distance	Effective pixel size	Number of pixels	Field of view	Scintillator
Value	8x	0.5	31.7mm	2.5µm	1024 × 1024	2.56mm × 2.56mm	LuAG: Ce (50 μm)

The detector consists of a 50 µm thick LuAG:Ce scintillator, a Photron FASTCAM SA-Z high-speed camera, and the abovementioned optical system. The camera has a native pixel size of 20 µm and a full resolution of 1024 × 1024 pixels, supporting frame rates of up to 120000 fps. Detector efficiency is primarily constrained by the light coupling efficiency of the optical system. Commercial long-working-distance microscope lenses typically achieve a coupling efficiency of around 5%. To improve this, a custom optical system was designed with a numerical aperture (NA) of 0.5 and 8× magnification, yielding an effective pixel size of 2.5 µm. In comparison, a typical Mitutoyo long-working-distance lens system offers 7.5× magnification and an NA of 0.21. The custom design thus provides approximately twice the NA and four times the coupling efficiency,¹⁹ significantly enhancing overall detector performance.

It is crucial to synchronize all instruments in white-beam dynamic SR-µCT to carefully protect the sample and detector from excessive heat load. Figure 7 illustrates the control system and timing chart designed for this purpose. A central PC initiates the experiment by sending a start signal to a Stanford Research Systems DG535 digital delay/pulse generator and subsequently receives the acquired data from the Photron FASTCAM SA-Z camera upon completion. The DG535 unit serves as the master clock synchronizer, coordinating the actions of the photon shutter, white-beam shutter, air-bearing rotation stage, and high-speed camera. Specifically, upon receiving the start signal from the PC (t = 0 s), the photon shutter opens and the rotation stage begins accelerating immediately. The FASTCAM SA-Z camera starts data acquisition with a delay of 0.1 s, followed by the opening of the white-beam shutter at 0.2 s. This timing sequence is designed to meet three key requirements: first, to minimize exposure of the detector and sample to the white beam; second, to allow the rotation stage to reach its desired speed before data collection begins; and third, to ensure that all collected dynamic µCT data are useful by initiating camera recording just prior to X-ray exposure. Once triggered, the rotation stage and camera operate independently at their preset speeds without further communication. After the experiment finishes, all instruments are simultaneously stopped or closed, and the acquired data are transferred from the camera to the PC for storage and processing.

Figure 7.

The control system (a) and timing chart (b) of dynamic SR-µCT.

Samples and data collection

Three samples were scanned using the dynamic SR-µCT system. The first sample consisted of polymethyl methacrylate (PMMA) spheres with diameters ranging from 50 μm to 200 μm, tightly packed in a plastic tube. This static sample was used to evaluate the capability and stability of the system under two distinct scanning modes: dynamic continuous rotation and conventional step-and-snap acquisition. In the dynamic continuous rotation mode, the rotation stage rotated continuously while the detector acquired projection images without synchronization between the two devices. In the step-snap mode, the stage rotated to each prescribed angle and paused, after which a trigger signal was sent to open the shutter and acquire a projection image. Upon receiving a completion signal from the detector, the shutter closed and the stage advanced to the next angle. This iterative process continued until the full angular range was covered. The inter-device communication and mechanical motion in step-snap mode resulted in a total acquisition time of approximately 600 s. Both datasets were acquired using the same hardware, with only the control software reconfigured between modes.

The second sample was a fully dried Pittosporum tobira stem, selected for its complex microarchitecture and structural stability during the brief scan time of approximately 0.36 s. The third sample was a live ant measuring approximately 1 mm × 3.5 mm, which was placed in a ventilated plastic container under normoxic conditions at room temperature during scanning.

All samples were mounted at the center of the rotation stage using 3 M double-sided tape and allowed to stabilize for at least 20 min to ensure secure adhesion and, in the case of the ant, sufficient calmness. Key acquisition parameters are summarized in Table 2. For the PMMA sample in step-snap mode, 1440 projections were acquired to satisfy the sampling requirements of Filtered Back Projection (FBP) reconstruction. In dynamic mode, both the PMMA and Pittosporum tobira stem were scanned with 940 projections, balancing rotation speed and detector frame rate. For the live ant, the frame rate was reduced to extend the total acquisition time to 5.45 s within the camera's fixed memory constraint.

Table 2.

The experimental parameters of two samples.

	Item
Sample	Rotation speed	Frame rate	Exposure time	Projectionnumber per CT	CT temporal resolution	Exposure time per CT	Total data collect time
PMMA spheres (Step-snap mode)	－	－	40 µs	1440	－	∼0.80 s	600 s
PMMA spheres (Dynamic CT mode)	4800°/s	25000 fps	40 µs	940	26.7 Hz	0.0375 s	0.36 s
Dry pittosporum tobira stem (Dynamic CT mode)	4800°/s	25000 fps	40 µs	940	26.7 Hz	0.0375 s	0.36 s
Live ant (Dynamic CT mode)	4500°/s	8000 fps	125 µs	320	25 Hz	0.04 s	5.45 s

The sample-to-detector distance was set to 33.5 cm for all experiments, and a 1 mm Al filter was used to shape the beam spectrum (Figure 2). This distance was optimized for propagation-based phase contrast imaging, considering the effective pixel size and beam spectrum. Flat-field and dark-field corrections were applied to all projections: flat-field images were acquired with the sample moved out of the beam path, and dark-field images were collected with the beam shut off, both using the same exposure time as their projection images.

Data processing

Data processing and 3D rendering for dynamic SR-µCT present substantial challenges due to the large data volumes involved—a single experiment can generate hundreds of gigabytes comprising hundreds or even thousands of individual CT scans. Manual processing and visualization of such datasets would be prohibitively time-consuming, underscoring the need for automated or semi-automated computational workflows.

In this study, dynamic SR-µCT data were processed using PITRE software,²⁰ a comprehensive and freely available software for synchrotron-based CT data. To meet the demands of dynamic imaging, PITRE incorporates an optimized processing pipeline: once key reconstruction parameters are manually defined, the software automatically generates and executes a batch processing task for the entire dynamic dataset. For instance, with a single command, PITRE can initiate reconstruction of all time-resolved CT volumes, specifying the number of projections per reconstruction and the interval between successive scans.

For the PMMA spheres and dried Pittosporum tobira stem, slices were reconstructed without phase retrieval to preserve edge-enhancement effects, which aid in visualizing fine structural details and identifying potential artifacts. In contrast, for the live ant sample, the phase-attenuation duality Paganin algorithm was applied to all projections prior to reconstruction.²¹ This approach yielded slices in which materials with higher refractive indices appeared brighter and those with lower indices darker. All reconstructions were performed using the filtered back-projection algorithm with a Shepp–Logan filter. The parameters of dynamic SR-µCT data processing are summarized in Table 3.

Table 3.

The parameters of dynamic SR-CT data processing.

Item	Phaseretrieval (PR)	PRalgorithm	δ/β value	Projection number	Interval projection number	Filter
PMMA spheres (Step-snap mode)	－	－	－	1440	－	Shepp Logan
PMMA spheres (Dynamic CT mode)	－	－	－	940	940	Shepp Logan
Dry pittosporum tobira stem (Dynamic CT mode)	－	－	－	940	940	Shepp Logan
Live ant (Dynamic CT mode)	yes	Paganin	500	320	320	Shepp Logan

3D rendering of the live ant dataset was carried out semi-automatically using Dragonfly software.²² Dragonfly supports Python scripting and allows the development of custom add-ons to implement specific workflows. Users can either execute commands directly in the integrated Python console or record a sequence of operations into a macro for automating repetitive tasks. This macro functionality was leveraged to achieve efficient, semi-automated 3D visualization of the time-resolved volumetric data.

Results

Reconstructed slices of the PMMA sphere sample obtained using both dynamic CT mode and step-snap mode are presented in Figure 8. The slice acquired in dynamic CT mode exhibits slightly more pronounced stripe artifacts, particularly in the region highlighted in orange, which can be attributed to the relatively limited number of projection images acquired in this mode. In addition, a slight blurring effect is visible in the dynamic CT results, arising from the continuous rotation of the sample during exposure, which leads to angular integration within each projection.

Figure 8.

Reconstructed slices of PMMA spheres sample with dynamic CT mode (a) and step-snap mode (b), (c) the differential image between (a) and (b), (d) and (e) the enlarged views of (a) and (b) respectively.

Despite these minor artifacts, the overall structural features of the PMMA spheres and the plastic tube remain clearly identifiable in the dynamic CT mode and are consistent with those obtained in the step-snap mode. The total hard X-ray exposure times were 0.36 s for the dynamic CT mode and about 0.80 s for the step-snap mode. These short exposure durations prevented any detectable sample damage or deformation, such as sphere displacement or distortion. It is worth noting that prolonged exposure to the hard X-ray white beam could cause significant sample alteration, including deformation or even rupture of the spheres.

Artifacts such as blurring—especially in spheres located farther from the center of rotation—can arise from several experimental factors, including inadequate sample fixation, centrifugal forces-induced motion, instability in rotation or camera acquisition speed, or incorrect projection number settings during reconstruction. Although it can be challenging to pinpoint the exact cause when such artifacts occur, the close agreement between the dynamic and step-snap reconstructions in this study confirms the stability and reliability of the dynamic CT system under the employed conditions.

A reconstructed slice, line profile, and enlarged views of the Pittosporum tobira stem are presented in Figure 9. The images reveal fine structural details with high clarity, capturing vascular tissues of varying dimensions. The line profile demonstrates sharp contrast transitions between different material, confirming excellent differentiation capability. Both the PMMA spheres and Pittosporum tobira stem were reconstructed using 940 projections, a value consistent with theoretical calculations based on the rotation stage speed and detector frame rate. The quality of these reconstructions demonstrates the robust performance of the dynamic SR-µCT system.

Figure 9.

A reconstructed result of pittosporum tobira stem sample: (a) slice, (b) the profile of green line in (a), (c) and (d) the enlarged areas images in (a).

Figure 10 presents reconstructed cross-sections along the reference yellow line (indicated in the 0 s image of Figure 11) of a live ant at different time points (0 s, 2.48 s, 2.64 s, 2.8 s, and 5.28 s). These time points were selected to highlight particularly significant motion dynamics. The slices clearly capture morphological changes resulting from the ant's movement during scanning, notably showing leg displacement (indicated by orange arrows) and opening/closing of vascular structures (marked with green arrows).

Figure 10.

Reconstructed slice at the yellow line position in the 0 s image in Figure 11 of a live ant at different times.

Figure 11.

3D rendering images of a live ant at different times with Dragonfly software.

Figure 11 shows 3D renderings of the live ant at corresponding time points, processed using Dragonfly software. While the ant's torso remained relatively stable due to confinement in the plastic container, significant leg movement is evident (yellow arrows). Despite these rapid motions, the reconstructed images maintain clear, smooth contours throughout the appendages, demonstrating that the system's 25 Hz temporal resolution is sufficient for four-dimensional in vivo studies of living ants.

Discussions

The dynamic SR-µCT has garnered significant interest due to its ability to capture three-dimensional structural evolution across diverse fields such as biomedicine, materials science, energy research, and semiconductor technology. A key advantage of SR-µCT lies in its high spatial and contrast resolution, yet there remains a growing need to image larger, centimeter-scale samples typical in these applications. Thus, an ideal dynamic SR-µCT system should combine high spatial, temporal, and contrast resolution with a sufficiently large FOV.

This is often achieved using white beam sources from bending magnets or wigglers at synchrotron facilities, which can readily provide centimeter-scale beam sizes at dedicated imaging beamlines. Notably, wiggler sources offer a flux density one to two orders of magnitude higher than bending magnets, enabling substantially improved temporal resolution in dynamic imaging. However, the increased heat load associated with wigglers must be carefully managed, for instance, through the use of water-cooled shutters.

When an undulator source is employed at a dedicated imaging beamline, the flux density can be several orders of magnitude higher than that of wigglers, though the available beam size is considerably smaller—typically around 2 mm at facilities like SSRF, compared to approximately 40 mm from bending magnets or wigglers. Despite this limitation, undulator-based beamlines enable high spatial-temporal resolution, and high-contrast dynamic SR-µCT using monochromatic beams, albeit within a smaller FOV. Each source type—bending magnet, wiggler, and undulator—offers distinct advantages, making these three approaches complementary in advancing dynamic SR-µCT capabilities.

In the current study, a temporal resolution of 26.7 Hz was achieved, limited primarily by the maximum speed of the rotation stage. There is significant potential to improve temporal resolution at SSRF through upgrades to the rotation stage. Increasing rotational speed would reduce the number of projections per CT scan, assuming detector performance and flux remain constant. For the PMMA sphere sample, 940 projections were used per reconstruction, though this number could be halved in practice using FBP algorithms and reduced even further with iterative reconstruction or artificial intelligence based processing methods. For instance, Sumitomo Rubber Industries Ltd has developed a custom high-speed rotation device achieving a temporal resolution of 10 ms, demonstrating the potential for substantial improvement beyond commercial stages.²³

Data processing for dynamic SR-µCT remains a considerable challenge due to the immense data volumes, which can reach terabytes per day. GPU-accelerated parallel computing on cluster systems is an effective strategy to enhance processing speed.²⁴ While many synchrotron facilities have established such computational clusters to handle these resource-intensive tasks, potential bottlenecks may include data transfer rates between PCs or clusters, as well as data read/write speeds between memory and hard disks. Addressing these limitations will be essential to fully leverage the capabilities of high-speed dynamic SR-µCT.

Conclusions

A dynamic SR-µCT system operating at 26.7 Hz has been successfully developed using a filtered white beam at the BL09B test beamline of SSRF. Experimental results demonstrate the system's stability and feasibility for in vivo four-dimensional studies, highlighting its strong potential for applications in biomedical research and related fields.

Footnotes

Acknowledgments

The dynamic CT instruments were preliminary tested at BL13W1 at the SSRF.

ORCID iDs

Rongchang Chen

Tiqiao Xiao

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by the National Key R&D Program of China (2022YFA1604002), the Guangdong Special Support Program (2023TQ07Z464), the National Natural Science Foundation of China (12405366), the Shenzhen Science and Technology Program (JCYJ20250527151304001).

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

References

Maire

Withers

. Quantitative X-ray tomography. Int Mater Rev 2014; 59: 1–43.

Chen

Longo

Rigon

, et al. Measurement of the linear attenuation coefficients of breast tissues by synchrotron radiation computed tomography. Phys Med Biol 2010; 55: 4993–5005.

Baruchel

Buffiere

J-Y

Cloetens

, et al. Advances in synchrotron radiation microtomography. Scr Mater 2006; 55: 41–46.

Tang

Yang

, et al. Synchrotron X-ray tomography for rechargeable battery research: fundamentals, setups and applications. Small Methods 2021; 5: 2100557.

Elmer

Barth

, et al. Synchrotron radiation microtomography for large area 3D imaging of multilevel microelectronic packages. J Electron Mater 2014; 43: 4421–4427.

Bertrand

Cotte

Stampanoni

, et al. Development and trends in synchrotron studies of ancient and historical materials. Phys Rep 2012; 519: 51–96.

Hettiarachchi

Donner

Doelsch

. Application of synchrotron radiation-based methods for environmental biogeochemistry: introduction to the special section. J Environ Qual 2017; 46: 1139–1145.

Catasta

Bianchini Massoni

Esposito

, et al. The role of dynamic computed tomography angiography in endoleak detection and classification after endovascular aneurysm repair: a comprehensive review. Diagnostics 2025; 15: 370.

Buurlage

J-W

Marone

Pelt

, et al. Real-time reconstruction and visualisation towards dynamic feedback control during time-resolved tomography experiments at TOMCAT. Sci Rep 2019; 9: 18379.

10.

Saif

Lin

Singh

, et al. Dynamic imaging of oil shale pyrolysis using synchrotron X-ray microtomography. Geophys Res Lett 2016; 43: 6799–6807.

11.

Zhang

Wang

Koe

, et al. Synchrotron X-ray imaging and ultrafast tomography in situ study of the fragmentation and growth dynamics of dendritic microstructures in solidification under ultrasound. Acta Mater 2021; 209: 116796.

12.

Marone

Schlepütz

Marti

, et al. Time resolved in situ X-ray tomographic microscopy unraveling dynamic processes in geologic systems. Front Earth Sci 2020; 7: 346.

13.

Dubsky

Hooper

Siu

KKW

, et al. Synchrotron-based dynamic computed tomography of tissue motion for regional lung function measurement. J R Soc Interface 2012; 9: 2213–2224.

14.

Scanziani

Singh

Menke

, et al. Dynamics of enhanced gas trapping applied to CO2 storage in the presence of oil using synchrotron X-ray micro tomography. Appl Energy 2020; 259: 114136.

15.

Chen

Yang

, et al. Monochromatic-beam-based dynamic X-ray microtomography based on OSEM-TV algorithm. J Xray Sci Technol 2017; 25: 1007–1017.

16.

Rack

Garcia-Moreno

Schmitt

, et al. On the possibilities of hard X-ray imaging with high spatio-temporal resolution using polychromatic synchrotron radiation. J Xray Sci Technol 2010; 18: 429–441.

17.

Fan

Xue

, et al. The design of the test beamline at SSRF. AIP Conf Proc 2019; 2054: 060040.

18.

Sánchez Del Río

Dejus

. XOP v2.4: recent developments of the x-ray optics software toolkit. In: Sanchez Del Rio M and Chubar O (eds) San Diego, California, USA, p. 814115.

19.

Xie

Luo

, et al. High-efficiency fast X-ray imaging detector development at SSRF. Synchrotron Radiat 2019; 26: 1631–1637.

20.

Chen

Dreossi

Mancini

, et al. PITRE: software for phase-sensitive X-ray image processing and tomography reconstruction. J Synchrotron Radiat 2012; 19: 836–845.

21.

Chen

Rigon

Longo

. Comparison of single distance phase retrieval algorithms by considering different object composition and the effect of statistical and structural noise. Opt Express 2013; 21: 7384–7399.

22.

The Dragonfly software . https://www.theobjects.com/index.html .

23.

Mashita

Yashiro

Kaneko

, et al. High-speed rotating device for X-ray tomography with 10ms temporal resolution. J Synchrotron Rad 2021; 28: 322–326.

24.

Zhang

Wang

, et al. Towards full-stack deep learning-empowered data processing pipeline for synchrotron tomography experiments. The Innovation 2024; 5: 100539.