The stakes are far higher than a technical quibble. Ground ice is the linchpin of permafrost stability. When it thaws, the ground can deform, subside, and collapse, damaging roads, pipelines, and buildings across the Arctic. Accurately knowing how much ice a given volume of soil holds is therefore fundamental to predicting thaw-related ground deformation and to engineering in cold regions. For decades, that knowledge came from visual descriptions, in which field geologists classified ice by hardness, structure, and colour. Those descriptive systems, dating back to the 1960s, proved unreliable and inconsistent, and a more modern cryostratigraphic approach, which classifies ice by its amount and spatial distribution, still depends on subjective visual interpretation. Laboratory measurement of volumetric ice content, based on weighing and drying samples, is accurate but destructive, slow, expensive, and limited in spatial coverage.

Computed tomography promised to break that bottleneck. By scanning frozen cores non-destructively, CT offers detailed three-dimensional images and the tantalizing possibility of computing the fractions of ice, sediment, and gas directly from density values. Medical CT scanners operate at millimetre to sub-millimetre resolution, while micro-CT systems resolve far finer pore-scale structures. But translating greyscale images into reliable estimates of soil components is anything but straightforward, and the new study quantifies exactly how badly the translation can go wrong.

The researchers collected permafrost cores on Bylot Island in Nunavut and near Beaver Creek in Yukon using a portable drill with a 10-centimetre barrel, keeping the samples below zero degrees Celsius from field to laboratory. The cores were subsampled into segments of roughly 5 to 20 centimetres and scanned at the Institut national de la recherche scientifique in Quebec City using a Siemens SOMATOM Sensation 64 medical scanner, with voxel resolutions of 0.18 to 0.24 millimetres in the horizontal plane and 0.4 to 0.6 millimetres vertically. Samples stayed frozen throughout, wrapped in plastic and foam to limit heat exchange, and no phase change was observed during the brief handling periods.

The physics of CT imaging is where the trouble begins. A CT image is built from voxels, three-dimensional pixels stacked from two-dimensional slices, and each voxel records a single averaged density value. A feature must be at least twice the voxel size to be detected at all. When a single voxel contains more than one material, their densities are averaged in what is known as the partial volume effect. In permafrost, pore ice occupying the tiny spaces between fine sediment grains is typically far smaller than a medical CT voxel, so its signal vanishes into the surrounding sediment, and the averaged density falls outside the range assigned to ice. The result is a systematic underestimation of ice content in sediment-rich samples.

Organic matter adds a second, opposite bias. The density of organic material spans a wide range depending on its degree of decomposition, from fibrous peat to fully humified soils, and portions of that range overlap with the density window assigned to ice in standard segmentation schemes. When researchers classify voxels by Hounsfield Unit thresholds, the standardised density scale of CT imaging, organic matter can be misclassified as ice, inflating the apparent ice content. The study found that ice content was underestimated in all low-organic-matter samples and overestimated in nearly all high-organic-matter samples, with the largest positive bias in samples containing more than 20 percent organic matter.

Across all 261 samples, the CT estimates were positively correlated with laboratory values but diverged systematically from the ideal one-to-one line. The mean bias was minus 12.9 percent, with a root mean squared deviation of 25.6 percent, meaning typical errors were enormous relative to the quantities being measured. Residuals depended strongly on composition: for samples with laboratory-measured ice content below about 75 percent, CT generally underestimated the true value, with negative residuals exceeding 60 percent in some cases. Above roughly 75 percent ice content, a tendency toward overestimation emerged. Agreement improved only when ice contents exceeded about 75 percent and organic matter sat in the 10 to 20 percent window, a narrow sweet spot that few natural samples occupy.

The team also tested how sensitive the results were to the choice of Hounsfield Unit thresholds, applying published threshold ranges from previous studies to two representative cores. The outcome was striking: estimated ice content swung from severe underestimation of up to minus 64 percent to marked overestimation of up to plus 31 percent relative to laboratory values, depending solely on which thresholds were used. Because permafrost is so compositionally heterogeneous, unlike relatively uniform sea ice, no single or transferable threshold can be defined for its materials. HU-based segmentation is inherently non-unique, and even calibrated corrections cannot overcome the fundamental limits imposed by scanner resolution and overlapping material densities.

The authors conclude that medical CT cannot reliably quantify volumetric ice content and should not be recommended for that purpose, given the errors, the cost, and the processing effort involved. Its genuine value lies elsewhere: in qualitative visualization of cryostructures, the distinctive patterns of lenticular, layered, reticulate, and suspended ice, sedimentary stratifications, and large-scale heterogeneity within cores, all of which are critical for interpreting how permafrost formed and how it will behave mechanically. Higher-resolution micro-CT systems may reduce partial-volume effects by resolving finer structures, and approaches based on sample-specific calibration or density normalization may partially improve estimates, but whether higher-resolution imaging can deliver accurate quantification remains an open question the authors identify as a key direction for future research.

For a field racing to understand how rapidly thawing Arctic ground will release water, reshape landscapes, and destabilize infrastructure, the message is clear. The scanner that so beautifully reveals the hidden architecture of frozen ground cannot be trusted to count the ice within it, and researchers who need numbers, not pictures, will still have to reach for the oven and the scale. The full dataset of laboratory and CT-derived measurements has been released on Zenodo, giving the community a benchmark for testing the next generation of imaging approaches.

Subject of Research: Accuracy of medical X-ray computed tomography for estimating volumetric ice content in permafrost samples

Article Title: Brief communication: Limitations of medical X-ray computed tomography for estimating ice content in permafrost samples

Article References: Roustaei, M., Darey, J., Mohammadi, Z., & Fortier, D. (2026). Brief communication: Limitations of medical X-ray computed tomography for estimating ice content in permafrost samples. The Cryosphere, 20(9), 5265-5270. https://doi.org/10.5194/tc-20-5265-2026

Image Credits: AI Generated

DOI: 10.5194/tc-20-5265-2026

Keywords: permafrost, ground ice, X-ray computed tomography, cryostratigraphy, Hounsfield Unit segmentation, partial volume effect, volumetric ice content, organic matter, Canadian Arctic, thaw settlement, micro-CT, The Cryosphere

The stakes are higher than they might appear. Grain morphometry underpins a remarkable breadth of Earth science. The sizes and shapes of clasts record how sediments are produced on hillslopes, how rivers hydraulically sort and transport their loads, and how deposits in the stratigraphic archive were laid down. Grain size data have even been used to reconstruct transport conditions on Mars, where the shapes of ancient pebbles hint at how far water once flowed. But the traditional ways of collecting these data are laborious and biased. Manual Wolman counts and line sampling yield relatively few observations and depend heavily on the observer. Point-cloud methods struggle with grain occlusion and require geometric assumptions. Earlier image-based tools either need field calibration or systematic manual correction, and they tend to over- or underestimate grain sizes in predictable but troublesome ways.

Machine learning promised to change this, and to some extent it has. Texture-based neural networks can predict summary statistics of a grain size distribution directly from image texture, while segmentation-based approaches delineate every grain individually, opening the door to datasets of tens of thousands of measurements. The catch, as Mair and colleagues emphasize, is that these segmentation models have behaved like narrow specialists. Trained on homogeneous, task-specific datasets, they lose accuracy the moment they encounter a new camera, a new lighting regime or a new sediment type. Each new setting has effectively demanded a new model and a freshly curated training set, which has blunted much of the promised efficiency.

The Bern team’s solution was to reach for a different class of model altogether. In recent years, transformer-based foundation models such as Meta’s Segment Anything Model, or SAM, trained on eleven million images and more than a billion annotation masks, have shown an extraordinary ability to generalize to unseen data. But SAM in its basic form is inefficient at densely segmenting large numbers of similar objects, which is precisely what a gravel bar photograph demands. The breakthrough came with Cellpose-SAM, a next-generation model developed for segmenting cells in biomedical imagery. It grafts SAM’s transformer image encoder onto the flow-based segmentation machinery of the original Cellpose framework, which tracks predicted vector flows to separate touching objects. The hybrid keeps SAM’s generalization while excelling at dense, prompt-free segmentation of many instances of one object class, and its architecture has grown to more than 304 million trainable parameters, roughly fifty times larger than the U-Net backbone it replaces.

Retraining such a model for geoscience required data, and lots of carefully annotated ones. The researchers assembled the ImageGrains 2.0 dataset, combining 81 previously labelled tiles with 162 new ones for a total of 243 image tiles containing more than 29,000 manually annotated grain masks. The imagery spans fluvial gravels photographed from uncrewed aerial vehicles and handheld cameras in Switzerland, Spain, Peru, New Zealand and Namibia; painted and unpainted pebbles used to track grain mobility; angular proglacial deposits; near-vertical outcrops of lithified conglomerates; X-ray computed tomography slices of glacial till; and micro-CT images of bioclastic marine sand. Crucially, the team deliberately included awkward content, such as vegetation, sieves, hands, scale objects, water bodies and harsh shadows, to test whether the model could distinguish genuine grains from everything else in a messy field photograph.

The performance gains are substantial. Benchmarked against the manual annotations using average precision at an intersection-over-union threshold of 0.5, the fine-tuned Cellpose-SAM model achieved a median score of 0.72 across all image tiles, roughly 20 percent higher than the second-best method, a Cellpose 2 model trained on the same data. The advantage held in both training and unseen test splits, and it proved largely independent of how many images each sediment type contributed to training. The model scored highest of all on X-ray CT tiles, which made up only about a tenth of the dataset. Even more telling was the generalization test: when the researchers withheld two entire subsets from training, the model still matched or beat specialist models that had been fine-tuned on those very images, evidence that the transfer-learning strategy preserves the broad visual competence of its SAM heritage.

Accurate masks translate directly into accurate measurements, and here the numbers are striking. The model recovered 93 percent of ground-truth grains, and mean differences in grain size were around 3 percent or less for both the long and short axes of ellipse fits to each grain. The resulting grain size distributions were statistically indistinguishable from the ground truth in 88 percent of cases for the a-axis and 82 percent for the b-axis, compared with only 54 and 57 percent for the second-best model. Shape metrics fared equally well, with average deviations below 2 percent for roundness, roughness, elongation and orientation. The team also identified a practical rule of thumb: perfect statistical agreement between predicted and true grain size distributions was only achieved when segmentation scores exceeded roughly 0.68 average precision, a threshold that varied with how well sorted the sediment was.

The framework extends into three dimensions as well. Applying the model to a stack of 400 X-ray CT slices from a drill core of glacio-fluvial sediment in southern Germany, the researchers segmented 4,647 well-defined coarse grains in full volumetric detail, something 2D photographs can never provide. Because CT scans capture complete grains rather than partially occluded surfaces, they sidestep the geometric extrapolations that plague LiDAR and photogrammetry-based approaches. The authors note that rigorous 3D benchmarking still awaits a manually annotated volumetric ground truth, whose creation has so far been impractical, but the demonstration suggests CT-based sediment analysis, from clast fabric to depositional history, is now within reach of automated workflows.

Accessibility was a design priority. ImageGrains 2.0 ships as an installable Python package with an interactive graphical user interface for annotating images and correcting model predictions, so researchers need not write code at all. Training runs in under ninety minutes on a high-end GPU, and although transformer models are demanding, the authors show that fine-tuning remains feasible on a mid-range desktop graphics card, with inference requiring as little as 3 gigabytes of memory. Because the model builds on already-trained foundation weights, its marginal energy footprint is modest compared with training such architectures from scratch. Adapting the model to a genuinely new sediment type or imaging style can require as few as seven additional annotated tiles, a dramatic lowering of the barrier that once made every new site a modelling project of its own.

The deeper message of the study concerns where progress will come from next. The authors argue that future gains in grain measurement will depend less on new architectures than on larger, more variable, publicly shared annotated datasets, ideally with annotations from multiple experts to establish consensus-level benchmarks. They explicitly frame their work as a step toward a foundation model for granular particles in geoscientific imagery, one that could serve any discipline needing dense segmentation and morphometric analysis, from sedimentology to planetary science. With the dataset, model weights and code all openly available, the Bern team has effectively invited the community to build on their groundwork, turning what was once a tedious field chore into a problem that a well-fed neural network can now handle in minutes.

Subject of Research: Deep-learning-based automated segmentation and morphometric measurement of sediment grains in geoscientific imagery

Article Title: ImageGrains 2.0: Improved precision and generalization for grain segmentation

Article References: Mair, D., Witz, G., Do Prado, A., Garefalakis, P., Wild, A., Ville, F., Schuster, B., Horn, M., Österle, J., Fabbri, S. C., Litty, C., Achleitner, S., Leistner, S., Hiller, C., & Schlunegger, F. (2026). ImageGrains 2.0: Improved precision and generalization for grain segmentation. Earth Surface Dynamics, 14(4), 527-551. https://doi.org/10.5194/esurf-14-527-2026

Image Credits: AI Generated

DOI: 10.5194/esurf-14-527-2026

Keywords: ImageGrains 2.0, grain segmentation, Cellpose-SAM, deep learning, sedimentology, grain size distribution, X-ray computed tomography, transfer learning, Segment Anything Model, fluvial sediment, geomorphology, open-source software

X-ray computed tomography, or XCT, has become a routine first step in core processing on board the Chikyu. The method is non-destructive, which makes it ideal for an early look at the three-dimensional structure of rocks and sediments before anyone commits to cutting, sampling, or splitting them. A medical-grade scanner, a GE Healthcare Discovery CT 750HG, is installed on the vessel. Core sections of roughly 1.5 meters are laid on a translating table and advanced through the scanner while an emitter-receiver pair rotates around them. The X-ray beam is attenuated as it passes through the sample through a combination of Compton scattering and the photoelectric effect, and the resulting attenuation coefficient is almost proportional to density. From hundreds of projections acquired at 360 degrees around the core axis, a computer reconstructs a full 3D volume that maps the density of the rock, slice by slice, from top to bottom.

The raw data, however, are unwieldy. Each core section produces DICOM files, the standard format for medical tomography, weighing several hundred megabytes apiece. Horizontal slices have a pixel size of roughly 175 micrometers, while consecutive slices are spaced 625 micrometers apart along the core axis. On board the ship, these volumes could only be explored with specialized software running on a handful of dedicated Macintosh computers located several meters from the core description tables. Scientists found themselves walking back and forth between the cores and the workstations, and checking whether a fracture or boundary was truly planar meant endlessly scrolling through parallel slices. The XCT images, in principle the most continuous and undisturbed record of the recovered material, had become a source of friction in a workflow that runs around the clock during an expedition.

Doan and colleagues propose a visual summary that strips away this friction. Their method, which they describe as unwrapping, converts the cylindrical 3D volume into a single flat 2D image, in much the same way that the Shroud of Turin is said to record a body wrapped around a cylinder. For every iso-Z slice perpendicular to the core axis, the XCT values are interpolated along a circular path centered on the core, traveling counter-clockwise to match the conventions used in borehole imaging. The circle radius is typically set to 64 pixels so that the path stays inside the core liner and samples only rock material. Recording the attenuation values along this circle for every depth produces an image in which the horizontal axis is angle around the core and the vertical axis is depth. Planar features such as fractures and lithological boundaries, which appear as complicated curves in individual slices, unfold into characteristic sinusoidal traces that trained eyes can measure directly.

The implementation was deliberately low-tech in spirit. The team wrote Python scripts embedded in Jupyter Notebook files that read the DICOM data and generate the summary figures. On a 2020 MacBook Pro laptop, producing both the unwrapped image and the longitudinal slices of a core section took less than three minutes. The outputs were then printed on paper and delivered to two key groups: the XCT watchdog scientist, who selects whole-round core intervals for priority sampling before the cores are split, and the structural geology and sedimentology teams who conduct visual core description. Because a single color range, spanning XCT values from 0 to 3000, was applied to all cores during the expedition, the printed images also allowed scientists to track how attenuation, and therefore density, changed with depth across the entire drilled sequence.

The test bed for the method was the first window of IODP Expedition 405, carried out in late 2024 as part of the JTRACK project tracking tsunamigenic slip across the Japan Trench. The unwrapped images quickly proved their worth across the scientific party. The structure team used them to identify and measure the orientation of planar features that were difficult or impossible to distinguish from the cut surface of the core. The sedimentology team relied on them as a continuous reference for tracing fine-scale lithofacies variations, defining interval boundaries, and choosing spots for smear-slide sampling. In one striking case, a thick clast of ash resting at the bottom of the archive half, entirely hidden from the surface, would have gone unnoticed without the XCT summary. The physical properties team found that subtle variations in XCT values flagged heterogeneities throughout the core, and that distinct mineral phases such as pyrite and calcite stood out as bright, strongly absorbing regions.

Beyond the immediate operational gains, the unwrapped images opened a door to more rigorous quantitative analysis. Because they are constructed along a cylinder, they resemble the electrical and acoustic borehole images produced by downhole logging tools, and they can be loaded directly into software such as TechLog that picks sinusoidal features automatically. This allowed the team to measure the dip and relative azimuth of planar structures along the borehole faster and more precisely than manual measurements on the split working halves. Crucially, the XCT scans were collected before the cores were split, preserving the geometry of structures with minimal disturbance and yielding more reliable orientation data. Comparing unwrapped core images with borehole images also helped reorient cores where bedding could be recognized and facilitated correlating structural features between core and log, even though each dataset has limitations: borehole images are oriented and quasi-continuous but lower in resolution, while cores are high-resolution but incomplete, since recovery is rarely a full 100 percent.

The authors are careful to note that the printed XCT summaries complement rather than replace direct observation of the cores. Fine features such as slickensides, the polished fault surfaces that record slip, are invisible on X-ray images, and different lithologies can share similar densities and therefore similar XCT values, making them hard to tell apart. Visual core description at the table remains a necessity. What the unwrapping method changes is the efficiency and continuity of that description: scientists no longer need to interrupt their work to consult a distant workstation, and they can annotate a paper summary with measurements as they go, keeping the digital 3D volumes available for later, more detailed analysis.

The expedition also exposed practical hurdles. Unwrapping was not originally part of the Expedition 405 workflow, and early access to the DICOM files was not planned. Generating the images required lengthy manual copying operations so that data could be processed on an external computer disconnected from the onboard laboratory network, and this maintenance burden was high enough that usage of the summaries tended to lapse during the second window of the expedition. The authors argue that installing the Python scripts directly on a dedicated computer on board would turn XCT unwrapping into an almost real-time service running between scanning and visual core description, automatically producing depth-referenced unwrapped images and combined summaries that could also incorporate data from the Multi-Sensor Core Logger to cross-validate measurements and guide sampling.

If that integration succeeds, the humble printed image may become a standard tool of scientific ocean drilling. The method’s code and notebooks have been released openly through a Zenodo archive, lowering the barrier for other drilling platforms and even land-based core repositories to adopt the approach. For a discipline in which decisions about irreplaceable samples must often be made within hours of the core arriving on deck, a technique that turns a half-gigabyte 3D scan into a sheet of paper in under three minutes is more than a convenience. It is a way of putting the invisible interior of the Earth, quite literally, into the hands of the scientists who need to read it first.

Subject of Research: Real-time unwrapping of 3D X-ray CT scans of scientific drilling cores for visual core description

Article Title: Real-time unwrapping of 3D X-ray CT scans for visual core description

Article References: Doan, M.-L., Brunet, M., Pizer, C., Okuda, H., Chang, Y.-C., Satolli, S., Nicholson, U., Yamamoto, Y., Conin, M., Fukuchi, R., Kirkpatrick, J., Toczko, S., & the IODP Expedition 405 Scientists (2026). Real-time unwrapping of 3D X-ray CT scans for visual core description. Scientific Drilling, 35(1), 55-60. https://doi.org/10.5194/sd-35-55-2026

Image Credits: AI Generated

DOI: 10.5194/sd-35-55-2026

Keywords: X-ray computed tomography, scientific ocean drilling, IODP Expedition 405, D/V Chikyu, visual core description, core scanning, image unwrapping, structural geology, sedimentology, borehole imaging, Scientific Drilling, Japan Trench

The technique at the heart of the story is X-ray computed tomography, or XCT, the same fundamental technology used in hospitals to image the human body. When a drill core arrives on the Chikyu’s core processing deck, it is loaded into an onboard medical-grade scanner, a Revolution Frontier VT1700 built by GE HealthCare. The machine rotates an X-ray source around the section and reconstructs a three-dimensional map of how strongly the material attenuates X-rays. Because attenuation depends on both density and elemental composition, the resulting images reveal features that are completely invisible from the outside: burrows left by seafloor organisms, the sharp bases of turbidite deposits, subtle faults and deformation bands, and even the ghostly outlines of gas-filled voids.

Physically, the scanner reconstructs the distribution of the linear attenuation coefficient, conventionally expressed as CT numbers in Hounsfield units, a scale on which water registers zero and air registers minus one thousand. Dense minerals or materials rich in heavy elements produce high values, so discontinuities defined by density contrasts pop out of the images. The Chikyu scanner produces two-dimensional slices of 512 by 512 pixels across a 90-millimetre field of view, giving a resolution of about 0.176 millimetres within each slice, with slices spaced 0.625 millimetres apart. A full-length core section of roughly 1.4 metres therefore consists of some 2,200 to 2,300 individual slices, each saved as a DICOM file — the standard format of medical imaging. The scanner is calibrated every 24 hours using mock-ups of air, water, and aluminium to keep the numbers honest.

Here lies the paradox that motivated the new study, led by Hanaya Okuda of the Japan Agency for Marine-Earth Science and Technology together with colleagues in Austria and France. Although XCT scanning is one of the very first measurements performed on every core, and although it captures the most complete and continuous record of the material before anything is cut away, the data were largely going to waste. The DICOM files are enormous — typically more than a gigabyte per section — the viewing software is unfamiliar to most geologists, and only a handful of dedicated computers on the vessel could even open them. As a result, the scans were consulted almost exclusively by designated watchdog scientists whose job was to flag intervals for time-sensitive whole-round sampling, such as interstitial water squeezing or microbiology work, before the core warmed and its chemistry drifted.

Okuda and her team, including Charlotte Pizer and Michael Strasser of the University of Innsbruck and Mai-Linh Doan of Université Grenoble Alpes, set out to democratize that data. They built a Python-based visualization workflow that runs on the ship’s existing computers and converts the raw DICOM stacks into a family of accessible images that any shipboard scientist can consult within minutes. From each three-dimensional section volume, the workflow generates four image types: slices along the split surfaces of both the archive half and the working half, a slice perpendicular to the split surface that reveals structures hiding beneath it, and an unwrapped image produced by sampling the volume along a circular path of about 22.5 millimetres diameter centred on the core axis — effectively flattening the curved outer surface of the core into a single panoramic strip, a scheme adapted from methodology developed during IODP Expedition 405.

Alongside the images, the workflow computes a depth profile of CT numbers by averaging pixel values within a 40-millimetre-diameter circular area, counting only voxels above 1,200 Hounsfield units to exclude liner material and air. These parameters can be tuned manually for different lithologies, and the profiles are exported both as images and as CSV files for quantitative analysis. The individual products are then assembled into three nested summaries. A section summary lines up all five products for one core section, ideal for pinpointing undisturbed, representative spots for sampling. A core summary aligns one image type across every section of a core, directly mirroring what scientists see on the core description tables. Finally, a hole summary stacks every core in the borehole into a continuous downhole image, generated for both of the programme’s depth scales, CSF-A and CSF-B, and vertically compressed to one-third length for the downhole views.

The practical payoff is immediate and concrete. A physical properties specialist hunting for a crack-free cube of sediment for P-wave velocity measurements can scan the perpendicular slice to dodge fractures that would sabotage the measurement. Sedimentologists describing the archive half can compare their observations against the split-surface images to spot burrows and event deposits that a knife might smear. Structural geologists can trace faults through the working half before sampling destroys the context. And because the summaries are saved on a local computer accessible to everyone, scientists on any shift, in any science team, can generate them independently rather than queuing for the few machines equipped with DICOM viewers.

Getting the software to run smoothly on a research vessel posed its own engineering challenges. The team chose a Mac mini running the open-source Horos DICOM viewer as the processing host, because Horos offers prompt access to the shipboard XCT server and can export data easily, unlike the locked-down Advanced Workstation used for watchdog duties. They also decompress the DICOM files into the standard uncompressed Explicit VR Little Endian format, ensuring the archives remain usable in post-cruise research with tools such as ImageJ. To sidestep the notorious difficulty of configuring Python environments on shipboard machines, the entire workflow is packaged in a Docker container — a self-contained digital ecosystem that can be deployed on any computer running the free Docker software. The code reads each section’s 18-digit J-CORES identification number, builds a look-up table automatically, and processes more than 2,000 DICOM files per section in parallel across all processor cores, finishing in a few minutes; the downhole summaries take about one minute each.

The workflow was implemented and tested during IODP³ Expeditions 502 and 503, and while the summaries cannot yet be produced before time-sensitive whole-round sampling — not all sections have been scanned by that point — they are ready well before non-time-sensitive sampling, section splitting, and visual core description begin. The authors are candid about the method’s limits: slicing a three-dimensional volume along fixed planes inevitably misses some structures, so an irregular, subvertical feature such as an injection structure near the core’s outer edge could escape detection. They therefore recommend complementing the summaries with examination in a full three-dimensional DICOM viewer. Looking ahead, they envision direct comparison of the unwrapped images with borehole logging data, and the integration of machine-learning techniques to automatically detect key features, potentially fusing XCT data with continuous core scan images and other track measurements for an unprecedented real-time picture of what the drill has recovered.

What makes this development resonate beyond one ship is its philosophy. The Chikyu’s scanner was already producing world-class data on every expedition; the bottleneck was not instrumentation but accessibility. By wrapping a robust visualization pipeline in a portable container and publishing both the Docker image and the Python scripts openly, the team has shown how a modest investment in software can multiply the scientific return of expensive, hard-won samples drilled from the seafloor. The same strategy, the authors note, can be transferred to any laboratory working with drill cores, promising rapid and consistent visualization of large tomographic datasets wherever scientists interrogate the Earth’s buried archives.

Subject of Research: Non-destructive X-ray computed tomography visualization of drill cores and its integration into the shipboard core workflow on the drilling vessel Chikyu

Article Title: Non-destructive core visualization using X-ray computed tomography scan and its implementation into the core workflow on D/V Chikyu

Article References: Okuda, H., Pizer, C., Doan, M.-L., & Strasser, M. (2026). Non-destructive core visualization using X-ray computed tomography scan and its implementation into the core workflow on D/V Chikyu. Scientific Drilling, 35(1), 119-125. https://doi.org/10.5194/sd-35-119-2026

Image Credits: AI Generated

DOI: 10.5194/sd-35-119-2026

Keywords: X-ray computed tomography, drill cores, D/V Chikyu, IODP, scientific drilling, DICOM, core visualization, Python workflow, Docker, marine geology, shipboard science, JAMSTEC

The research team, led by Jinrong Cao of the Geological Carbon Dioxide Storage Technology Research Association and the Research Institute of Innovative Technology for the Earth in Japan, combined three observational streams in a single experiment: distributed fiber-optic strain sensing, time-lapse X-ray computed tomography, and coupled numerical modeling of two-phase flow and poroelastic deformation. Their specimen was a cylindrical tuffaceous sandstone core, 34.83 millimeters in diameter and 179.50 millimeters long, drilled from the Sarukawa oil field in Oga, Akita Prefecture, an area selected for a Japanese advanced carbon storage program targeted to begin operations by 2030. The rock had a porosity of 25.06 percent and an intrinsic permeability of 5.41 millidarcys, with a mineral framework dominated by the zeolite clinoptilolite at roughly 63 percent, alongside quartz, albite, anorthite, and heulandite.

The sensing element was a single-mode silica fiber bonded to the core surface with epoxy at an optimized wrap angle of 80 degrees relative to the core axis. This helical geometry converts one continuous fiber into a dense array of quasi-independent strain sensors, resolving circumferential, or hoop, strain at one-centimeter spatial resolution along the full specimen length. The interrogator, a Neubrex 7020 unit, works by measuring Rayleigh frequency shifts in backscattered laser light using a tunable wavelength coherent optical time-domain reflectometer. Because strain and temperature both shift the Rayleigh spectrum, the team held the experiment at a stable 40 degrees Celsius and applied a calibrated strain coefficient of minus 0.153 gigahertz per microstrain, yielding a strain resolution of about half a microstrain for comparable measurements, far finer than the tens to roughly one hundred microstrains observed during injection.

Before introducing carbon dioxide, the researchers ran a single-phase water flow test to pin down permeability and validate their poroelastic model. When upstream pressure reached 10.5 megapascals, circumferential strain near the inlet jumped to approximately 75 microstrains within a single sampling interval of 36 seconds or less, while the downstream end of the core registered only a few microstrains. Numerical simulations showed that the sub-second initial response could not be resolved experimentally, but by roughly half a second to one second the simulated strain gradient matched the first measured profile, and a steady inlet-to-outlet strain gradient established itself within about one second. This close agreement between measurement and model confirmed that, in fully water-saturated rock, deformation is governed by pore water pressure gradients acting through Biot-style poroelastic coupling.

The main event was the supercritical carbon dioxide flood. Time-lapse CT scans, taken at 60 seconds per volume using an iodide-rich contrast solution to sharpen fluid-rock contrast, tracked the drainage process in detail. During the first 1.6 minutes, corresponding to 0.036 injected pore volumes, carbon dioxide accumulated mainly near the inlet. By 2.67 minutes the saturation field had become spatially nonuniform, reflecting variability in pore accessibility and capillary entry conditions across the zeolite-rich framework. At 4.27 minutes the front reached the specimen midpoint, and breakthrough occurred at 8.0 minutes, when average carbon dioxide saturation hit 19.4 percent. Differential pressure during drainage held steady between 0.45 and 0.48 megapascals at a flow rate near one milliliter per minute.

The striking result was how differently the strain field evolved from the saturation field. At the onset of injection, a pronounced strain gradient appeared immediately: strain near the inlet climbed to roughly 80 microstrains, while even the outlet showed a small but detectable response. The strain front was far more spatially diffuse than the comparatively sharp saturation front imaged by CT. By breakthrough, strain had reached about 100 microstrains near the inlet and 20 microstrains near the outlet. Most tellingly, at a position 0.09 meters from the inlet, a distinct strain increase appeared near the very start of injection, while local carbon dioxide saturation remained at zero for the first four minutes. The delay between strain onset and saturation onset grew with distance from the inlet, exactly the pattern expected if pressure disturbances propagate through the connected aqueous phase faster than the capillary-controlled gas front advances.

The coupled numerical model, a two-dimensional axisymmetric finite-element representation of the full specimen, provided the physics-based bridge between the observable strain and the unmeasured internal pressure. The team calibrated two-phase flow parameters exclusively against the CT-derived saturation profiles, then compared the predicted strain against the fiber measurements as an independent test. Simulated pore gas and pore water pressures evolved concurrently but with different spatial patterns, related through capillary pressure, and the modeled strain reproduced both the magnitude and the timing of the measured response. Because the sandstone’s apparent drained bulk modulus of roughly one to two gigapascals makes it relatively compliant, a given pressure perturbation produces a large, measurable strain. The authors are careful to stress that the fiber does not measure pressure directly; the strain signal is a model-constrained poroelastic proxy for the evolving pressure footprint, not a standalone pressure gauge.

The engineering implications reach well beyond the laboratory. Operational surveillance of storage sites typically relies on wellhead and bottomhole pressure, flow rates, sparse downhole gauges, and episodic plume imaging, none of which uniquely constrains the internal pressure field. Distributed strain sensing adds a spatially continuous constraint that can be folded into coupled flow and geomechanical models, helping to characterize uncertain formation properties and to evaluate deformation relevant to reservoir, caprock, and wellbore integrity. Field deployments already hint at feasibility: distributed fiber sensing has been demonstrated under field-scale injection conditions at the CO2CRC Otway site in Australia, a pumping test at Mobara in Japan captured pressure-related reservoir deformation, and behind-casing water-injection measurements show that injection-induced strain is detectable through a casing-cement-formation system.

The authors are equally candid about the limitations that separate their benchtop result from routine field practice. The adhesive layer bonding the fiber to the core was not independently calibrated, so perfect strain transfer was assumed. Field fibers are usually installed axially behind casing, where recorded amplitude depends on cable orientation, completion design, cement properties, gauge length, and strain-transfer efficiency, and axial strain is not generally equivalent to the hoop strain measured here. Stiffer formations or rocks with a lower Biot-Willis coefficient would produce smaller strains that may fall below detection thresholds, meaning an absent strain signal cannot be read as absent pressure communication. The homogeneous axisymmetric model also cannot capture localized effects of bedding, fractures, or anisotropic permeability, and longer-term chemical interactions between carbon dioxide, brine, and reactive minerals may gradually alter rock stiffness and fracture behavior.

Even with those caveats, the study delivers a compelling proof of concept: the mechanical consequences of injection arrive before the plume does, and a helically wound strand of glass can watch them happen. Future priorities include experiments on rocks with contrasting stiffness and capillary behavior, direct internal pressure measurements, three-dimensional models that explicitly represent the helical fiber geometry, and field studies that integrate distributed strain with pressure gauges, time-lapse geophysics, distributed acoustic and temperature sensing, and surface deformation observations. As global net-zero pathways call for roughly ten gigatons of carbon dioxide storage per year by 2050, tools that convert invisible pressure transients into measurable, interpretable strain signals could become an essential part of the monitoring toolkit, giving operators and regulators earlier and more defensible warnings about where the subsurface is being mechanically stressed.

Subject of Research: Distributed fiber-optic sensing of poroelastic strain precursors during supercritical CO2 core flooding in sandstone

Article Title: Poroelastic strain precursors during laboratory carbon dioxide core flooding revealed by helically bonded fiber sensing

Article References: Cao, J., Park, H., Xue, Z., Nakajima, T., Amer, R., & Ouchi, W. (2026). Poroelastic strain precursors during laboratory carbon dioxide core flooding revealed by helically bonded fiber sensing. Communications Engineering, 5(1), Article 164. https://doi.org/10.1038/s44172-026-00783-5

Image Credits: AI Generated

DOI: 10.1038/s44172-026-00783-5

Keywords: carbon capture and storage, distributed fiber-optic sensing, poroelasticity, CO2 core flooding, X-ray computed tomography, pore pressure, geomechanics, sandstone, capillary pressure, geological storage, strain sensing, two-phase flow

The technology at the heart of the study is called NIR-II fluorescence molecular tomography integrated with X-ray computed tomography, or NIR-II FMT-XCT. Fluorescence molecular tomography is an optical imaging technique in which a fluorescent probe is delivered into the body and light emitted from the probe is collected at the tissue surface. Mathematical reconstruction algorithms then solve an inverse problem to localize and quantify the fluorescence source deep inside the tissue. The key innovation here is the use of the second near-infrared window, spanning roughly 1,000 to 1,700 nanometers, where biological tissues scatter and absorb far less light than in the conventional first near-infrared window. Lower scattering means photons can travel farther and more predictably through tissue, which translates into better penetration depth and higher contrast. In head-to-head comparisons reported in the paper, three-dimensional NIR-II FMT achieved a signal-to-background ratio 1.73 plus or minus 0.27 times higher than three-dimensional NIR-I FMT, a substantial gain in imaging contrast that matters enormously when the target is a faintly expressed protein buried inside dense tumor tissue.

To make PD-L1 visible, the team engineered a targeted probe called aPD-L1-ICG, which links an anti-PD-L1 antibody to indocyanine green, a clinically approved fluorescent dye already used in humans for angiography and other applications. Once injected, the antibody portion seeks out and binds PD-L1 molecules on tumor cells and immune cells, carrying the fluorescent tag to sites of expression. The XCT component of the hybrid system supplies anatomical context: it images the structural skeleton of the chest and nodes, allowing the optical signal to be mapped onto defined three-dimensional regions of interest. The reconstruction pipeline incorporated a standardized imaging space-based elastic net method with an alternating direction method of multipliers solver, designed to stabilize the notoriously ill-posed fluorescence reconstruction problem and produce quantitative, reproducible measurements rather than vague blobs of brightness.

The researchers validated the platform across a demanding spectrum of models and specimens. They began with lung cancer cell lines, then moved to patient-derived xenografts, in which fragments of human lung tumors are grown in immunocompromised mice, preserving much of the heterogeneity of the original cancer. They then applied the technique to resected human lung tumors and lymph nodes obtained from patients. Throughout this progression, the reconstructed three-dimensional fluorescence signals tracked with PD-L1 immunohistochemistry, the current clinical gold standard, but with an important added dimension: the imaging revealed spatial variation of PD-L1 within tumors and within lymph nodes, information that a slide-based method simply cannot capture because it samples only a thin plane of the specimen.

The most clinically consequential part of the study involved 28 patients with lung cancer who had received neoadjuvant chemoimmunotherapy, meaning chemotherapy plus immune checkpoint blockade given before surgery to shrink the tumor and prime the immune system. After treatment, when surgical specimens were resected, the team performed ex vivo NIR-II FMT measurements on the tumors and compared the resulting signal-to-background ratios with two benchmarks: the radiographic response seen on CT scans before surgery, and the pathological response measured in the resected tissue as the proportion of residual viable tumor cells. The results were striking. Tumor NIR-II FMT signal-to-background ratio correlated with CT-based tumor shrinkage with a correlation coefficient of 0.6076 and a P value below 0.001, indicating a moderately strong and highly statistically significant association between the molecular signal and the anatomical response to therapy.

Even more provocative was the comparison with conventional PD-L1 immunohistochemistry. The ex vivo NIR-II FMT signal-to-background ratio was inversely associated with the residual viable tumor proportion, with a Pearson correlation coefficient of minus 0.5332 and a P value of 0.0035, meaning patients whose tumors showed stronger PD-L1-targeted fluorescence after treatment tended to have less surviving tumor tissue. By contrast, tumor-cell PD-L1 expression measured by standard immunohistochemistry showed no significant association with residual viable tumor, with a coefficient of minus 0.1119 and a P value of 0.571. In other words, the three-dimensional whole-tissue fluorescence measurement captured information about therapeutic response that the conventional two-dimensional biomarker assessment missed, presumably because it integrates expression across the entire specimen rather than averaging over selected microscopic fields.

The biological rationale for this advantage is worth unpacking. PD-L1 is not a static property of tumor cells alone; it is dynamically induced by inflammatory signaling, particularly interferon gamma released by activated T cells. When checkpoint blockade works, T cells flood into the tumor, and the resulting immune activity can actually drive PD-L1 expression upward on surviving tumor and immune cells. A high post-treatment PD-L1 signal measured across the whole tumor volume may therefore reflect an engaged, drug-perturbed immune microenvironment rather than simply more target protein, and the correlation with tumor shrinkage and low residual disease is consistent with that interpretation. Immunohistochemistry, by contrast, examines a limited tissue plane and reports tumor proportion scores that may be discordant with deeper regions of the same tumor, especially after treatment has reshaped the tissue architecture.

The authors are careful, appropriately so, about what the study does and does not prove. All patient findings are retrospective and exploratory, based on post-treatment specimens measured after surgical resection. The associations support further prospective investigation but do not establish pretreatment predictive utility, meaning the technique has not yet been shown to identify which untreated patients will respond, nor do they establish clinical net benefit, meaning no evidence yet exists that using this imaging to guide decisions improves outcomes. Building that evidence will require prospective studies in which the fluorescence signal is measured before or early during therapy and used, ideally in a controlled trial design, to inform or stratify treatment. The regulatory and logistical path is also nontrivial: although indocyanine green is clinically established, the antibody-dye conjugate would need to be produced under good manufacturing practice conditions, validated for safety, biodistribution and specificity, and cleared through the appropriate regulatory frameworks before any routine clinical use.

Nevertheless, the technical achievement is considerable. By combining the deep-tissue advantages of second near-infrared fluorescence with the anatomical precision of X-ray CT and a robust reconstruction algorithm, the team has demonstrated a platform that turns a qualitative biomarker slide into a quantitative, three-dimensional molecular map. The intraclass correlation analyses and comparisons across preclinical models, xenografts and human resections described in the study suggest the measurements are reproducible enough to support larger trials. For a field in which roughly half of PD-L1-positive patients do not respond to checkpoint inhibitors and some PD-L1-negative patients respond unexpectedly, any tool that reads the tumor more completely has obvious appeal. If prospective validation confirms the promise of this exploratory work, surgeons and oncologists may one day assess not a single biopsy of a tumor but the whole tumor and its draining lymph nodes in three dimensions, bringing immunotherapy selection closer to the true, messy biology of the disease it is meant to conquer.

Subject of Research: Three-dimensional NIR-II fluorescence molecular tomography with X-ray CT for quantitative PD-L1 imaging in lung cancer and tumor-draining lymph nodes to assess immunotherapy efficacy

Article Title: 3D NIR-II FMT-XCT imaging for quantitative analysis of PD-L1 expression in lung cancer and tumor-draining lymph nodes facilitating immunotherapy efficacy assessment: an exploratory study

Article References: 3D NIR-II FMT-XCT imaging for quantitative analysis of PD-L1 expression in lung cancer and tumor-draining lymph nodes facilitating immunotherapy efficacy assessment: an exploratory study. (n.d.). https://doi.org/10.1186/s12916-026-05200-4

Image Credits: AI Generated

DOI: 10.1186/s12916-026-05200-4

Keywords: PD-L1, lung cancer, immunotherapy, fluorescence molecular tomography, NIR-II imaging, X-ray computed tomography, molecular imaging, immune checkpoint inhibitors, tumor-draining lymph nodes, neoadjuvant chemoimmunotherapy, biomarkers, precision medicine

The study, conducted at the Wiregrass Research and Extension Center in Headland, Alabama, was set up on a Dothan sandy loam in the Coastal Plain region, where more than 85 percent of the soil is sand and organic matter levels are low. The team, led by Ranveer Singh and Jasmeet Lamba of Auburn’s Department of Biosystems Engineering, together with crop scientist Charles Y. Chen and Sushil Adhikari, spread biochar produced by pyrolysis at roughly 980 degrees Celsius at rates of 0, 7.5, 15 and 30 megagrams per hectare, equivalent to 0, 1.1, 2.2 and 4.4 percent by weight of the top soil layer. The biochar was incorporated into the top 50 millimeters of soil after peanut emergence in May 2024, and undisturbed soil cores were collected just before harvest in October 2024.

The centerpiece of the investigation was the X-ray CT scanning, performed with a GE EVO Revolution 128-slice medical scanner at Auburn’s College of Veterinary Medicine. Operating at 140 kilovolts and 210 milliamperes, the scanner produced three-dimensional image stacks of the top 100 millimeters of soil at a resolution of about 0.195 millimeters. The raw grayscale data were then processed in ImageJ, where a local thresholding algorithm was carefully calibrated using artificial pores of known diameter embedded in the soil columns, ensuring that the segmentation of soil matrix and pore space was accurate to within 1.5 percent. From these images the researchers extracted macroporosity, macropore number density and pore circularity, the geometric fingerprints of the soil’s plumbing system.

The results were striking in their subtlety. In the top 50 millimeters, where the biochar had been incorporated, macroporosity showed no significant change at any application rate. But in the layer just below, between 50 and 100 millimeters, the macropore number density dropped dramatically: by 57 percent at the 15-megagram rate and 64 percent at the 30-megagram rate compared with untreated control plots. Pore circularity, a measure of how close pores are to a perfect circle, fell by about 5 percent at the moderate 15-megagram rate in the surface layer, indicating that biochar particles had disrupted the geometry of existing pores, merging them into elongated, irregular cavities rather than leaving them untouched.

The researchers interpret these patterns through two competing mechanisms they call the occupying effect and the expansion effect. Under the occupying effect, biochar particles physically fill existing soil pores, reducing the volume of large, freely draining channels. Under the expansion effect, the biochar’s own inherited internal porosity adds new pore space to the bulk soil. The data suggest that at rates up to about 15 megagrams per hectare, the occupying effect dominates in this sandy soil, while at 30 megagrams per hectare the pore indices began converging back toward control values, hinting that beyond that threshold the expansion effect may take over. The response, the authors argue, is likely a function of the soil-to-biochar ratio.

Perhaps the most intriguing finding concerns where the biochar actually went. Differences in pore indices in the 50-to-100-millimeter layer, below the incorporation depth, point to the preferential downward movement of biochar particles through the coarse pores of the sandy soil. This migration may explain why the amendment failed to deliver the expected benefits: the particles simply did not stay put. Previous studies have documented vertical transport of biochar in coarse-textured soils, and the Auburn team speculates that if biochar particles remained immobile at their depth of incorporation, improvements in bulk density and total porosity might have materialized, as they have in controlled pot experiments where such losses are negligible.

The hydro-physical measurements told a similarly sobering story. Soil water retention curves, generated by combining high-resolution evaporation-method data from HYPROP2 tensiometers in the wet range with WP4C dewpoint measurements extending to oven dryness, showed no significant effect of biochar at any rate or depth. Bulk density, total porosity, wet aggregate stability and plant-available water were all statistically unchanged four months after application. Even the highest dose of 30 megagrams per hectare, a substantial and costly amount, left the soil’s capacity to hold water against gravity and to supply it to plants essentially untouched.

Why did biochar fail here when it has succeeded elsewhere? The authors point to the soil itself. Clay particles, with their high surface area and reactive surfaces, provide the binding sites where organic matter and biochar can knit stable aggregates; sandy soils with low clay and low organic matter offer few such opportunities. Pyrolysis temperature matters too: biochar produced at high temperatures, as in this study, can be hydrophobic, repelling water rather than absorbing it, at least until environmental exposure gradually wears that hydrophobicity down. Other researchers have found that water retention in sandy soils improves only at much higher application rates, in the range of 40 to 65 megagrams per hectare, doses that would be economically prohibitive for most farms.

The implications reach beyond agronomy into environmental policy. Biochar’s ability to reduce nutrient leaching and greenhouse gas emissions depends in part on how it reshapes the soil’s pore network. If biochar occupies large pores and shifts flow from rapid preferential channels to slower matrix flow, solutes spend more time in contact with biochar surfaces and can be adsorbed. But if the expansion effect creates abundant new preferential pathways, contaminants could bypass the amendment entirely. Knowing which regime prevails, at which dose, in which soil, is therefore essential for anyone counting on biochar to deliver its promised environmental services.

The Auburn team is candid that their hypothesis, that biochar would improve the hydro-physical properties of a poorly structured sandy soil within a single growing season, was rejected by the data. They argue that the field needs more refined prescriptions: biochar types and rates matched to specific soil textures, targeted application methods such as banding near the root zone or subsurface placement to reduce runoff losses, and longer observation windows to capture slow structural changes. They also call for micro-CT imaging at micrometer resolution to resolve the pores inside biochar particles themselves, which current medical scanners cannot detect. For now, the message for farmers on sandy ground is one of tempered expectations: biochar remains a promising tool for carbon sequestration, but transforming it into a reliable water-management strategy will require far more precision than simply spreading charcoal and hoping for the best.

Subject of Research: Short-term effects of pine biochar application rates on soil pore networks and hydro-physical properties in sandy soil

Article Title: Elucidating short-term impact of biochar addition on soil pore networks using X-ray computed tomography and soil water retention characteristics

Article References: Singh, R., Lamba, J., Chen, C. Y., & Adhikari, S. (2026). Elucidating short-term impact of biochar addition on soil pore networks using X-ray computed tomography and soil water retention characteristics. Discover Soil, 3(1), Article 102. https://doi.org/10.1007/s44378-026-00252-7

Image Credits: AI Generated

DOI: 10.1007/s44378-026-00252-7

Keywords: biochar, soil pore networks, X-ray computed tomography, macroporosity, soil water retention, sandy soil, pyrolysis, soil physics, peanut production, soil amendment, hydro-physical properties, pore circularity

The research team, led by Huan Xu and Juanjuan Ren with colleagues including Yanquan Wu, Wengao Liu, Jun Ye, and Shijie Deng, approached the problem through a staged, progressive optimization strategy. They prepared thirteen concrete mixtures based on the mix design method specified in the Chinese railway standard TB/T 3275, using ordinary Portland cement, Class F fly ash, ground granulated blast-furnace slag, manufactured sand, and natural crushed stone. In the first stage, polyvinyl alcohol fiber content varied from 0.05 to 0.5 percent by volume; in the second, basalt fiber was added at 0.1 to 0.4 percent on top of the best PVA dosage; in the third, nano-silica was introduced at 1 to 4 percent by mass of cementitious material. This incremental design allowed the researchers to trace the incremental contribution of each component, moving from single-fiber to hybrid-fiber to ternary fiber-and-nanoparticle systems.

Each fiber brings distinct strengths and trade-offs. Polyvinyl alcohol fibers are lightweight, non-conductive, and economical, with an elastic modulus of 40 gigapascals and a tensile strength of 1,830 megapascals, making them compatible with the track circuits that steel fibers would interfere with. Basalt fibers are stiffer still, with an elastic modulus of 87.2 gigapascals and tensile strength of 2,180 megapascals, offering superior load transfer across cracks, though excessive single-fiber use can increase brittleness. Nano-silica, with a specific surface area of 300 square meters per gram and an average particle size near 20 nanometers, acts at an entirely different scale: it physically fills capillary pores, accelerates cement hydration through its pozzolanic activity, and reacts with calcium hydroxide to generate additional calcium-silicate-hydrate gel, densifying the matrix and improving fiber-matrix interfacial bonding.

The first challenge was workability, and here the trade-offs became apparent. All thirteen mixtures satisfied the standard’s slump requirement of no more than 200 millimeters, but fiber and nanoparticle additions steeply reduced fluidity. Increasing PVA fiber content from 0.05 to 0.5 percent cut slump by between 16.1 and 58.7 percent relative to the reference mix, largely because the hydrophilic hydroxyl groups on PVA absorb free water and the flexible fibers intertwine during mixing. Basalt fiber partially mitigated the loss, with one hybrid mixture showing a slump 34.4 percent higher than its PVA-only counterpart, likely because basalt’s lower hygroscopicity reduces PVA agglomeration. Nano-silica caused the steepest decline, with the 3 percent nano-silica mixture reaching only 62 millimeters of slump, a consequence of its enormous surface area adsorbing free water and superplasticizer molecules.

Mechanical testing after 28 days of standard curing revealed a consistent pattern: properties rose and then fell with increasing dosage, reflecting an optimum beyond which fiber agglomeration creates weak zones. The best ternary mixture, designated P2B1N3 with 0.2 percent PVA, 0.1 percent basalt fiber, and 3 percent nano-silica, achieved a cube compressive strength of 60.5 megapascals, a splitting tensile strength of 5.2 megapascals, and a flexural tensile strength of 7.0 megapascals, representing gains of 33.3, 30.7, and 52.2 percent respectively over the unmodified reference concrete. The flexural improvement was particularly striking, nearly 6.5 times the gain achieved by PVA fiber alone, underscoring how nanoparticle-driven matrix densification multiplies the effectiveness of hybrid-fiber bridging.

The heart of the study lay in fracture mechanics. Notched three-point bending beams were tested under closed-loop displacement control on an MTS 810 system, while a 2D digital image correlation system tracked crack initiation and propagation on the specimen surface. The load versus crack mouth opening displacement curves displayed three classic stages: linear elastic deformation, stable crack propagation, and unstable failure. The crack initiation load of the ternary mixture reached 5.11 kilonewtons, 40.4 percent above the reference, and its peak load reached 7.8 kilonewtons, a 22.3 percent increase. Applying the double-K fracture criterion, the researchers found that crack initiation fracture toughness rose by 36.7 percent and unstable fracture toughness by a remarkable 75 percent compared with the reference concrete, while fracture energy determined by the RILEM work-of-fracture method climbed 77.6 percent and the ductility index improved 45.4 percent.

Digital image correlation revealed how fundamentally the modified mixtures changed crack behavior. The unmodified reference specimen showed textbook brittle fracture: its crack length leapt from 3.29 millimeters before peak load to 43.48 millimeters at peak, and the specimen failed with a through crack nearly perpendicular to the loading point. In contrast, the hybrid-fiber mixtures displayed distributed strain fields, delayed strain localization, and crack tip opening displacements that grew far more gradually. At the peak load stage, the ternary mixture’s crack length was 28.93 millimeters, and horizontal strains near the notch tip spread over a broad region rather than concentrating along a single line, evidence that fibers and nanoparticles were sharing the burden of resisting deformation across multiple scales.

X-ray computed tomography provided the mesoscopic explanation. Scanning cylindrical core samples at a voxel size of approximately 45 micrometers, the team found that PVA fiber alone actually increased volumetric porosity, from 1.49 percent in the reference to 2.04 percent, because fiber incorporation entrains air and weakens local matrix-aggregate contact. Adding basalt fiber brought porosity back down to 1.45 percent, and the ternary mixture achieved the lowest porosity of all, 1.40 percent, along with the smallest coarse-pore fraction: only 3.21 percent of pores exceeded 500 micrometers and 0.91 percent exceeded 1,000 micrometers, compared with 5.00 and 1.01 percent in the reference. Layer-by-layer analysis showed the ternary specimen had the most uniform pore distribution with no high-porosity weak layers, and a normalized multi-indicator comparison confirmed an inverse association between coarse-pore content and mechanical-fracture performance.

The authors are candid about the limits of their work. The study examined 28-day laboratory-scale specimens only, used one CT specimen per representative mixture, and did not include complete single- and binary-component control groups, so the independent and interaction effects of the three components could not be statistically separated. Mechanisms such as interfacial transition zone refinement and improved fiber-matrix bonding are inferred from prior literature rather than directly characterized. Future work will need to address early-age shrinkage and thermal cracking, freeze-thaw resistance, sulfate attack, coupled environmental-loading conditions, cost-effectiveness, and ultimately full-scale track slab tests. Nevertheless, the consistency across macroscopic fracture parameters, DIC-based crack evolution, and CT-derived pore structure provides a compelling, cross-scale case that the PF-BF-NS ternary system, with stage-dependent and functionally complementary roles, offers a robust materials-level foundation for optimizing the durability and safety of the concrete slabs that carry high-speed trains across China and beyond.

Subject of Research: Fracture performance and toughening of nano-silica-modified hybrid fiber reinforced track slab concrete for high-speed railway ballastless tracks

Article Title: Fracture performance of nano-modified hybrid fiber reinforced concrete for strengthening and toughening of high-speed railway track slab

Article References: Xu, H., Ren, J., Wu, Y., Liu, W., Ye, J., & Deng, S. (2026). Fracture performance of nano-modified hybrid fiber reinforced concrete for strengthening and toughening of high-speed railway track slab. Case Studies in Construction Materials, 25, Article e06524. https://doi.org/10.1016/j.cscm.2026.e06524

Image Credits: AI Generated

DOI: 10.1016/j.cscm.2026.e06524

Keywords: track slab concrete, hybrid fiber reinforcement, polyvinyl alcohol fiber, basalt fiber, nano-silica, fracture toughness, double-K fracture criterion, digital image correlation, X-ray computed tomography, fracture energy, crack propagation, high-speed railway