The research team was led by Associate Professor Feng Peng of Chongqing University, with colleagues from the 26th Institute of China Electronics Technology Group Corporation and Stevens Institute of Technology. The peer-reviewed study, “Scalable Photonic Crystal Scintillators for High-Performance X-ray Imaging in Photon-Starved Regimes,” was published online in Opto-Electronic Advances on September 16, 2026.

X-ray scintillators absorb high-energy X-ray photons and convert them into visible light that can be recorded by a camera or photodetector and used to reconstruct an image. High-density inorganic scintillators such as cerium-doped gadolinium aluminum gallium garnet, or GAGG:Ce, offer strong X-ray stopping power and high luminescence efficiency. However, a mismatch between the refractive index of the scintillator and that of the surrounding air or packaging medium can trap generated light through total internal reflection. The source reports that as many as 90% of photons may remain confined inside the material in waveguide modes, reducing the light available to the detector.

This limitation is especially relevant to thin scintillators used for high-resolution or low-dose imaging. Increasing scintillator thickness can improve X-ray detection sensitivity, but it can also increase lateral diffusion of scintillation light and reduce spatial resolution. The researchers therefore focused on the light-extraction interface rather than the bulk of the scintillator, testing whether a periodic structure could release optical modes otherwise confined inside the material.

To create the structure, the team used an interfacial self-assembly process driven by a surface-tension gradient in an ethanol-water binary solvent system. The process combined surfactant-assisted interfacial densification with vibration-free drainage transfer. The result was a macroscopically continuous monolayer of hexagonal close-packed polystyrene nanospheres on a 25 mm-diameter GAGG:Ce scintillator wafer. The periodic photonic crystal supplies additional in-plane momentum, allowing modes trapped within the scintillator to couple into free-space radiation modes through momentum matching and Bragg diffraction.

The fabrication approach was intended to address limitations in two established methods. Top-down lithography is precise but requires costly equipment and complex processing, and it is difficult to apply to large-area brittle scintillator substrates. Bottom-up self-assembly can be less expensive, but large-area films formed by that approach may contain cracks, vacancies, and multi-crystal domains that reduce brightness gains or introduce imaging artifacts. The reported strategy was proposed and experimentally validated for producing large-area photonic crystal scintillators through interfacial self-assembly.

In experiments, the modified scintillator produced an approximately 4.64-fold increase in average light output and a 13 dB improvement in X-ray imaging signal-to-noise ratio. It also maintained spatial resolution better than 25 μm. These measurements indicate that the interface treatment improved light extraction without sacrificing spatial resolution in the tested device; they do not establish clinical performance, patient benefit, or the minimum radiation dose required for a particular diagnostic application.

The source identifies photon-starved X-ray imaging as the intended operating context, including possible applications in low-dose medical diagnosis, rapid security inspection, and high-precision industrial non-destructive testing. These are proposed application areas rather than demonstrated clinical or field deployments in the reported experiment. The technology was also presented as a potential route toward lower-dose and higher-definition X-ray imaging under weak-signal conditions.

Several limitations remain. Residual defects in the self-assembled structure may still have a minor effect on resolution, and the authors state that the process requires further optimization to remove those defects. They also call for investigation of compatibility with a broader range of scintillator materials. The study was supported by the Natural Science Foundation of Chongqing, the Development Foundation of the 26th Institute of China Electronics Technology Group Corporation, the Beijing Natural Science Foundation, and China’s National Key Research and Development Program.

Subject of Research: Experimental photonic crystal scintillators for high-performance X-ray imaging under photon-starved conditions

Article Title: New photonic crystal scintillator paradigm boosts low-cost X-ray imaging sensitivity

Article References: New photonic crystal scintillator paradigm boosts low-cost X-ray imaging sensitivity. Original publication (Bibliographic details unavailable.)

Image Credits: AI Generated

DOI: Not provided

Keywords: photonic crystals, X-ray imaging, scintillators, light extraction, GAGG:Ce, self-assembly, signal-to-noise ratio, spatial resolution, low-dose imaging, photonic, crystal, scintillator

The research, led by Songyuan Tang of the APS in collaboration with Tekin Bicer, Kamel Fezzaa, and Samuel Clark of Argonne National Laboratory, tackles the problem by fusing two x-ray image streams of the same physical process. One stream comes from a high-speed (HS) camera with excellent spatial resolution but a modest frame rate. The other comes from an ultra-high-speed (UHS) camera that films at megahertz rates but with far fewer pixels. The team’s deep learning model, called SWIN-XVR, takes a handful of consecutive low-resolution (LR) frames from the UHS camera and two high-resolution (HR) frames from the HS camera, and reconstructs a new image sequence in which every frame has the pixel size of the HR images and the timing of the LR images. In effect, the model consolidates the capabilities of both cameras into a single virtual detector.

The architecture builds on the Shifted Window (SWIN) transformer, a vision transformer design that distributes self-attention across spatially local, shifting patch groups rather than relating every pixel to every other pixel. Standard vision transformers scale quadratically with image size in two dimensions, which makes them impractical for dense vision tasks such as video restoration. SWIN transformers sidestep this bottleneck by computing attention within local spatio-temporal windows and alternately shifting those windows between transformer blocks to expand the effective receptive field. SWIN-XVR consists of three modules: an input feature extraction module that converts multi-resolution inputs into shallow features through convolution branches, an attention-based deep feature enhancement module that stacks SWIN transformer blocks in multiple stages, and an output reconstruction module that selects the target feature map and upscales spatial dimensions by a factor of four using pixel shuffle layers.

Inside each SWIN transformer block, a multi-head self-attention (MSA) layer jointly attends to multiple feature subspaces and aggregates features within each local spatial window and across all feature maps. A two-layer feed-forward network cascades with the MSA for fine adjustment, layer normalization regularizes features throughout the deep network, and a learnable relative position bias is added to the self-attention maps to better model spatial relationships among features. Stochastic depth randomly drops samples within each training batch for regularization, and residual connections between subsequent transformer stages enhance feature propagation. The convolution outputs are used directly as tokens with a tokenization patch size of just one by one, allowing the model to work at fine granularity.

Training data came from 1,358 videos recorded with a Photron FastCam SA-Z camera operating at frame rates of 1 to 140 kilohertz and a Shimadzu HPV-X2 camera operating at 1 to 5 megahertz, both deployed during high-speed synchrotron x-ray imaging experiments at the 32-ID beamline of the APS. The videos cover a wide range of advanced manufacturing processes. The Photron videos contain 101 to 8,901 frames, while the Shimadzu videos contain 128 frames. Low-resolution images were synthesized by binning high-resolution images with a four-by-four window, and random crops of 96 by 96 and 24 by 24 pixels were made on corresponding HR and LR images to cover spatially consistent regions. The model was trained to minimize the L1 loss with the Adam optimizer and an initial learning rate of 0.0002, updated by a cosine annealing scheduler with warmup.

The decisive step was moving model training from beamline-local computing resources to the Polaris supercomputer at the Argonne Leadership Computing Facility (ALCF). The team implemented data parallelization with PyTorch’s distributed data parallelization (DDP) library, initiating multiple processes across four Nvidia A100 SXM4 GPUs on a single node and across multiple nodes, with the Nvidia collective communications library (NCCL) and message passing interface (MPI) synchronizing GPU devices across the cluster. Using 40 Nvidia A100 GPUs, the end-to-end training time was sped up by roughly a factor of ten compared to training with beamline-local resources. The team monitored GPU utilization using the HTA software package and found that utilization exceeded 90 percent with more than five samples in a batch, so they fixed the batch size at ten for the scalability analysis.

Strong scaling experiments, in which a fixed training job of 1,206,240 images was distributed across increasing numbers of GPUs from four to 256, revealed that scaling efficiency remained above 90 percent for up to 32 GPUs. Beyond that threshold, communication overhead among GPUs began to erode the gains. Kernel time analysis showed that the percentage of computation kernel time generally decreased with the number of GPUs, with the opposite trend in communication kernel time, confirming that distributed training faces a lower bound in wall time as GPU counts continue to rise.

Hyperparameter tuning on Polaris identified the optimal configuration: three input LR frames, a spatial attention window of four by four, a hidden token size of 384, and six attention heads per transformer block. Doubling the hidden size significantly improved reconstruction fidelity, by up to 0.8 decibels in peak signal-to-noise ratio (PSNR) on one test case, at the proportional cost in inference time. Doubling model depth yielded only a slight improvement of up to 0.12 decibels while nearly doubling inference time, so the team doubled the hidden size and used six attention heads for the final architecture. With three subsequent LR frames and two HR frames differing in spatial and temporal resolution by factors of four and twenty respectively, the algorithm achieved average PSNRs of 37.40 and 35.60 decibels on the two testing datasets.

The validated model was tested on two independent x-ray videos from real applications, one of additive manufacturing and one of friction stir welding, both previously uploaded to the tomobank repository. Testing conditions varied the PSNR of the input LR images from roughly 20 to 60 decibels by simulating Poisson noise, and varied the temporal down-sampling of the HR sequence by factors of 2, 10, 20, and 450. Compared against four independent algorithms, namely bicubic interpolation, Bayesian fusion, EDVR, and EDVR-STF, SWIN-XVR performed close to or significantly better than EDVR-STF on the two cases. Bayesian fusion demonstrated higher accuracy only when input LR images were nearly noise-free and in the absence of scene motion, but its performance dropped significantly in the presence of complex motion dynamics. The deep learning-based algorithms consistently outperformed bicubic interpolation.

The practical implications are substantial. Currently, reconstructing 500 high-resolution images of dimension 400 by 1,024 takes about 115 seconds on four Nvidia A100 GPUs. Because the UHS and HS cameras operate in parallel, users avoid frequent camera switching, which could take a prohibitive amount of time during experiments. With total average experiment durations ranging from about ten minutes to an hour, the proposed image acquisition and reconstruction framework could double the data throughput within the same time budget. The algorithm is implemented in Python and publicly available on GitHub. Looking ahead, the team envisions imaging data accumulated at the beamline being transferred to HPC systems to train foundation models at scale, which could then be deployed back to local application servers and fine-tuned for specific tasks, supporting user services with improved accuracy, automation, and robustness. The pipeline is intended to be reproducible during routine operations of ultra-high-speed imaging beamlines such as 32-ID at the APS, and techniques such as domain adaptation could generalize the method to other detectors and modalities, including computerized tomography and magnetic resonance imaging.

Subject of Research: A SWIN-based vision transformer for high-fidelity and high-speed imaging experiments at light sources

Article Title: A SWIN-based vision transformer for high-fidelity and high-speed imaging experiments at light sources

Article References: Tang, S., Bicer, T., Fezzaa, K., & Clark, S. (2025). A SWIN-based vision transformer for high-fidelity and high-speed imaging experiments at light sources. Frontiers in High Performance Computing, 3, Article 1537080. https://doi.org/10.3389/fhpcp.2025.1537080

Image Credits: AI Generated

DOI: 10.3389/fhpcp.2025.1537080

Keywords: SWIN-based, vision, transformer, high-fidelity, high-speed, imaging, experiments, light, sources, scientific research

CSN5 occupies a peculiar and powerful position within the Italian research landscape. Unlike commissions devoted purely to fundamental physics, it is chartered with applied, technological, and interdisciplinary research, with advanced sensing and detection as its central pillar. This mandate places it at the junction where the brutal requirements of experimental particle physics meet the practical needs of medicine, environmental monitoring, security, and space science. The logic is straightforward: detectors built to survive the radiation fields of a hadron collider are, almost by definition, among the most robust and precise sensing devices humans know how to make. Technologies conceived for fundamental experiments routinely find second lives in hospitals, satellites, and industrial settings.

Among the most striking contributions to the collection is work on oriented-crystal electromagnetic calorimetry. In conventional calorimeters, incoming high-energy particles generate showers of secondary particles as they cascade through dense material. When particles are aligned with the crystal axes of certain materials, a phenomenon known as channeling dramatically accelerates shower development, allowing compact crystals to do the work of much bulkier detectors. The INFN team pursued this concept through an extensive research and development program, progressing from single-crystal studies to the construction of a full matrix of lead tungstate, or PWO, crystals. The result points toward compact, high-performance calorimeters particularly suited to collider experiments, beam-dump facilities, and high-energy gamma-ray astrophysics, where every centimeter of detector volume comes at a premium.

Timing and radiation hardness emerge as recurring themes throughout the collection. A comprehensive review examines the development of Low-Gain Avalanche Diodes, or LGADs, and thin silicon diodes for particle therapy. These solid-state detectors offer dramatically superior sensitivity and timing capabilities compared to the gas ionization chambers that have long served as the workhorses of clinical beam monitoring. The implications extend from beam commissioning to dosimetry in ultra-high dose rate radiotherapy, the so-called FLASH regime, and spatially fractionated techniques where dose is delivered in patterns designed to spare healthy tissue. The work illustrates a tight synergy between detector physics, electronics development, and clinical requirements, a collaboration that INFN has cultivated deliberately.

Pushing the LGAD concept further, the CSN5 eXFlu project has produced ultra-thin sensors with an active thickness of just 15 micrometers, roughly a fifth the width of a human hair. At this scale, charge carriers traverse the device so quickly that exceptional timing resolutions become possible, while the thin active volume inherently resists radiation damage that would degrade thicker sensors. The demonstrated performance positions these devices as leading candidates for future high-precision timing applications, both in fundamental physics experiments where particle trajectories must be resolved with picosecond precision and in applied environments characterized by extreme radiation levels.

Medical applications receive further attention through a review of silicon carbide detectors for advanced dosimetry. Silicon carbide’s wide bandgap, the energy gap between its valence and conduction bands, makes it extraordinarily tolerant to radiation and thermally stable, properties that conventional dosimetric materials struggle to match. The survey covers detector architectures and experimental results spanning photon, proton, and carbon-ion radiotherapy, as well as FLASH treatment techniques. As radiotherapy evolves toward ever more demanding dose rates and particle species, the physical robustness of silicon carbide positions it as a serious alternative for the next generation of clinical dosimeters.

Looking toward future collider experiments, another contribution addresses the challenge of four-dimensional tracking, capturing both the spatial position and precise arrival time of particles passing through a detector. The review highlights three-dimensional silicon sensors, in which electrode columns penetrate the sensor bulk rather than sitting on its surface, as meeting key performance criteria for spatial and temporal resolution under unprecedented radiation levels. Achieving this performance requires careful co-design of sensors and readout electronics alongside advanced fabrication techniques, an integrated approach that INFN has championed across its detector programs.

Beyond traditional semiconductors, the collection showcases novel materials and unconventional approaches. A mini-review examines polysiloxane-based scintillators, flexible polymers that emit light when struck by ionizing radiation. These materials combine radiation hardness, mechanical flexibility, and chemical versatility, with emerging applications ranging from neutron detection to real-time dose monitoring in proton therapy. Recent advances in photocurable and 3D-printable formulations mean researchers can now print custom scintillator geometries, dramatically expanding the design space of polymer-based radiation sensors and opening possibilities for detectors shaped to fit specific anatomical or experimental constraints.

The ARCADIA collaboration contributes advances in fully-depleted CMOS monolithic sensors, devices that integrate sensing and electronics on a single chip of silicon. By combining fast charge collection, low power consumption, and embedded processing, these sensors address the needs of high-rate particle and photon detection, with applications spanning time-of-flight measurements to X-ray imaging. Meanwhile, the ASIX analog spectral imager for X-rays pushes imaging performance in a different direction, adopting an ultra-low-noise, fully analog, asynchronous readout architecture to overcome the charge-sharing limitations inherent to small-pixel devices. The technology demonstrator targets simultaneous high spatial and energy resolution for demanding applications in synchrotron science, materials research, and soft X-ray astrophysics.

The collection also ventures into quantum territory. One article describes integrated optics for generating squeezed vacuum states in lithium-niobate waveguides, compact and robust sources that reduce quantum noise below classical limits in one quadrature of an optical field. Such sources could enhance precision measurements, including gravitational-wave detection, and represent a promising platform for future quantum sensing. Rounding out the collection, miniaturized tissue-equivalent proportional counters developed at the INFN National Laboratories of Legnaro enable microdosimetry in hadron therapy environments, linking physical dose to biological effectiveness at the cellular scale, while INFN’s activities within the international Medipix4 collaboration advance next-generation hybrid pixel detectors based on Timepix4 and Medipix4 application-specific integrated circuits, promising enhanced spatial, temporal, and energy resolution across a broad range of applications. Together, these twelve contributions paint a picture of a research community whose sensor innovations ripple outward from fundamental physics into medicine, space science, and quantum technology, confirming that the instruments built to catch the faintest signals from the universe are also among the most transformative tools for life on Earth.

Subject of Research: Particle and radiation sensor technologies developed within INFN's CSN5 commission, spanning silicon detectors, scintillators, dosimetry, and quantum sensing

Article Title: Editorial: Particle and radiation sensors developed at INFN-CSN5

Article References: Adriani, O., Guidi, V., & Quaranta, A. (2026). Editorial: Particle and radiation sensors developed at INFN-CSN5. Frontiers in Sensors, 7, Article 1814474. https://doi.org/10.3389/fsens.2026.1814474

Image Credits: AI Generated

DOI: 10.3389/fsens.2026.1814474

Keywords: INFN, particle detectors, radiation sensors, LGAD, silicon carbide, dosimetry, radiotherapy, CMOS sensors, quantum sensing, scintillators, X-ray imaging, microdosimetry

The study, published in BMC Medical Imaging by a multidisciplinary group from Xinjiang Medical University and its Sixth Affiliated Hospital, introduces a model called WEM-Mamba, short for Wavelet-Enhanced MedMamba. The work sits at the intersection of two of the hottest trends in medical deep learning: state space models, the efficient new architecture family that includes Mamba, and frequency domain analysis, which extracts information about textures and edges that spatial-domain models often overlook. According to the authors, most existing approaches, whether convolutional neural networks, vision transformers, or earlier Mamba variants, concentrate almost exclusively on modeling features in the spatial domain, leaving the rich frequency content of medical images underexploited. For a task like rotator cuff classification, where subtle textural signatures in the radiograph may hint at tendon damage, that omission could mean losing exactly the evidence that matters most.

To understand why frequency matters, it helps to think about what an X-ray actually encodes. Radiographs are dense grids of intensity values, and while clinicians read them as pictures of bones and soft tissue shadows, mathematically they can be decomposed into components that vary at different spatial scales and orientations. Fine, rapidly changing patterns correspond to high-frequency content such as edges and fine texture; smooth, slowly varying patterns correspond to low-frequency content such as broad anatomical structures. The Haar wavelet transform, a classical mathematical tool that dates back decades, provides a principled way to split an image into these frequency bands while keeping the spatial location of each feature intact. WEM-Mamba exploits this by embedding a Haar wavelet decomposition step directly inside its core processing module, so the network can examine the image simultaneously through spatial and frequency lenses.

Technically, the framework builds on MedMamba, a medical imaging variant of the Mamba architecture. Mamba and its relatives are built on state space models, which process sequences, including sequences of image patches, using recurrent-style computations that scale linearly with input length rather than quadratically as attention mechanisms in transformers do. That efficiency makes state space models attractive for medical imaging, where high-resolution images and limited training data are the norm. The researchers’ key innovation is the Wavelet-Enhanced SS-Conv-SSM module, or WESS. In this module, an input feature map is decomposed by the Haar wavelet transform into multiple frequency sub-bands. These sub-bands are processed and then synergistically fused with the global context modeling that the state space model provides, before an inverse wavelet transform can reconstruct the enhanced representation. The result is a hybrid design in which convolutional operations capture local detail, the state space model captures long-range dependencies across the image, and the wavelet pathway injects multi-band frequency features that neither of the other two components would extract on their own.

The team evaluated WEM-Mamba on a dataset of shoulder X-ray images collected retrospectively from routine clinical examinations at the Sixth Affiliated Hospital of Xinjiang Medical University, with ethics approval and anonymized patient data. They benchmarked their model against fifteen representative competitors spanning the major architecture families: convolutional neural networks, transformers, multi-layer perceptrons, and Mamba variants. WEM-Mamba came out on top, achieving an accuracy of 0.8950, an F1-score of 0.9309, a precision of 0.9078, a recall of 0.9552, and an area under the curve of 0.9116. Notably, the recall figure, which measures how many true tears the model successfully flags, is the highest of its headline metrics, a property that matters greatly in screening contexts where missed diagnoses carry the greatest cost. Equally important for real-world deployment, the model is compact: it contains 14.92 million parameters and requires only 2.04 gigafloating-point operations per inference, a footprint modest enough to run on ordinary clinical hardware rather than specialized computing clusters.

The performance-to-cost ratio is one of the study’s most compelling aspects. Deep learning models in medical imaging have often chased accuracy at the expense of size and speed, producing architectures too heavy for the very hospitals that need them most. By combining the linear-scaling efficiency of state space models with the lightweight Haar wavelet transform, the authors demonstrate that careful architectural design, rather than brute-force scale, can deliver strong classification performance on modest computational budgets. For clinics in underserved regions, where shoulder X-rays are already the first-line examination but specialist radiology expertise may be scarce, a compact model that augments the initial radiograph with a tear probability estimate could meaningfully accelerate referrals for MRI and definitive treatment.

The research also carries a broader message for the machine learning community. Wavelet-based feature extraction has a long pedigree in signal processing and image compression, but it has been somewhat eclipsed in the deep learning era by purely learned representations. The success of WEM-Mamba suggests that classical mathematical transforms and modern sequence models are not rivals but complements. The wavelet decomposition imposes a physically meaningful inductive bias, telling the network explicitly where the fine textures and edges live, while the state space model supplies the flexible global reasoning that fixed transforms cannot. In domains where training data are limited, as is nearly always the case in medicine, building such structural knowledge into the architecture may prove more data-efficient than expecting a generic network to rediscover it from scratch.

Caveats remain, and the authors are candid about them. The study was conducted on a single-center dataset, and all conclusions are explicitly framed as valid under the data and experimental settings of this particular investigation. The researchers state that further validation on larger-scale, multi-center, and external datasets is required before the approach can be considered robust, and that the clinical application value of the framework needs continued assessment in subsequent research. Single-center datasets can encode site-specific scanner settings, patient demographics, and imaging protocols, all of which can inflate apparent performance when a model is tested on data from the same source. Multi-center trials with external test sets are the standard next step for any diagnostic AI aspiring to clinical use, and this work is no exception.

Nevertheless, the study represents a thoughtful and timely contribution to computer-aided diagnosis of musculoskeletal injuries. It addresses a genuine clinical bottleneck, the gap between cheap first-line X-rays and definitive but costly MRI, with a method that is technically novel, computationally lean, and empirically competitive. The work was supported by a range of Chinese research funding programs, including the National Natural Science Foundation of China and multiple Xinjiang regional initiatives, and the article is published open access, making the full technical details available to researchers worldwide. If the promised multi-center validation bears out the single-center results, wavelet-enhanced state space models like WEM-Mamba could become a standard tool in the radiology toolbox, helping ensure that a patient’s first, cheapest image is not the last word on their shoulder pain.

Subject of Research: A wavelet-enhanced Mamba-based deep learning framework for classifying rotator cuff tears in shoulder X-ray images

Article Title: WEM-Mamba: a Wavelet-Enhanced MedMamba framework for rotator cuff tear classification in X-ray images

Article References: Wang, Z., Liu, Z., Li, J., Aboduaini, Y., Abulaiti, A., Li, S., Yakufu, M., Halike, A., Xv, L., & Li, G. (2026). WEM-Mamba: a Wavelet-Enhanced MedMamba framework for rotator cuff tear classification in X-ray images. BMC Medical Imaging. https://doi.org/10.1186/s12880-026-02740-2

Image Credits: AI Generated

DOI: 10.1186/s12880-026-02740-2

Keywords: rotator cuff tear, deep learning, Mamba, state space model, wavelet transform, frequency domain features, X-ray imaging, medical image classification, computer-aided diagnosis, shoulder pain, MedMamba, BMC Medical Imaging

The idea was proposed by David Klyshko in 1988 and verified experimentally in 1995 by Pittman and colleagues, who used pairs of entangled photons to form images of an object that neither photon had individually interacted with in a conventional way. For years, ghost imaging seemed inseparable from quantum entanglement. But researchers soon discovered that classical light sources—thermal lamps, polarized lasers, and laser beams passed through rotating diffusers to create pseudo-thermal speckle—could do the same job. The pivotal moment came in 2008 with computational ghost imaging, which dispensed with the reference arm entirely. Instead of measuring a reference beam, the system projects known patterns onto the object and reconstructs the image from the bucket signal alone, using the patterns as a computational reference. That simplification turned ghost imaging into a practical single-pixel imaging technology.

Mathematically, the reconstruction is straightforward. The image value at each position is obtained by correlating the bucket-detector readings with the known or measured intensity patterns, subtracting the product of the mean signals to isolate genuine fluctuations. Refinements followed quickly. Differential ghost imaging weights the correlation by the total source intensity, suppressing background noise and improving results for highly transparent or reflective objects. Normalized ghost imaging standardizes the reference intensity distribution, making the reconstruction insensitive to absolute brightness fluctuations. Compressive sensing, also introduced in 2008, exploits the fact that most natural images are sparse in some mathematical basis: by acquiring far fewer random measurements than the Nyquist sampling theorem would demand and solving a nonlinear optimization that favors sparse solutions, researchers can reconstruct images from dramatically reduced data. Deep learning entered the field in 2017, when Lyu and colleagues trained a neural network to denoise and refine low-sampling-rate ghost images, opening the door to reconstruction methods that now routinely outperform handcrafted algorithms.

Why go to all this trouble when cameras are everywhere? The answer, the review argues, is that conventional multi-pixel detectors such as CMOS and CCD sensors are optimized for narrow wavelength ranges and degrade badly outside them. In the extreme ultraviolet, at terahertz frequencies, and in much of the mid-infrared, fast high-resolution detector arrays are expensive, scarce, or simply nonexistent. A single-pixel detector, by contrast, can be chosen for peak sensitivity at almost any wavelength, and the simplicity of the setup reduces cost and complexity. Ghost imaging therefore thrives precisely where traditional imaging hardware struggles—decoupling illumination from detection and letting computation do the heavy lifting.

The technique has even crossed from photons to matter. In 2016, Khakimov and colleagues produced correlated pairs of metastable helium atoms by colliding a Bose–Einstein condensate, achieving micron-scale ghost images of a mask with massive particles. Follow-up work showed that higher-order correlations among ultracold atoms can boost image visibility without sacrificing resolution. Neutron ghost imaging, demonstrated on a reactor beamline by Kingston and colleagues, exploits the penetrating power of thermal neutrons, which pass through dense metals yet highlight light elements such as hydrogen—complementing X-rays in nondestructive materials evaluation while reducing dose. Electron ghost imaging, realized by Li and co-workers using patterned electron beams driven by a digital micromirror device, promises to cut acquisition time and specimen damage by orders of magnitude relative to conventional electron microscopy, a boon for radiation-sensitive samples.

X-ray ghost imaging may be the most consequential branch for medicine and materials science. X-rays penetrate soft tissue far more readily than bone or metal, but they also ionize, making radiation dose a critical constraint. In 2016, two groups independently achieved X-ray ghost imaging: Yu and colleagues performed lensless Fourier-transform ghost imaging with pseudo-thermal hard X-rays, while Pelliccia and colleagues split a synchrotron beam in Laue diffraction geometry and exploited natural speckle in the source. By 2018, Zhang and colleagues had demonstrated tabletop X-ray ghost imaging with dramatically reduced radiation, and a follow-up study combined Hadamard-patterned illumination with a multi-level wavelet convolutional neural network to reach roughly 10-micrometer resolution using only 18.75 percent of the Nyquist sampling rate—conditions the authors suggest could suit early-stage cancer detection. Polycapillary optics have since tripled resolution by shrinking the illumination speckle from 166 to 55 micrometers, and ghost tomography has extended the method to three-dimensional volumetric reconstruction. At free-electron lasers, ghost imaging has reached the extreme ultraviolet, enabling nanometer-scale imaging of radiation-sensitive samples.

In the visible and near-infrared, where cameras are cheap and excellent, ghost imaging finds its niche in hostile environments. Underwater, blue and violet light between 400 and 500 nanometers penetrates best, and compressive computational ghost imaging with Hadamard patterns and wavelet-domain enhancement has recovered images through turbid water; temporal ghost imaging has even carried error-free optical data across meter-scale underwater channels. Structured beams such as pseudo-Bessel rings and Lorentz beams, which self-heal after obstruction, improve contrast at depth. Deep learning has transformed the visible band: self-supervised dual networks now extract task-relevant features from raw speckle without labeled training data, block-wise networks iteratively refine multi-scale sub-images using only the bucket signal as supervision, and multi-input mutual-supervision frameworks co-train random-pattern and bucket-signal branches to correct artifacts. Multi-polarization fusion networks capture bucket signals under several polarization states and exploit their statistical differences to reconstruct scenes through dynamic scattering media.

The infrared story is equally striking. Because infrared light rides atmospheric windows, it sees through fog, smoke and dust while avoiding photo-damage—ideal for surveillance, remote sensing and diagnostics. Radwell and colleagues replaced the focal-plane array with a single indium-gallium-arsenide detector in a dual-band microscope, then pushed to 10-hertz real-time video through smoke and tinted glass, and later captured video-rate images of methane gas leaks at the 1.65-micrometer absorption line. Near-infrared ghost-imaging LiDAR has produced centimeter-resolution depth maps from a moving aircraft, and broadband single-pixel hyperspectral cameras classify concealed chemicals across 900 to 1700 nanometers at milliwatt illumination. In the mid-infrared, where digital micromirror devices fail and fast detectors are rare, Wu and colleagues used difference-frequency generation to transfer random patterns from a near-infrared signal to a mid-infrared idler between 3.2 and 4.3 micrometers, enabling ultrafast temporal ghost imaging without high-speed electronics. Graphene metasurface modulators now steer centimeter-scale beams at gigahertz rates, and passive thermal ghost imaging reconstructs hot objects in complete darkness from their self-emitted radiation.

Terahertz ghost imaging completes the spectrum. Terahertz waves pass through textiles, polymers and paper while carrying molecular-fingerprint contrast, making them attractive for security screening, pharmaceutical inspection and cultural heritage diagnostics. Chan and colleagues inaugurated the field by replacing pixelated arrays with a single photoconductive antenna and random printed masks; spinning disks, optically controlled silicon modulators, electrically tunable metamaterials and spintronic emitter arrays have since pushed acquisition toward video rate and achieved deep-sub-wavelength near-field resolution. Nonlinear ghost imaging schemes now fuse terahertz generation with time-resolved field sampling, targeting hyperspectral micro-volumetry of semitransparent samples. Water-vapor absorption, scarce modulators and heavy computational loads remain hurdles, but reconfigurable metamaterials, low-noise detectors, wavefront shaping and physics-informed neural networks are converging toward compact, real-time systems. Taken together, the review concludes, multi-wavelength ghost imaging has become a powerful low-dose alternative wherever conventional focal-plane arrays fall short—imaging with light, and with matter, in places no camera can go.

Subject of Research: Multi-wavelength ghost imaging using single-pixel correlation-based reconstruction across the electromagnetic spectrum and matter waves

Article Title: Multi-wavelength ghost imaging: a review

Article References: Tian, T., Oh, S., & Spielmann, C. (2025). Multi-wavelength ghost imaging: a review. Vicinagearth, 2(1), Article 4. https://doi.org/10.1007/s44336-025-00013-0

Image Credits: AI Generated

DOI: 10.1007/s44336-025-00013-0

Keywords: ghost imaging, single-pixel imaging, computational ghost imaging, X-ray imaging, terahertz imaging, infrared imaging, compressive sensing, deep learning, quantum imaging, matter waves, low-dose imaging, correlation imaging

Writing in Nature Photonics, Jiaqi Liu and colleagues, working under the supervision of Guangda Niu of the Wuhan National Laboratory for Optoelectronics at Huazhong University of Science and Technology and Mengling Xia of the Wuhan University of Technology, describe a design strategy built on so-called zero-dimensional structured materials. In such crystals, the functional building blocks are isolated molecular units that do not share bonding directions with their neighbours in any crystallographic orientation. Because each unit is electronically self-contained, cutting or growing the crystal in any direction never leaves behind broken chemical bonds at the surface. The result is a semiconductor that is, in the researchers’ terminology, dangling-bond-free along all crystallographic directions.

The significance of this property becomes clear when one considers how defects arise in ordinary semiconductors. A conventional three-dimensional crystal, such as silicon, shares covalent bonds in every direction, so any grain boundary or surface necessarily terminates those bonds and creates dangling bonds, chemically unsatisfied sites that trap charge carriers and dissipate energy as heat rather than light. Two-dimensional materials such as transition-metal dichalcogenides avoid this problem within their atomically thin planes, but their edges remain problematic. One-dimensional wires solve the edge problem along their axis yet still expose discontinuities at their tips and side surfaces. Even the celebrated halide perovskites, which have revolutionized solar cells and radiation detectors with their remarkable defect tolerance, rely on passivation strategies to tame their interfacial defects rather than eliminating them structurally.

Zero-dimensional materials sidestep this entire problem, but until now they carried a penalty of their own. Because the functional units are electronically isolated, charge and energy transport within them tends to be isotropic, meaning that a crystal grows with similar rates in all directions and adopts compact, blocky morphologies. That isotropy is excellent for electronic performance but awkward for imaging, where the geometry of the scintillator film matters enormously. In X-ray scintillation imaging, light generated deep inside a thick film scatters as it travels to the detector, and lateral spreading degrades spatial resolution. Columnar structures, in which parallel rods act like optical fibres channeling light toward the detector, are the standard remedy, as demonstrated by the structured caesium iodide scintillators used commercially since the late 1990s. Achieving such columnar growth without sacrificing the defect-free electronic character of the material seemed impossible, because the two requirements pull in opposite directions.

The Chinese team’s insight was that the geometry of a crystal and the electronic dimensionality of its structure can be decoupled. To do so, they exploited a subtle chemical lever built into the uranyl ion, a linear O=U=O unit in which a uranium atom is flanked by two strongly bonded oxygen atoms. In the model compound caesium uranyl chloride, Cs2UO2Cl4, the uranium–oxygen double bonds are dramatically stronger and more directional than the uranium–chlorine bonds that link neighbouring uranyl groups. This bonding disparity means the crystal’s growth kinetics differ from one crystallographic direction to another, even though the uranyl units themselves remain electronically isolated. By tuning the balance between the fast-growing and slow-growing facets, the researchers drove the crystals to elongate into one-dimensional rods while preserving the zero-dimensional, dangling-bond-free electronic framework intact.

The concept draws on classical crystal-growth theory that dates back more than a century. In 1901, Georg Wulff proposed that the equilibrium shape of a crystal is determined by the relative surface energies of its facets, and Carl Herring’s theorems of 1951 formalized how surface free energy governs morphology. The team applied this thermodynamic framework to the chemically anisotropic uranyl system, using it to predict which facets would dominate the equilibrium rod shape. They then validated the prediction at the atomic scale with transmission electron microscopy, characterizing the crystal boundaries to confirm that the expected surfaces form and that no problematic bonding discontinuities appear. The work was a collaboration among the Wuhan University of Technology, Huazhong University of Science and Technology and Beijing University of Technology, with Wei Wang and Yue Lu contributing the atomic-scale electron microscopy characterization.

On the computational side, the researchers employed density functional theory calculations with the DFT-1/2 correction method, a technique developed to overcome the systematic underestimation of band gaps in conventional approximations to density functional theory. This allowed them to model the electronic structure of the material accurately enough to confirm that the anisotropic crystal growth does not compromise the isotropic, defect-free electronic properties that make zero-dimensional semiconductors attractive in the first place. The combination of thermodynamic growth modelling, atomic-resolution microscopy and corrected band-structure calculations gave the team confidence that their design principle is physically robust rather than a lucky accident of one particular compound.

The payoff came when the columnar Cs2UO2Cl4 crystals were tested as an X-ray scintillator. Uranyl compounds have attracted growing interest as scintillator materials since 2018, when researchers first highlighted uranium as a distinct metal centre for building intrinsic X-ray scintillators, with the heavy uranium nucleus providing strong X-ray absorption and the uranyl luminescence providing efficient light emission. Subsequent work explored uranium-organic frameworks and uranyl-cluster compounds for flexible and thermally adaptive scintillators, but spatial resolution remained limited by the film structures those materials formed. The new columnar geometry changes that picture decisively: the team’s Cs2UO2Cl4 scintillator achieved a spatial resolution of 34 line pairs per millimetre, a figure that places it among the highest-resolution scintillation imaging media reported and corresponds to the ability to distinguish features finer than fifteen micrometres across.

The practical implications extend across medical diagnostics, industrial inspection and scientific instrumentation. In clinical X-ray imaging and computed tomography, higher spatial resolution translates directly into the ability to visualize finer anatomical structures and subtler lesions at lower radiation doses, because fewer photons are wasted on scattered, blurred light. In industrial and security screening, fine-resolution scintillators enable detection of small defects or threats. The researchers also emphasize the generalizability of their approach: the underlying principle, that anisotropic chemical bonding can be introduced within a zero-dimensional framework to sculpt crystal geometry without disturbing electronic isotropy, is not limited to uranyl halides. Any material family containing chemically inequivalent bonds between electronically isolated units could, in principle, be engineered the same way, opening a route to defect-tolerant semiconductors with customizable shapes for light-emitting diodes, photodetectors and other optoelectronic devices where interfacial defects have long been the performance bottleneck.

The study, published on 25 September 2026 in Nature Photonics, arrives amid intense global competition to improve radiation detection materials, with halide perovskites and their derivatives dominating recent advances. What distinguishes this work is its conceptual reframing: rather than passivating defects after they form, the team designed them out of the material at the structural level, then recovered the anisotropic geometry needed for high-resolution imaging through chemistry rather than through growth tricks that would reintroduce broken bonds. The authors declare no competing interests, and the work was supported by the National Natural Science Foundation of China, the National Key Research and Development Program of China, the Beijing Natural Science Foundation, the Shenzhen Science and Technology Program and Huazhong University of Science and Technology. If the zero-dimensional design philosophy proves as transferable as the authors anticipate, the humble dangling bond, long the bane of semiconductor engineering, may finally lose its grip on the next generation of imaging technology.

Subject of Research: Dangling-bond-free zero-dimensional Cs2UO2Cl4 scintillators for ultrahigh-resolution X-ray imaging

Article Title: Dangling-bond-free, chemically anisotropic semiconductors for ultrahigh-resolution X-ray imaging

Article References: Liu, J., Liao, M., Liu, X., Ma, N., Li, H., Wang, W., Lu, Y., Xu, Y., Luo, G., Tang, J., Xia, M., & Niu, G. (2026). Dangling-bond-free, chemically anisotropic semiconductors for ultrahigh-resolution X-ray imaging. Nature Photonics. https://doi.org/10.1038/s41566-026-02013-y

Image Credits: AI Generated

DOI: 10.1038/s41566-026-02013-y

Keywords: X-ray imaging, scintillators, zero-dimensional materials, dangling bonds, Cs2UO2Cl4, uranyl halide, crystal growth, spatial resolution, defect tolerance, semiconductors, Nature Photonics, optoelectronics

STIX is not a conventional telescope. It is an indirect imaging spectrometer, a device that never forms a direct picture of the Sun. Instead, it observes the bremsstrahlung X-rays emitted by hot flare loops and accelerated electrons in the energy range from 4 to 150 kiloelectronvolts, using pixelated cadmium telluride detectors with an energy resolution of about 1 kiloelectronvolt. The imaging trick relies on 30 pairs of grids, each a stack of tungsten foils etched with fine slits, placed in front of the detectors. As X-rays stream through a grid pair, they cast a shifting Moiré pattern of light and shadow onto the detector behind it. By measuring the contrast and phase of that pattern, each sub-collimator, as a detector-grid unit is called, records one Fourier component of the X-ray source, much as a radio interferometer records a single visibility. Combine enough of these components and an image of the flare emerges from the mathematics.

The trouble is that every Fourier component depends on knowing precisely how transparent each grid pair really is. Before Solar Orbiter launched, engineers measured the grids optically, characterizing the pitch, slit width, and orientation of each one, and those measurements have underpinned the data analysis software ever since. But optical measurements have a fundamental blind spot: they can only see the top and bottom layers of a stack. The STIX grids for the 24 sub-collimators with the coarsest angular resolution are built from tungsten foils 33 or 50 micrometers thick, stacked to reach a total thickness of 400 micrometers, with the pattern etched into each layer. If the layers are not perfectly aligned, or if the etching leaves slits that vary slightly from layer to layer, the effective opening of the grid narrows in ways no optical inspection of the outer surfaces can reveal. Unlike its predecessors on Yohkoh, RHESSI, and the Chinese ASO-S mission, STIX also lacked a full pre-launch X-ray calibration at the small incident angles relevant to solar flare observations, a constraint of time and budget.

The new work turns that limitation into an opportunity. The key insight is that STIX carries a built-in flux monitor it was not entirely designed to be one: the Coarse Flare Locator, or CFL. This sub-collimator uses an H-shaped front grid and an empty rear grid, and for certain flare locations on the solar disk, the shadow cast by the front grid leaves up to three of the CFL’s large detector pixels fully illuminated. A fully illuminated pixel, by definition, sees the total incident X-ray flux, unmodulated by any grid. By comparing the flux recorded by the fully illuminated CFL pixels with the flux transmitted through each imaging sub-collimator, the team could compute the true effective transmission of every grid pair directly from flight data. The transmission is simply the ratio of the transmitted flux to the total flux, a beautifully direct measurement that sidesteps the uncertainties of the ground-based optical characterization.

Building the calibration dataset required patient mining of the STIX archive. From roughly 25,000 flares recorded between January 2021 and February 2025, the team selected events with at least one fully illuminated large CFL pixel, accurate flare locations, and enough counts to keep statistical errors at or below about 3 percent. The winnowing left 91 flares, a small fraction of the total, because only flares at particular solar latitudes illuminate the CFL pixels completely. The team focused on the 10 to 15 kiloelectronvolt band, where the transmission of the Solar Black coating on the spacecraft heatshield is high enough that any inhomogeneities introduce only second-order errors, and where photon statistics remain strong. Careful corrections for background counts, detector live time, and onboard energy channel binning were applied, and a self-consistency check of the CFL pixels themselves showed the total flux measurements are accurate to about 2.3 percent.

What the flares revealed was unambiguous. The measured transmission of the sub-collimators is systematically lower than the values predicted from the optical grid characterization, and the internal shadowing effect, the progressive narrowing of the effective slit width as photons arrive at increasing angles, is far weaker than a model of perfect grids would predict. Simple geometric simulations explain why. When the team modeled grids with random etching and stacking imperfections of a few micrometers, drawn from Gaussian distributions with a standard deviation of about 2 micrometers, the simulated on-axis transmission dropped and the shadowing effect flattened out, exactly matching the pattern in the flight data. For sub-collimator 5, with slits nominally 83 micrometers wide, imperfections of this scale reduce the effective slit width by roughly 6 micrometers. In some sub-collimators, such as 6b, a slight tilt of the stacked layers, plausibly caused by tilted alignment pins during assembly, produces a sloped transmission profile that the simulations reproduce as well.

The practical payoff is substantial. When the team applied the new transmission calibration to 25 independent flares observed between March and December 2025, the total flux estimates from the different sub-collimators agreed within about 2 percent, whereas the old calibration produced systematic biases ranging from 3 to 39 percent for the coarse-resolution sub-collimators and up to 175 percent for the finest ones. A second test on the powerful X2.2-class flare of December 8, 2024 drove the point home: photon spectra derived independently from the individual sub-collimators now agree within roughly 2 percent across the 8 to 20 kiloelectronvolt range, a dramatic improvement over the 6.9 percent scatter of the old calibration. The new calibration also means that derived photon fluxes are about 13 percent higher than previously computed, and fitted emission measures, a measure of the amount of hot plasma in the flare, increase by a similar 13 percent. Fitted flare temperatures, reassuringly, do not change, and the team emphasizes that conclusions from previously published STIX spectral analyses remain valid.

There are honest caveats. The calibration is valid only at low energies, below about 20 kiloelectronvolts, and only for offset angles between roughly minus 0.5 and plus 0.5 degrees, because at higher energies the tungsten grids become partially transparent and the CFL statistics grow too thin. Below 8 kiloelectronvolts, inhomogeneities in the Solar Black coating introduce discrepancies that the team hopes to self-calibrate in a future study using the dependence of transmission on flare location. Three sub-collimators, 3a, 5a, and 5b, showed energy-dependent discrepancies likely tied to detector rather than grid calibration and are recommended for exclusion from spectral analyses for now. The high-energy grid calibration, crucial for studying whether flare X-ray emission is anisotropic, will follow in a companion paper, and the team notes that cross-calibration better than 10 percent between Solar Orbiter and Earth-orbiting observatories is needed to settle that question.

Perhaps the most enduring lesson is a design philosophy. The authors strongly recommend that future indirect X-ray imagers include a dedicated total flux monitor, so that every recorded flare can contribute to self-calibration, as the Hard X-ray Imager aboard ASO-S already does. A flux monitor with sufficient effective area would even allow the same technique to be extended to high energies. For now, the new transmission model will be released in the upcoming version of the STIX analysis software, quietly improving every flare image and spectrum the mission produces. It is a striking demonstration that a spacecraft instrument, far from any laboratory, can diagnose its own imperfections using nothing more than the raw fury of the Sun, and that a few micrometers of tungsten misalignment, invisible to any optical microscope on Earth, can be measured from 100 million kilometers away by letting solar flares do the talking.

Subject of Research: In-flight self-calibration of the tungsten grid transmission of the STIX hard X-ray imaging spectrometer on board Solar Orbiter using solar flare observations

Article Title: In-Flight Self-Calibration of the STIX Grid Transmission

Article References: Massa, P., Volpara, A., Stiefel, M. Z., Bilgili, E., Degen, M., Limousin, O., Hurford, G. J., & Krucker, S. (2026). In-Flight Self-Calibration of the STIX Grid Transmission. Solar Physics, 301(9), Article 138. https://doi.org/10.1007/s11207-026-02739-6

Image Credits: AI Generated

DOI: 10.1007/s11207-026-02739-6

Keywords: Solar Orbiter, STIX, solar flares, X-ray imaging, instrument calibration, tungsten grids, Solar Physics, space instrumentation, hard X-rays, spectroscopy, ESA, helio physics