Optics Letters

The Transport-of-Intensity Equation (TIE) and polarization differential phase contrast (pDPC) exhibit transfer characteristics that limit their use as standalone quantitative phase imaging modalities. TIE provides stable recovery of low spatial frequencies but attenuates fine structure, while pDPC enhances phase gradients but is intrinsically insensitive near zero frequency. We present a hybrid reconstruction framework that combines a single-shot pDPC measurement with an in-focus-defocused TIE image pair and fuses the resulting phase estimates in the Fourier domain using frequency-selective weighting. The illumination geometry is co-designed to enforce angular separation between modalities, enabling high-frequency phase-gradient acquisition in a single exposure while preserving low-frequency sensitivity through low-numerical-aperture symmetric illumination. The proposed method requires three intensity measurements without requiring interferometric stability or sequential asymmetric illumination. Experimental results demonstrate improved phase fidelity across spatial frequencies, reduced halo artifacts, and enhanced recovery of both smooth and fine structural features. This approach provides a compact and robust method that moves towards extended spatial-frequency phase imaging for dynamic and resource-constrained applications.

Optical coherence tomography angiography (OCTA) has been extensively explored for the detection of pathological retinal changes arising in various ophthalmological diseases. Due to phase instability of conventional systems, most OCTA signal processing methods to date have relied on intensity-based averaging, despite the resulting lower contrast and dynamic range compared to coherent averaging approaches. In this work, we (1) present a novel computation pipeline that enables fully-coherent processing in any phase-stable or phase-unstable system, and (2) demonstrate the first angiography metric with integrated signal-to-noise ratio (SNR) correction and coherent spatiotemporal ensemble averaging effective in any OCT system. Our method does not rely on 2D en face projections or binarization to produce high-contrast en face views and it significantly enhances the contrast-to-noise ratio of computed OCTA volumes, particularly in regions with reduced SNR levels. Our OCTA volumes also have high cross-sectional quality, enabling the use of single-depth en face views and the localization of capillary paths in the axial dimension. We can now identify capillaries below the retinal pigment epithelium, near the lamina cribrosa in the optic cup, which was not previously possible due to low SNR. We also conducted an analysis of strategies for bias reduction in OCTA and developed the optimal method utilizing coherent spatiotemporal ensemble averaging. Because our method leverages coherent averaging across all dimensions, it provides flexibility in shaping the averaging kernel to target specific workflows. The value of this additional degree of freedom is illustrated in our vessel tracing investigations, in which the depth resolution was maximized by limiting coherent ensemble averaging to the two lateral dimensions even in presence of strong phase noise, enabling the identification of 3D features such as overlapping capillary paths.

Background and Objective: High-speed and high-resolution retinal imaging using adaptive optics line confocal ophthalmoscopy (AOLCO) resolves individual red blood cells (erythrocytes) flowing through the smallest retinal capillaries, but quantifying their velocity first requires removing linear and nonlinear distortions caused by eye motion that may result in a displacement of tens to hundreds of pixels between frames. We present a computational pipeline that stabilizes AOLCO capillary video and measures erythrocyte velocity in true physical units, delivered as an interactive 3D Slicer extension and a batch command-line tool sharing one analysis core. Methods: Our video analysis pipeline has four stages: preprocessing, registration, postprocessing, and velocity measurement. During preprocessing, illumination is corrected, and closed-eye frames are identified. During registration, each frame is aligned to a reference frame using affine and, when needed, B-Spline registration, initialized by our robust and fast (GPU-based) translation-registration algorithm to accommodate large eye movements (e.g., microsaccades). During postprocessing, adjacent registered frames are differenced to suppress stationary structures, and a 2D projection of the frame-difference video reveals the capillary network. During velocity measurement, an operator manually identifies capillary segments on the 2D projection image. Spatiotemporal images are then generated along these segments from the frame-difference video, and blood-flow velocity curves are obtained from these images using the Radon transform. Results: We validate the velocimetry of the pipeline against synthetic ground-truth flow data, and the experiments show that the velocimetry recovers the speed within an error of 4% The GPU-based translation registration algorithm aligns the large eye motions (~90 um per frame on average) and runs about 3.6 x (100 Hz) to 38 x (400 Hz) faster than the baseline method. Across the full cohort (2344 videos, 57 subjects, 30 Hz to 400 Hz), our registration approach improves structural similarity (SSIM) of the video frames on every acquisition from 0.73 to 0.86. The recovered blood flow velocities are physiologically plausible, of the same order as published AO measurements. Conclusions: Our computational pipeline enables AOLCO velocimetry on the capillary-level.

Gas-bubble-enhanced focused ultrasound (FUSGB) is advancing clinically across a myriad of applications, including targeted drug delivery through transient blood-brain barrier opening (BBBO). Its broader translation requires a better understanding of the bioeffects that enable reproducible enhancement of barrier permeability while avoiding vascular injury, underscoring the need to clarify the relationship between acoustic exposure, bubble activity, and the biological response central to treatment optimization. These interactions are difficult to isolate in vivo, and acoustic reflections, geometry, and field distortion in conventional cultureware hinder quantitative in vitro studies. Herein, we present a theoretically and experimentally characterized ultrasound-transparent organ-on-chip device with >99.9% US transparency that enables quantitative in vitro modeling of the FUSGB procedure and study of BBBO. Furthermore, submicron bubbles (sMB, ~800nm) were fabricated and characterized. Acoustic dose-dependent subharmonic and broadband cavitation dose curves (SCD and BCD), as well as temporal responses, were studied. A 3-day accelerated Caco--2 surrogate barrier with a high TEER (~1200xcm2) was used to demonstrate US dose-dependent targeted barrier opening, visualized using an in vitro Evans Blue assay. Together, this study and its findings establish a foundation for quantitative in vitro modeling of FUS BBBO, as well as other US interactions in microphysiological systems, supporting systematic evaluation of procedural regimens, large-parameter-space optimization, acoustic safety window, and exposure conditions for brain drug delivery, and additionally enabling mechanistic insight into FUSGB BBBO.

Significance: Fluorescence imaging remains largely qualitative and device-specific, limiting reproducibility and intersystem comparisons. SI-traceable radiometric characterization requires calibrated sources suitable for transferring known radiance to imaging systems. Aim: Establish and evaluate a compact, calibrated solid-state radiometric emitter target (RET) for SI-traceable radiance transfer, and assess methods for evaluating radiance invariance, imager responsivity, and extension of radiance transfer to digitally defined regions of interest (ROIs). Approach: The radiance invariance of an LED-based RET was evaluated by measuring its radiance across varying source-to-aperture distances and collection apertures. A previously described radiance-transfer formalism based on calibrated source radiance and collection geometry was implemented to determine imager responsivity (R{lambda}), which was evaluated across varying working distances and lens apertures. Grubbs and MAD outlier tests and one-way ANOVA were used to assess radiance and responsivity invariance. Finally, responsivity obtained using digitally defined ROIs was compared with matched physical apertures to assess ROI-based radiance transfer. Results: The measured RET radiance remained invariant across the tested source-to-aperture distances and collection apertures, with no significant differences observed across distance (p = 0.88) or aperture (p = 0.15). Imager responsivity remained stable across working distances for both tested lens configurations (p = 0.48 and p > 0.99). Responsivity was also invariant across lens apertures, except for deviations at f/11 that were consistent with practical aperture tolerances. Digitally defined ROIs reproduced responsivity values obtained using matched physical apertures, demonstrating that the radiance-transfer approach can be extended to image-defined regions. Conclusions: The presented methods provide an approach for evaluating compact LED-based sources for invariant-radiance emission and suitability for SI-traceable radiance transfer. The characterized RET supported stable imager responsivity across changes in imaging geometry and extension to digitally defined ROIs. These methods can be adapted to other LED-based sources, with further angular characterization and uncertainty analysis needed to establish broader applicability.

Image scanning microscopy (ISM) is emerging as the natural evolution of confocal laser-scanning microscopy, combining optical sectioning and a two-fold improvement in lateral resolution with high signal-to-noise ratio. Single-photon avalanche diode (SPAD) array detectors enable an effective implementation of ISM by recording a micro-image at each scan position with single-photon sensitivity, negligible readout noise, and without the fixed frame-readout constraints of conventional cameras. Nevertheless, SPAD array-based ISM remains largely confined to laboratories with expertise in microscope development. Here, we present BrightEyes-Scope, a reference implementation of a SPAD array-based ISM system. We provide step-by-step guidelines for optical design, system construction, alignment, calibration, data acquisition, and image reconstruction. Presented as a complete ISM microscope, BrightEyes-Scope also serves as a reference for upgrading existing laser-scanning microscopes. We anticipate that it will accelerate the adoption of ISM for quantitative super-resolution imaging.

This study presents a systematic investigation of the brain's own microwave emissions in the 0.85-5.0 GHz range, aiming to validate microwave encephalography (MWEG) as a novel tool for cognitive state assessment. Early experiments (2016-2019) using comparisons with background noise, a control phantom, and different antenna types confirmed the endogenous origin of the recorded signals. In a 2021 experiment (118 measurements on one subject, right and left occipital areas), we observed a significant functional hemispheric asymmetry: the number of "substantial spikes" was significantly higher in darkness for the right hemisphere (p<0.001) and higher in light for the left hemisphere (p<0.001). A 2025 group study (three subjects) confirmed the asymmetry (p<0.05 for right, p<0.01 for left). We describe the signal processing pipeline, including spike detection and statistical analysis. Two complementary mathematical models are presented: an energy-based model linking optical input to microwave emission via neural activity modes, and an information-based model interpreting spectral changes in terms of structural pattern overlap with memory engrams. An estimation of the physical dimensions of possible brain resonators (2-22 mm) is provided. The results are discussed in the context of hemispheric specialization and the information commutation theory. This work opens new perspectives for developing low cost, contactless brain monitoring systems.

Abstract Background Advanced in vitro models of the human heart have become promising tools for studying cardiac diseases and drug discovery. Three-dimensional (3D) engineered heart tissues (EHTs) provide more adult cardiac tissue-like structural organization compared to traditional two-dimensional (2D) in vitro models with straightforward contractile readouts. However, the assessment of the architecture of EHTs, especially the quantification of cellular and tissue structural components, has remained poorly studied. Here we demonstrate the use of light-sheet microscopy for structural imaging of intact EHTs and provide quantification of the EHT cellular and subcellular structures. Methods We developed an in-house inverted selective plane illumination microscopy (iSPIM) platform and combined it with iterative indirect immunofluorescence (IT-IF) staining for better signal acquisition. We developed an image-processing and analysis pipeline dedicated to volumetric EHT analysis. Results With this integrated platform, we achieved volumetric imaging of intact EHTs and quantification of 3D tissue architecture, revealing aligned cardiomyocytes within the imaged volume and increasing cardiac nuclei content towards the core of the tissue. The developed methods allowed the detection of heterogeneities between tissue sections, demonstrating the sensitivity required for future disease modeling or maturation studies. Conclusions The developed integrated platform opens new possibilities for volumetric analysis and characterization of 3D tissue structure using iSPIM, providing new insights into EHT architecture and the use of EHTs as disease modeling tools.

In this paper, we introduce Nagini3D, a 3D biological image segmentation method that hybridizes two high-performing yet highly complementary segmentation paradigms: deep learning-based approaches and snake-based methods. Our approach leverages both the high efficiency of data-driven techniques and the convenient shape representation provided by active surfaces, which proves highly practical for incorporating a priori knowledge as well as for the post-analysis of segmented objects. By considering a complete 3D volume, our method ensures segmentation consistency across all three dimensions, unlike many approaches that handle volumes in a slice-by-slice manner. Additionally, it theoretically enables the representation of any object with spherical topology, overcoming the limitations of common 3D methods that rely on object convexity assumptions. Furthermore, the continuous representation of objects allows for a resolution-independent description, facilitating the extraction of various geometric features of interest (e.g., local curvature, principal deformations). We demonstrate both the quality of the segmentation produced by our method and the richness of the accessible geometric features through an extensive set of experiments conducted on multiple real and simulated datasets, including objects with challenging geometries.

Objective: To develop and validate a dual-probe Diffuse Correlation Spectroscopy (DCS) system for non-invasive and simultaneous, monitoring of cerebral blood flow (CBF) in the bilateral Middle Cerebral Artery (MCA) territories, and expanding the utility of conventional DCS limited to cortical-volume-based CBF measurements to vessel-specific cerebral perfusion monitoring. Methods: A dual-probe DCS system was designed for non-invasive monitoring of MCA-specific perfusion. Probe placement and protocol optimization study has been performed using anatomical landmarks, motor and speech activation tasks in healthy volunteers. System stability and repeatability were further evaluated in a pilot cohort of 30 healthy (age, 25{+/-}7 years) participants using optimized probe position and protocol. A bilateral MCA ischemic Lacunar Infract stroke case report also validated the feasibility of the system in clinical settings. Results: Measurements demonstrated superior sensitivity towards MCA-territory perfusion at targeted probe locations compared to off-MCA positions. In pilot cohort, significant increase of 30.34 {+/-} 21.56% and 36.48 {+/-} 21.22% in rCBF corresponding to hand squeeze and speech task respectively showed reproducible physiological responsiveness of the system (p<0.001). Measurement done on a patient with bilateral MCA ischemic Lacunar Infract stroke showed a significant change of 30% during speech for both the MCAs but no significant change is observed for hand squeeze tasks. Conclusion: The custom built dual-probe DCS system enables non-invasive, operator-independent, targeted and continuous monitoring of rCBF within bilateral MCA territories. Significance: This approach enables the potential use of DCS system for bilateral and vessel-specific monitoring of cerebral perfusion in the MCA territories.

Point-of-care (POC) molecular detection has enabled rapid and accessible testing at or near the site of need, facilitating timely decision-making in clinical, field, and industrial settings. Despite these advancements, molecular detection platforms remain limited in their ability to prepare DNA from complex tissue matrices, restricting their applicability to a narrow range of sample types. To address sample preparation inefficiencies, focused ultrasound extraction (FUSE) has been developed for the rapid release of DNA from robust tissue matrices. FUSE disintegrates target tissue and releases DNA by delivering high pressure ultrasound pulses that generate a cavitation bubble cloud. In prior work, FUSE was performed using large, laboratory-based equipment. Here, a miniaturized FUSE device was designed and tested to enable POC DNA sample preparation. Design constraints were defined based on prior FUSE experiments, and based on these metrics, a 750 kHz cylindrical transducer was fabricated. The pressure output was measured, and high-speed optical imaging demonstrated that the transducer generated sustained cavitation within the defined focal region. The feasibility of DNA extraction from piscine and timber tissues was evaluated, and results show that the device can release DNA with quality suitable for PCR-based detection from both tissue types. These results demonstrate the potential of this device to rapidly extract DNA from complex samples and expand the applicability of molecular detection platforms.

Abstract: Purpose: Superparamagnetic iron oxide nanoparticles (SPION) are widely used as MRI contrast agents for cell tracking and molecular imaging. However, their detection using conventional T2/T2* weighted and susceptibility-weighted MRI relies on negative contrast (signal voids), which lacks specificity because similar hypointense signals can arise from other sources of magnetic susceptibility. This study introduces Frequency Shift Imaging (FSI), a positive-contrast technique based on an adiabatic Shinnar-Le Roux (SLR) self-refocused pulse that addresses the limited specificity of conventional SPION imaging. Methods: An adiabatic self-refocused RF pulse was designed using the SLR algorithm for frequency-selective positive-contrast imaging. The optimized pulse had a spectral bandwidth of 370 Hz, a peak RF amplitude of 14 T, and a duration of 15 ms. FSI data were acquired on a 7T MRI scanner. Phantom experiments used agarose-embedded 30 nm cell-labeled SPION at 0%, 25%, 75%, and 100% concentrations (pure). In vivo validation was performed following intramuscular SPION injection into the hind limb of a mouse. FSI images were compared with conventional acquisitions. Results: Quantitative phantom analysis showed normalized FSI signal intensities of 25.02% and 75.16% for samples containing 25% and 75% labeled cells, respectively, demonstrating an approximately proportional relationship between FSI signal and labeled-cell concentration. In vivo, FSI generated localized hyperintense positive contrast at the SPIO injection site, providing improved conspicuity and localization compared with the corresponding negative-contrast signal void. Discussion and Conclusion: FSI enables robust, B1-insensitive, positive-contrast SPION imaging at 7T with reduced echo time and SAR compared to paired adiabatic approaches, supporting its potential for quantitative cell tracking and molecular imaging applications.

In magnetic particle imaging (MPI), signal is generated from the non-linear magnetic response of superparamagnetic iron oxide (SPIO) nanoparticles. The optimization of SPIOs for MPI is an active area of investigation. Tracer performance can be evaluated by magnetic particle relaxometry (MPR) and image-based metrics. This study characterized and compared the performance of ten commercially available tracers, spanning a broad range of magnetic core sizes, hydrodynamic diameters and surface coatings, using the peak signal intensity measured from MPR, and the total signal and maximum signal intensity measured from 2D MPI images. Synomag-D tracers exhibited the highest MPR peak signal intensities and highest maximum signal per {micro}g iron. ProMag, a micron-sized iron oxide particle (MPIO), yielded a low MPR peak signal intensity but had the highest total MPI signal per {micro}g of iron. A strong correlation was observed between MPR peak signal intensity and MPI maximum signal intensity (R2=0.94). In contrast, there was a weak correlation between MPR peak signal intensity and MPI total signal intensity (R2=0.38), though this correlation improved when MPIOs were excluded from the analysis (R2=0.91). While MPR generally predicts the tracer performance, it does not completely replicate the complex imaging conditions. As such, comprehensive tracer evaluation requires combining both MPR and image-based metrics.

Cardiac amyloidosis is caused by deposits of misfolded proteins in the heart and can lead to heart failure. Histological confirmation typically involves staining tissue sections with Congo red and examining them under polarized light for apple-green birefringence; however, achieving an interpretable, label-free, and observer-independent assessment remains challenging. Here we show that Pearson kurtosis of channel intensities within the amide I window provides an interpretable, fitting-free contrast for cardiac amyloid in unstained human cardiac tissue sections from ten participants imaged by ultrabroadband multiplex coherent anti-Stokes Raman scattering (CARS). In a discovery participant, high-kurtosis regions corresponded with Congo red-positive tissue in two CARS-histology pairs of adjacent sections (descriptive spatial AUCs, 0.79 and 0.73). Across 15 fields from nine additional participants, field-level kurtosis values did not overlap between severe-deposition and Congo red-negative tissue fields. Among the descriptors examined, kurtosis showed the largest standardized participant-level separation. Exploratory multivariate curve resolution yielded a deposit-associated component with aromatic-residue features shared with reagent transthyretin. In one field from the discovery participant, the second harmonic generation (SHG) signal, acquired simultaneously with CARS, showed a broadly inverse spatial pattern relative to the amide I kurtosis map, revealing distinct tissue-architectural contrast. Together, these findings support a potential label-free approach to detecting cardiac amyloid deposits that combines amide I kurtosis derived from broadband CARS with SHG-derived architectural contrast.

Vector flow mapping (VFM) reconstructs left-ventricular (LV) blood velocity from color-Doppler echocardiography by combining the measured beamwise component with physical and regularizing constraints. Analysis of the discrete VFM formulation shows that the inverse problem is intrinsically ill posed: the occurrence of singular modes can be predicted from the geometry of the segmented blood-pool domain, the imposed boundary conditions, and the degree of smoothing. These modes can propagate uncertainty along entire transverse bands of the reconstructed velocity field, yet conventional VFM neither quantifies this uncertainty nor allows for correcting the blood-pool segmentation. We introduce Bayesian VFM (B--VFM), a hierarchical framework that jointly infers radial and transverse velocities, a probabilistic blood-pool mask, their spatially resolved uncertainties, and hyperparameters weighting Doppler and segmentation fidelity, mass conservation, boundary conditions, and smoothness. The discretized posterior admits a closed-form gradient and exact Hessian, enabling computationally efficient, gradient-based MAP estimation, sampling, and direct analysis of ill-posed modes. Posterior inference combines Gibbs sampling of conjugate Gamma-distributed hyperparameters with conditional maximum-a-posteriori estimation and a Laplace approximation for the high-dimensional velocity and mask fields. To accommodate systematic departures from planar mass conservation, B-VFM can learn the covariance of the planar divergence residual from an ensemble of flows and incorporate it as a structured model-discrepancy prior. Independent chains converged reproducibly, while covariance priors learned from flow ensembles illustrated how model discrepancies can be incorporated into the inference. B--VFM was evaluated using Lamb-Chaplygin dipoles under ideal conditions and with Doppler corruption, Doppler voids, and segmentation defects, and using the Hicks-Moffatt family of spherical vortices to assess violations of planar mass conservation. The method produced smooth reconstructions, localized uncertainty near unreliable measurements and regions of model inconsistency, and used flow information to correct segmentation errors. Within the tested vortex family, the data-informed planar divergence prior reduced velocity bias and mask distortion. B--VFM thus provides an uncertainty-aware reconstruction method and a flexible foundation for future VFM formulations incorporating additional priors, observations, and physical models. Future work will evaluate the method using clinical data and more complex three-dimensional benchmark flows.

Purpose: Magnetic Resonance Acoustic Radiation Force Imaging (MR-ARFI) encodes focused ultrasound (FUS)-induced tissue displacements to MR phase shifts. Because these phase shifts are small, optimizing displacement sensitivity and minimizing motion artifacts requires characterization of the temporal dynamics of acoustic radiation force (ARF). Beyond sequence design, such characterization enables MR-ARFI to serve as a probe of brain viscoelastic behavior and can inform the design of therapeutic ultrasound pulsing schemes by quantifying how rapidly tissue deforms and relaxes. Methods: We developed a dynamic model assuming a gamma-variate impulse response for ARF(t), with time constant ({tau}) to be estimated. A single-shot spin echo ARFI acquisition with a bipolar motion encoding gradient (MEG) was implemented, incorporating retrospective corrections for head motion and physiological noise. A 500 kHz FUS pulse was applied in on/off blocks while systematically varying the timing between the MEG and sonication. The resulting phase time series were fit to the model to estimate time constant in five human participants. To assess sensitivity to tissue stiffness, {tau} was additionally measured in two customized gelatin phantoms of differing stiffness, characterized independently by MR elastography (MRE). Results: The average measured time constant in human participants was {tau} 3.7 {+/-} 0.27 ms. Comparing the two phantoms, the stiffer phantom exhibited a shorter {tau} (3.0-3.5 ms) than the softer phantom (4.0-4.5 ms), which tracked the MRE-derived Kelvin-Voigt viscoelastic time constants. In addition, the peak ARFI displacement was lower in the stiffer phantom (0.17 vs 0.74 m for the less stiff phantom), and a broader displacement focal spot size (FWHM 22 mm vs 15 mm). Conclusion: This time series approach with motion and physiological noise corrections validates an MR-based technique capable of characterizing the brain's temporal response to sonication-induced mechanical forces, providing a foundation for optimizing the design of ultrasound pulsing schemes for ARFI, as well as for future studies of brain biomechanics.

Capturing fast hemodynamics in deep organs is essential for understanding microvascular regulation of organ responses to physiological demands and pathological stress. However, non-invasive three-dimensional (3D) imaging of these microscale processes remains challenging due to trade-offs between spatial resolution, imaging speed, and penetration depth. Here, we present fluctuation-based high-order super-resolution acoustic microscope (FLAME), a tracking-free 3D ultrasound imaging technology capable of fast microvascular angiography and flow measurement. Using as few as 30 volumes, FLAME improves 3D ultrasound resolution by 8-fold (~50 m) and shortens data acquisition time by over two orders of magnitude, from tens of seconds to tens of milliseconds, compared with conventional tracking-based super-resolution approaches1-6. This high-throughput capability supports high-fidelity imaging of tissue functions under flexible experimental settings. FLAME achieves an unprecedented 3D super-resolution frame rate of ~40 Hz, capturing transient hemodynamic responses to various physiological and pathological stimuli in different mouse organs, from the microvascular level to the whole-body scale. With open-sourced implementation, FLAME provides an accessible platform for real-time, in-depth hemodynamic monitoring across diverse biomedical applications and clinical translation.

Trans-arterial radioembolization using yttrium-90 (Y-90) microspheres treats unresectable liver cancer with radiation. The tumor-to-parenchyma dose ratio is mainly governed by the patient-specific hepatic arterial flow distribution, which transports the microspheres. Even though computational fluid dynamics (CFD) simulation can predict this flow distribution, its high computational cost and time limit intraprocedural application. To overcome this computational limitation, a physics-informed neural network (PINN) surrogate was developed to provide rapid predictions of flow distribution in a patient-specific hepatic artery model. The surrogate maps spatial coordinates to the velocity vector U with Cartesian components u, v, w and the static pressure p. Reference velocity and pressure fields came from a converged finite-element CFD solve of the same geometry at about 1.05 x 105 nodes; a controlled fraction was used to supervise the training while the remainder was held out for validation. Using the NVIDIA PhysicsNeMo framework, we compared four coordinate networks (MLP, Modified Fourier, Multi-Scale Fourier and SIREN) under soft and hard inlet enforcement across a supervised data sweep (data fraction df = 0 to 0.20) and repeated df = 0.20 at a matched 300,000-step budget. Predicted flow splits were propagated through an established Y-90 dose-kernel pipeline. Accuracy is governed primarily by the supervised-data fraction, not the architecture: at df [&ge;] 0.10 the backbones agree to within one to two R2 points, and R2(u) rises from {approx} 0 under pure physics (df = 0) to {approx} 0.95 by df = 0.01 for the best backbones, saturating near 0.97 to 0.99 by df = 0.20. At a matched 300,000-step budget the Multi-Scale Fourier network leads (R2(u) = 0.978) ahead of SIREN (R2(u) = 0.976) and the MLP (R2(u) = 0.968); the best value seen (R2(u) = 0.985, Modified Fourier) needed 1.8 times that budget. On the df = 0.20 checkpoint the per-outlet flow split matches CFD (r = 0.999, L1 = 4.4 percentage points) and the derived dosimetry agrees to within 2.4% for perfused volume, mean dose and every isodose volume. At df = 0 training collapses toward the trivial near-zero state, which satisfies continuity, momentum and the zero-pressure outlets exactly and is ruled out only by the inlet and data terms; this, not architecture choice, is the limiting failure mode. Once trained, the network evaluates the full 105,284-node field in about 2 s on a four-core CPU. The supervised-data fraction, with a practical floor between df = 0.05 and 0.10, is the primary design lever for such hepatic-arterial PINN surrogates.

Functional ultrasound (fUS) imaging provides deep, wide-field access to brain function, but instrument cost and complex acquisition and analysis workflows restrict its wider adoption. Here we introduce @fUS, an agentic transcranial imaging platform with an open architecture that integrates purpose-built hardware with executable experimental skills and persistent structured memory. The system combines 128-channel, 16-bit acquisition at 125 MHz with GPU-based processing and enables functional imaging through the intact scalp and skull of mice, with approximately 100-m spatial resolution and 10-Hz temporal sampling, providing a basis for longitudinal studies without cranial-window surgery. Its agent connects scientific objectives to experimental design, direct instrument control and data analysis, retaining context across interactions and supporting inspectable workflows through text and hands-free voice control. Neuroscience trainees with limited engineering experience independently configured the system within 30 min. Agent-assisted exploration of whisker-stimulation datasets revealed low-frequency vascular changes beyond conventional response mapping, supported by complementary two-photon measurements of single-vessel diameter. By combining transcranial imaging with accessible instrument control and analysis, @fUS provides a foundation for wider adoption and larger, more diverse neuroscience datasets. More broadly, it offers a framework for scientific instruments in which measurement, computation and experimental reasoning are developed as parts of the same system.

Medical ultrasound is a portable, non-invasive, and cost effective imaging modality that is particularly well suited for resource-limited settings. Optic nerve sheath diameter (ONSD) measurement from ultrasound is used as a point-of-care assessment tool for intracranial pressure (ICP), which is associated with several neurological conditions. However, valid measurement depends on selecting a frame in which the optic nerve sheath is clearly visible. Manual selection requires a high level of expertise. To enable the use of ONSD in low resource settings by medical personnel of all levels, we present a sparsely supervised AI framework that scores each frame of an ONSD ultrasound video and selects the optimal frame for ONSD assessment. Per-frame embeddings from an ultrasound foundation model (USFM) are passed to a lightweight temporal head and trained using Gaussian soft labels, which assign graded targets around labeled key frames, rather than hard binary (0/1) per-frame targets. We evaluate the model performance using subject-level cross-validation on 18 subjects and 323 ultrasound videos spanning nine acquisition sweep types. Top-k frame selection is used as a metric for direct comparison against the baselines. While only requiring sparse labels, our method can identify an optimal frame in 82.2% of the evaluated sweeps, exceeding the strongest training-free baseline (49.2%) and hard-label classification (71.9%).