Synchrotron X-Ray Multi-Projection Imaging (XMPI) for High-Resolution 4D Characterization of Multiphase Flows

This paper introduces Synchrotron X-ray Multi-Projection Imaging (XMPI), a novel, rotation-free technique that enables high-resolution, four-dimensional tracking of microparticles in opaque multiphase flows by capturing simultaneous multi-angle projections, thereby overcoming previous limitations in observing microscale dynamics for applications in rheology, medicine, and materials science.

The Gist

Imagine trying to watch a single grain of sand swim through a jar of thick, muddy honey. If you look at it with your eyes, you see nothing but a blur of brown. If you try to use a standard camera, the mud blocks the light. Even if you could see through the mud, most 3D cameras require you to spin the jar around to get a full picture. But if you spin the jar, you change how the sand moves, ruining the experiment.

This is the problem scientists have faced for years when studying "multiphase flows"—mixtures where tiny particles, bubbles, or droplets float inside a fluid. These mixtures are everywhere: in blood, paint, ketchup, and even lava. Understanding how the tiny bits move inside these thick, opaque liquids is crucial, but it has been nearly impossible to see without disturbing them.

The New "Magic Flashlight"

The researchers in this paper have built a new tool called XMPI (Synchrotron X-ray Multi-Projection Imaging) that solves this puzzle. Here is how it works, using a simple analogy:

Think of a standard X-ray machine as a single flashlight shining through a wall. You get a flat, 2D shadow. To get a 3D picture, you usually have to rotate the object (like a CT scan at the hospital).

The XMPI team, however, used a super-powerful "flashlight" at a massive research facility called MAX IV in Sweden. Instead of one beam, they used special crystals to split one beam of X-rays into two separate beams, like a prism splitting white light into a rainbow. These two beams hit the sample from two different angles at the exact same time.

• The Setup: Imagine holding two flashlights at different angles, shining them through a jar of muddy blood simultaneously.
• The Result: Two cameras on the other side catch two different "shadows" (projections) at the exact same instant.
• The Magic: Because they have two views at once, they can mathematically figure out exactly where every single tiny particle is in 3D space, without ever having to spin the jar.

What They Actually Saw

The team tested this on two very different "muddy" liquids:

1. Glycerol (Thick Syrup): They mixed tiny, hollow glass beads (about the width of a human hair) into thick glycerol. Because the beads are hollow, the X-rays passed through them differently than the liquid, making them stand out like glowing dots. They successfully tracked hundreds of these beads as they flowed, creating a 4D movie (3D space + time) of their paths.
2. Human Blood: This is the real challenge. Blood is opaque and thick. You cannot see through it with a normal camera. However, the X-rays pierced right through. Even though the red blood cells themselves were too small to see individually, the tiny glass beads floated inside the blood were clearly visible. The team tracked these beads as they swam through the blood, proving the method works even in the most difficult, "muddy" fluids.

Why This Matters (According to the Paper)

The paper highlights three main achievements:

• No Spinning Required: They can watch fast-moving fluids in real-time without rotating the sample, which means they don't accidentally create fake currents by spinning the jar.
• Seeing the Invisible: They can track individual particles in fluids that are completely opaque to light (like blood or paint), which was previously impossible.
• Two Ways to Look:
  • The "Spotter" Method: In thinner mixtures, they tracked individual particles one by one (like following specific runners in a race).
  • The "Flow Map" Method: In very thick, crowded mixtures where you can't see individual beads, they used a computer vision technique called "Optical Flow." This is like looking at a crowd of people and seeing the general direction the crowd is moving, even if you can't pick out one specific person.

The Bottom Line

This paper doesn't claim to cure diseases or build new engines yet. Instead, it claims to have built a new "eye" that can see inside thick, dark, moving fluids. By splitting X-rays into two beams, they created a way to take high-speed, 3D movies of tiny particles flowing through opaque liquids like blood and syrup, all without ever touching or spinning the sample. This gives scientists a new, clear window into the microscopic world of fluids that was previously hidden in the dark.

Technical Summary

Technical Summary: Synchrotron X-Ray Multi-Projection Imaging (XMPI) for High-Resolution 4D Characterization of Multiphase Flows

Problem Statement
Multiphase flows involving particles, bubbles, or droplets suspended in fluids are critical to processes in biology, medicine, and industry (e.g., blood flow, papermaking, and concrete mixing). However, observing the microscale dynamics of these systems within opaque fluids remains a fundamental challenge. Optical methods are restricted to transparent systems or require index-matching strategies often incompatible with real-world suspensions. While techniques like Ultrasound Doppler Velimetry (UDV) and Magnetic Resonance Velocimetry (MRV) can access opaque systems, they typically offer limited spatial or temporal resolution and provide only averaged or indirect flow information. Furthermore, standard time-resolved X-ray Computed Tomography (XCT) requires sample rotation, which limits applicability to slow flows and introduces inertial artifacts in the rotating frame.

Methodology
The authors present Synchrotron X-ray Multi-Projection Imaging (XMPI), a novel approach enabling four-dimensional (3D + time) tracking of microparticles in dense suspension flows without sample rotation. The experiments were conducted at the ForMAX beamline of the MAX IV synchrotron facility (Lund, Sweden).

• Experimental Setup: A direct X-ray beam (16.55 keV) was split into two angularly separated beamlets (~48° apart) using Bragg reflection from Silicon and Germanium crystals. These two beams simultaneously illuminated a stationary sample (a Kapton tube with an inner diameter of ~0.72 mm) containing flowing suspensions.
• Detection: Two identical indirect X-ray microscopes, equipped with LuAG:Ce scintillators and sCMOS cameras, captured simultaneous projections from the two angles. The effective pixel size was 1.3 µm, and the system operated at 40 Hz.
• Samples: The study utilized Silver-coated Hollow Borosilicate Glass Spheres (SHGS, mean diameter 10 µm) suspended in two media:
  1. Glycerol (dilute and dense concentrations) to demonstrate tracking in a controlled fluid.
  2. Human whole blood (~40% hematocrit) to demonstrate capability in highly opaque, complex biological fluids.
• Data Analysis:
  • Dilute Suspensions: Individual particle tracking was performed using the open-source MyPTV library. Particles were identified in both projections and matched to reconstruct 3D trajectories and velocities.
  • Dense Suspensions: Due to the difficulty of tracking individual particles in crowded fields, the authors applied Optical Flow (OF) analysis using the RAFT algorithm in MATLAB to estimate projected velocity fields. These were compared against theoretical Poiseuille flow profiles.

Key Contributions

1. Rotation-Free 4D Imaging: The paper demonstrates the first application of XMPI at a synchrotron source for multiphase flows, achieving time-resolved, micrometer-scale 4D imaging of opaque fluids without the need for sample rotation.
2. Tracking in Opaque Media: The method successfully resolved individual microparticle positions and trajectories in both glycerol and human blood, overcoming the opacity limitations of optical techniques.
3. Statistical Flow Characterization: The study validated the extraction of statistical flow properties, including velocity profiles and particle number density distributions, in both dilute (via 3D Particle Tracking Velocimetry) and dense (via Optical Flow) regimes.
4. Validation of CFD Models: The high-resolution experimental data provides ground truth for validating particle-resolved Computational Fluid Dynamics (CFD) models, particularly for complex phenomena like inertial focusing and microstructural dynamics.

Results

• Dilute Flow (Glycerol): The system tracked over 2,400 particles in a 0.1 wt.% suspension. The reconstructed 3D trajectories confirmed a laminar Poiseuille velocity profile. A particle-free layer near the tube wall was observed, attributed to non-ideal upstream conditions or wall fouling rather than inertial focusing (given the low Reynolds number, $Re \approx 4.7 \times 10^{-5}$).
• Biological Flow (Blood): Individual SHGS particles were successfully tracked within human blood. While red blood cells (RBCs) were not directly resolved as distinct particles, a speckle pattern generated by the RBCs was visible in both projections, consistent with previous X-ray blood flow studies.
• Dense Flow (10 wt.% Glycerol): Individual particle tracking was challenging due to particle overlap. However, Optical Flow analysis successfully generated projected velocity profiles that closely matched the theoretical Poiseuille profile when accounting for a 0.02 mm particle-free layer. The agreement between the two projection angles confirmed the cylindrical symmetry of the flow.
• Performance Limits: The current setup (40 Hz) limits the maximum trackable velocity to ~20 mm/s to avoid motion blur. The authors note that with faster cameras (kHz range) and higher flux sources, the system could theoretically probe velocities up to 8000 mm/s.

Significance and Claims
The authors claim that XMPI opens a new spatiotemporal frontier for high-speed, rotation-free microtomography of complex multiphase flows. By combining XMPI with recent AI-supported reconstruction algorithms and sparse-projection techniques, the method offers a pathway to study time-resolved micron-sized features in dense suspensions at kHz acquisition rates.

The paper emphasizes that this methodology provides previously inaccessible experimental validation for CFD models and allows for the observation of microstructural dynamics relevant to suspension rheology and biomedical flows. While the current study is modest in its temporal resolution (40 Hz) and field of view (<10 mm), the authors posit that the technique is scalable. They envision future applications extending beyond particle suspensions to gas-liquid flows (foams) and immiscible liquid-liquid flows (emulsions), provided that higher flux densities and faster detection schemes are utilized. The work establishes a proof-of-concept for using synchrotron-based multi-projection imaging to resolve the 4D dynamics of opaque multiphase systems.