Synchrotron Tomography Overview

• Synchrotron-based tomography is a high-resolution, non-destructive 3D imaging technique that exploits synchrotron radiation to reveal internal microstructures using various contrast mechanisms.
• It employs advanced approaches like phase contrast, chemical mapping, and magnetic vector imaging to provide rapid, quantitative visualization for materials science, biomedicine, and dynamic process studies.
• Innovative image reconstruction and deep learning segmentation techniques enhance data quality, enabling accurate morphometric analysis and real-time monitoring of transient phenomena.

Synchrotron-based tomography is a high-resolution, non-destructive three-dimensional (3D) imaging technique that exploits the unique properties of synchrotron radiation—high brilliance, tunable energy, coherence, and polarization—to enable detailed morphological, chemical, and even magnetic investigation of materials across a broad spectrum of applications. By delivering unparalleled spatial and temporal resolution through approaches such as X-ray microtomography, scanning transmission X-ray microscopy, and advanced phase-contrast modalities, synchrotron-based tomography has become central in materials science, biomedical research, engineering, and physics, allowing comprehensive visualization, segmentation, and quantitative analysis of internal microstructures, compositional gradients, and dynamic processes.

1. Fundamental Principles and Modalities

Synchrotron-based tomography typically involves the acquisition of a series of projection images as a sample is rotated with respect to a monochromatic or polychromatic synchrotron beam. High photon flux and coherence provide several imaging contrasts:

• Absorption Contrast: Relies on differential X-ray attenuation (predominant for high-Z or heavily absorbing samples).
• Phase Contrast: Leverages the phase shift of X-rays due to differences in refractive index, dramatically enhancing sensitivity to low-Z and weakly absorbing specimens. Inline phase-contrast methods, such as those utilizing Paganin retrieval filters, are routine for soft matter, biological tissues, and polymer/organic samples (Weitkamp et al., 2020, Quereilhac et al., 2023, Donato et al., 16 Dec 2024).
• Chemical and Spectroscopic Tomography: By tuning X-ray energies near elemental absorption edges and collecting tomograms at multiple energies, chemical mapping or speciation in 3D is achieved (e.g., spectrotomography with soft X-ray scanning transmission setups) (Leontowich et al., 2018).
• Magnetic Vector Tomography: Utilizing X-ray magnetic circular dichroism (XMCD) in a tomographic context allows vector-resolved magnetic imaging at the nanoscale, facilitated by advanced tools such as MARTApp (Herguedas-Alonso et al., 21 Jan 2025).

Instrumentation includes full-field (parallel beam) systems for large volumetric samples and focused-beam modalities for sub-micron to nanometer resolution, as exemplified by zone-plate-based transmission X-ray microscopes (Leontowich et al., 2018, Weitkamp et al., 2020).

2. Data Acquisition, Reconstruction, and Quantitative Analysis

The synchrotron environment provides significant advantages for rapid, high-fidelity tomographic imaging:

• Rapid Scanning and Temporal Resolution: High flux and effective detector synchronization enable acquisition of thousands of projections in seconds to minutes, supporting in situ and time-resolved studies—including MHz-rate 3D imaging using multi-beam configurations like XMPI (Bellucci et al., 8 Feb 2024).
• Three-dimensional Reconstruction: Common reconstruction approaches include filtered backprojection (FBP), Fourier-based methods (e.g., Gridrec), and advanced algebraic or iterative techniques such as customized SART (cSART) and Unified Tomographic Reconstruction (UTR) (Donato et al., 16 Dec 2024, Ganguly et al., 2021). Implementation-specific discretization and filtering variants influence reproducibility, necessitating calibration routines such as implementation-adapted filters to minimize inter-software discrepancies (Ganguly et al., 2021).
• Phase and Contrast Enhancement: Reconstruction may integrate Paganin’s TIE-Hom phase retrieval for phase-contrast datasets, with parameter tuning ensuring the balance between noise suppression and spatial resolution (Donato et al., 16 Dec 2024, Quereilhac et al., 2023).
• Quantitative Morphometry: Extracted morphological parameters include equivalent diameter ($D_{eq}$), tortuosity ($\tau$), specific surface area ($S_v$), sphericity, and curvature metrics, directly linking 3D structure to physical or mechanical properties (Zhao et al., 2018, Zhao et al., 2018, Zhao et al., 2019).

3. Application Domains and Case Studies

Materials Science and Alloys

• Microstructure–Property Correlation: Synchrotron X-ray microtomography was employed to characterize Fe-rich intermetallics and pore networks in Al–Cu alloys, revealing that increased Fe content decreases interconnectivity and equivalent diameter of Fe phases while generating sharp, stress-concentrating morphologies that reduce mechanical performance (Zhao et al., 2018).
• Processing Effects: Ultrasonic melt processing (USP) analyzed by synchrotron microtomography demonstrated grain refinement, reduced phase interconnectivity, and lower porosity, all quantifiable by 3D morphometry (Zhao et al., 2018).
• Phase Nucleation and Growth: Synchrotron tomography combined with electron microscopy established the morphology, distribution, and nucleation sites (e.g., $\alpha$-Al$_2$O$_3$) for Al$_3$Sc phases in Al-Sc alloys, elucidating their role in grain refinement via direct visualization and orientation relationship mapping (Zhao et al., 2019).

Life Sciences and Biomedicine

• Organ-scale Imaging at Cellular Resolution: Whole mouse brains imaged at 0.65 $\mu$m voxel size (3.3 teravoxels) enable digital histology and precise anatomical registration to resources such as the Allen Mouse Brain Atlas, greatly expanding brain atlas fidelity and accessibility (Humbel et al., 22 May 2024).
• In Vivo CNS Fluid Space Imaging: SR$\mu$CT provides dynamic, high-resolution visualization (6.3 $\mu$m voxels) of cerebrospinal fluid spaces, their solute distribution, and tissue motion in live mice, outperforming both MRI and histology for integrated anatomical and functional analysis (Alarcón et al., 3 Jul 2025).
• 3D Defect Mapping in Biomaterials: Phase-contrast microtomography uncovers ultrastructural defects (e.g., kink-band-induced pores) in flax fibres, revealing their concentric organization, sub-micron dimensions, and spatial correlation with mechanical property degradations (Quereilhac et al., 2023).

Dynamic Phenomena and Process Imaging

• High-speed Volumetric Probing: X-ray Multi-Projection Imaging (XMPI) with beam-splitting optics realizes volumetric imaging at kHz–MHz rates without rotation, suitable for capturing stochastic events such as fracture dynamics, shock propagation, or rapid biological processes inaccessible to conventional tomography (Bellucci et al., 8 Feb 2024).
• In Situ Environmental Control: Contactless, modular high-temperature furnaces integrated with tomography stages facilitate real-time imaging of thermal transformations, as demonstrated by grain growth and strain relaxation in annealed Al1050 monitored with dark-field X-ray microscopy (Lesage et al., 1 Jul 2025).

4. Advanced Image Processing, Segmentation, and Analysis Workflows

• Deep Learning Enhancement and Segmentation: Generative adversarial networks (TomoGAN), residual U-Nets, SegFormer-based transformers, and attention-equipped architectures (ResAttUnet) are applied to denoise, reconstruct, or segment low-dose or artifact-laden synchrotron tomography. These approaches improve image quality, segmentation performance (MeanIoU up to ~84% on in situ data), enable efficient processing (e.g., 30 s per 3D volume), and facilitate generalization across changing sample morphologies (Liu et al., 2019, Wu et al., 2020, Li et al., 2021, Manchester et al., 27 Apr 2025).
• Segmentation in Complex or Low-dose Scenarios: Ex situ–to–in situ domain adaptation mitigates limited training data or poor image quality by simulating synchrotron artifacts on pristine ex situ scans for deep network training. Modified SegFormer architectures support high-throughput binary or multiphase segmentation and delineation of evolving structures in dynamic experiments, such as copper oxide dissolution (Manchester et al., 27 Apr 2025).
• Quantitative Visualization and Data Exploration: High-dimensional secondary data (e.g., particle morphometrics from >20 million objects) can be explored using AccuStripes, which apply adaptive binning and composition strategies to histogram representations, crucial for spatially resolved, tile-wise comparison in massive tomographic datasets (Heim et al., 15 May 2025).

5. Technical Innovations in Instrumentation and Facility Design

• Cryo-Optical Integration: Cryo-STXM instruments operating from 100–4000 eV enable 30–75 nm spatial resolution, high spectral sensitivity, cryogenic sample handling, and near-diffraction-limited performance, supporting advanced 3D chemical imaging (“4D spectrotomography”) (Leontowich et al., 2018).
• Flexible Beamline Engineering: Multi-modal setups (ANATOMIX at Synchrotron SOLEIL) offer absorption and phase-contrast imaging across 20 nm to 20 $\mu$m scales, variable beam sizes up to 40 mm, and high-speed acquisition compatible with in situ mechanical, environmental, or thermal experiments (Weitkamp et al., 2020, Lesage et al., 1 Jul 2025).

6. Limitations, Reproducibility, and Data Accessibility

• Reproducibility Concerns: Implementation-specific differences in reconstruction algorithms (such as filtering or backprojection nuances) lead to systematic quantitative discrepancies between software solutions. Optimization of implementation-adapted filters enables more comparable results across platforms and improves segmentation consistency (Ganguly et al., 2021).
• Handling of Teravoxel-Scale Data: Hierarchical, multi-resolution data architectures, blockwise non-rigid registration, and public repositories with browser-based access (e.g., Neuroglancer, siibra-explorer) address the practical challenges of making multi-terabyte datasets accessible, navigable, and usable for the broader research community (Humbel et al., 22 May 2024).
• Radiation Dose, Resolution Trade-offs, and Domain Adaptation: Advanced approaches (GANs, hybrid-dose measurements, ResAttUnet) enable maintaining high resolution at lower exposure, reducing sample damage, and shortening acquisition time, at the cost of algorithmic complexity and increased processing demands (Liu et al., 2019, Wu et al., 2020, Li et al., 2021, Manchester et al., 27 Apr 2025).

7. Future Perspectives

Ongoing developments in synchrotron-based tomography include further advances in spatial, temporal, and chemical sensitivity, integration of AI-powered segmentation in ultra-large datasets, expansion into multi-modal imaging (combining absorption, phase, magnetic, and chemical contrasts), and engineering of sample environments for in situ and operando studies at extreme conditions. Optimization of data processing pipelines,, continued efforts to standardize reconstruction for multi-facility reproducibility, and broader accessibility through online interactive tools represent essential directions for maximizing the impact of synchrotron tomography throughout scientific disciplines.