Magneto-acousto-electrical tomography (MAET) is a hybrid imaging technique that combines the ultrasound transmission with electrical detection to capture the conductivity information of soft tissue. However, most previous studies have focused mainly on reconstructing conductivity boundary information. To address this limitation, we propose a novel method termed Quantitative Magneto-Acousto-Electric Computed Tomography (qMAE-CT), which aims to achieve quantitative mapping of the full conductivity distribution. The proposed method delineates the forward process into three sequential steps, each addressed by a dedicated reconstruction strategy: model-based deconvolution, back-projection, and conductivity reconstruction using a Physics-Informed Neural Network (PINN). The proposed method has been rigorously validated through numerical simulations, phantom experiments, and in vitro tissue studies. The results show that qMAE-CT can accurately reconstruct the conductivity distribution of irregular shaped targets, with a SSIM of 0.815, and the spatial resolution higher than 4 mm.

OBJECTIVES:The primary objective of this review is to summarize research cases and recent advances in ultrasound technology for overcoming the blood-brain barrier (BBB) and facilitating drug delivery to the brain, thereby providing insights to promote research on enhancing drug permeation across the BBB. KEY FINDINGS:This review summarizes recent advances in focused ultrasound combined with microbubbles to enhance BBB permeability for brain-targeted drug delivery. It covers the mechanisms by which ultrasound and microbubbles open the BBB, the factors influencing this process, and their applications in brain diseases. Finally, existing challenges in the field are highlighted, and the prospects for clinical translation are discussed. SUMMARY:Ultrasound combined with microbubble technology can safely, non-invasively, and reversibly open the BBB and deliver drugs, offering novel strategies and methods for the treatment of brain diseases.

The development and progression of gliomas are intimately linked to abnormal tumor angiogenesis. Different grades of gliomas exhibit distinct vascular distribution and structural characteristics. Therefore, analyzing vascular architecture and hemodynamics can provide crucial insights into tumor growth and invasion. Ultrasound localization microscopy (ULM) was applied to image the cerebrovascular changes in an orthotopic glioma mouse model at two different stages of tumor progression (14- and 21-days post-implantation). A total of twenty mice with intracranial glioma were imaged on the 14th day (n = 9) and on the 21st day (n = 11) after implantation. Pathological staining, vascular endothelial immunofluorescence examinations, and fluorescence dye perfusion were conducted to compare the ULM imaging results of the microvascular network morphology. By separating slow and fast blood flow signals, the ULM images revealed distinct vascular changes that visually correlated with H E-derived pathological features. The vascular densities in glioma regions showed a significant increase compared to the contralateral side: 30.4

Within the brachial plexus, the C6&C7 nerves play crucial roles in peripheral nerve block. The precise identification of nerve locations ultrasound images is critical for drug injections in anesthesia. However, their segmentation remains challenging due to limited annotated datasets, small target size in noisy images, and underutilized anatomical context. In this paper, we proposed LA-BPNet for the segmentation of the brachial plexus C6&C7 nerves in ultrasound images, which is able to use the anatomy information as the indicator of their segmentation task. We first introduce the Hybrid Spatial Perception (HSP) Module to solve the small target (nerves) segmentation problem in the low signal-to-noise ultrasound images, which consists of an Axial Spatial Attention (ASA) module and a Multi-scale Asymmetric Local Spatial Attention (MA-LSA) Module. To use the anatomical priors of nerves in the segmentation task adequately, we present the Anatomical Information Guidance (AIG) Module consisting of two parallel task branches to supervise the nerve centers and the coarse segmentation, additionally, the Interactive Guidance Mechanism (IGM) is integrated the information interaction in these two branches. In this paper, a private dataset is constructed, which contains 863 ultrasound images of the brachial plexus C6&C7 nerves. Extensive experiments with state-of-the-art (SOTA) models on our private dataset demonstrates that the LA-BPNet achieves better performance. The code will be available at https://github.com/zhankn/LA-BPNet.

Affective brain-computer interfaces (aBCIs) are increasingly recognized for their potential in monitoring and interpreting emotional states through electroencephalography (EEG) signals. Current EEG-based emotion recognition methods perform well with short segments of EEG data (known as segment-based emotion analysis). However, these methods encounter significant challenges in real-life scenarios where emotional states evolve over extended periods. To address this issue, we propose a novel Dual Attentive (DuA) transformer framework for long-term continuous EEG emotion analysis. Unlike segment-based approaches, our model processes an entire EEG trial as a whole, identifying emotions at the trial level, referred to as trial-based emotion analysis. This framework is designed to adapt to varying signal lengths, providing a substantial advantage over traditional methods. The DuA transformer incorporates three key modules: the spatial-spectral network module, the temporal network module, and the transfer learning module. The spatial-spectral network module simultaneously captures spatial and spectral information from EEG signals, while the temporal network module detects temporal dependencies within long-term EEG data. The transfer learning module enhances the model's adaptability across different subjects and conditions. To verify the effectiveness of the proposed DuA transformer, we construct a long-term continuous EEG emotion database and extensively evaluate our model using the self-constructed database along with two benchmark EEG emotion databases. On the basis of the trial-based leave-one-subject-out cross-subject cross-validation protocol, our experimental results demonstrate that DuA significantly outperforms existing methods in long-term continuous EEG emotion analysis, with an average improvement of 2.8%. The DuA transformer's ability to adapt to varying signal lengths and its superior performance across diverse subjects and conditions highlight its potential for real-world applications, enhancing the overall user experience and efficacy of aBCI systems.

Exosomes have emerged as critical mediators of intercellular and inter-organ communication in bone biology. Secreted by bone-resident cells such as osteoblasts, osteoclasts, osteocytes, and mesenchymal stem cells (MSCs), these nanosized vesicles carry diverse molecular cargos that regulate bone remodeling, regeneration, and skeletal homeostasis. In addition to mediating local communication within the bone microenvironment, exosomes also participate in systemic crosstalk communication between bone and other tissues, including skeletal muscle, adipose tissue, gut microbiota, the immune system, the nervous system, and vasculature. Disruption of these exosome-mediated pathways contributes to the development and progression of bone diseases, including osteoporosis, osteoarthritis, osteonecrosis of the femoral head, and bone metastases. This review summarizes current advances in exosome-mediated signaling in both physiological and pathological contexts, with particular emphasis on their roles as biomarkers, therapeutic agents, and drug delivery vehicles. We also discuss the emerging contribution of artificial intelligence (AI) to exosome research, especially in biomarker discovery, disease classification, and target identification, as well as the major challenges that currently limit clinical translation. Together, these insights highlight the potential of exosome-based strategies for precision medicine in bone diseases.

Despite years of methodological progress, how far AI has come in liver fibrosis staging has never been systematically evaluated under the heterogeneous, multi-center conditions that define clinical practice. To address this gap, we introduce LiFS, a large-scale dataset and benchmark derived from the MICCAI 2025 CARE-Liver challenge, comprising 610 patients across multiple centers and scanners with multi-sequence MRI. To the best of our knowledge, LiFS is the first benchmark providing complete gadoxetic acid-enhanced sequences with histopathology-confirmed annotations from diverse real-world scanners. Through systematic evaluation of 9 independently developed methods selected from 96 registered teams against in-cohort radiologist reference results, our findings address how far current AI has progressed toward clinical-level liver fibrosis staging from three complementary perspectives. First, against radiologists, the best AI methods were broadly comparable to the senior radiologist and significantly exceeded the junior radiologist in selected settings, while median AI performance generally approached junior-radiologist levels. Second, from a data perspective, cross-center heterogeneity, label imbalance, and contrast-enhanced sequence variability emerge as the dominant challenges for AI methods. Third, from a technical perspective, methodological design choices, including spatial registration, input dimensionality, multi-modal fusion strategy, and backbone architecture, appear to modulate cross-center robustness, although no single choice alone closes the gap. Overall, LiFS provides a rigorous real-world benchmark for positioning the current state of AI in liver fibrosis staging and for enabling future research on the key challenges that limit clinically reliable deployment.

To develop and validate a machine learning model integrating ultrasound radiomics and clinicopathological parameters to predict intrahepatic recurrence in colorectal cancer liver metastases (CRLM) patients after curative hepatectomy. This retrospective study enrolled 278 eligible CRLM patients (age, 55 ± 12 years; male, 188) from two centers, including a main cohort (n = 224, July 2010–February 2021) and an external cohort (n = 54, February 2015–October 2020). Patients were stratified by recurrence status during a 2-year follow-up. Preoperative ultrasound images and clinicopathological parameters were collected. Radiomics features were extracted from liver metastases, peri-tumor areas, and disease-free liver parenchyma. Using least absolute shrinkage and selection operator (LASSO) analysis and support vector machine algorithms, three predictive models were developed: clinical, radiomics, and clinical-radiomics combined (cRadiomics) models. Model performance was assessed using five-fold cross-validation (main cohort) and external validation (external cohort), with metrics including receiver operating characteristic (ROC) curve, the area under the ROC curve (AUC), accuracy, sensitivity, and specificity. Six clinical parameters (pathological lymph node positivity, synchronous liver metastases, bilobar liver metastases, preoperative chemotherapy, use of targeted drugs, and preoperative CA19-9 > 200 U/mL) and seven radiomics features were identified as strong predictors. The cRadiomics model achieved AUC values of 0.811 (95

Quantitative viscoelasticity imaging via shear wave elastography (SWE) remains challenging due to complex wave physics and limitations of conventional reconstruction methods. To address this, we present SW-VEI-Net, a physics-informed neural network (PINN) that simultaneously reconstructs the shear elastic modulus and viscous modulus by integrating viscoelastic wave equations into a dual-network architecture. The framework employs a dual-loss function to balance data fidelity and physics-based regularization, significantly reducing reliance on empirical data while improving interpretability. Extensive validation on tissue-mimicking phantoms, rat liver fibrosis model, and clinical cases demonstrates that SW-VEI-Net outperforms state-of-the-art SWE methods. Compared to SWENet (a PINN-based method using a linear elastic model), SW-VEI-Net not only enables simultaneous assessment of shear elastic and viscous moduli, but also achieves higher accuracy in shear elastic modulus reconstruction. Furthermore, when benchmarked against the dispersion fitting (DF) method (based on a viscoelastic model), SW-VEI-Net produces comparable viscoelastic parameter maps while exhibiting enhanced robustness and consistency. For liver fibrosis staging, SW-VEI-Net achieves AUC values of 0.85 ($\geq$F2) and 0.91 ($=$F4) based on elastic modulus classification, surpassing both SWENet (0.84, 0.85) and DF (0.78, 0.88). Additional validation in healthy volunteers shows strong agreement with a commercial ultrasound system. By synergizing deep learning with fundamental wave physics, this study represents a significant advancement in SWE, offering substantial clinical potential for early detection of hepatic fibrosis and malignant lesions through precise viscoelastic biomarker mapping.

Bubble size and morphological attributes are closely tied to flotation performance indicators and are key metrics for assessing the operational performance of froth flotation. However, most existing semantic segmentation methods adopt an encoder-decoder architecture, where repeated downsampling and coarse multiscale fusion cause excessive feature loss and hinder precise edge and detail segmentation. To address this issue, we propose a multistage fusion U-Net++ (MF-UNet++) framework. Based on multiscale feature extraction, the framework employs a two-stage fusion strategy combining an inter-level feature fusion (IFF) module and a dense feature fusion (DFF-UNet++) module, which enhances multiscale representations while preserving local details. Moreover, we further introduce a multiscale Convolutional Block Attention Module (MS-CBAM), which captures multiscale contextual information and more effectively focuses on salient features. On the validation set, MFUNet++ achieved an mIoU of 88.0 %, improving by 6.6 % over U-Net++. The errors in bubble count and bubble size were 5.98 % and 4.65 %, respectively. On the test set, the average inference time per image was 60.99 ms.

Acute extremity compartment syndrome (AECS) is a severe surgical emergency with no effective non-invasive therapies available. Here, we demonstrate that low-intensity pulsed ultrasound (LIPUS) attenuates muscle damage by suppressing oxidative stress and ferroptosis. In skeletal muscle cells following hypoxia-reoxygenation (H/R), LIPUS reduced reactive oxygen species (ROS), malondialdehyde, lipid peroxides, and ferrous iron accumulation. In a rat AECS model, LIPUS (1 MHz, 0.5 W/cm2, 20% duty cycle) mitigated histopathological injury and iron deposition compared with untreated controls, without affecting healthy muscle. Mechanistically, LIPUS activated the Nrf2/Gpx4 antioxidant pathway and downregulated p53 expression. These findings identify LIPUS as a non-invasive intervention that inhibits ferroptosis and improves tissue outcomes in AECS, suggesting its potential as a bedside therapeutic strategy to complement surgical management.

The Segment Anything Model 2 (SAM2) demonstrates remarkable universal segmentation capabilities on natural images. However, its performance on ultrasound images is significantly degraded due to domain disparities. This limitation raises two critical challenges: how to efficiently adapt SAM2 to ultrasound imaging while maintaining parameter efficiency, and how to deploy the adapted model effectively in resource-constrained clinical environments. To address these issues, we propose UniUltra for universal ultrasound segmentation. Specifically, we first introduce a novel context-edge hybrid adapter (CH-Adapter) that enhances fine-grained perception across diverse ultrasound imaging modalities while achieving parameter-efficient fine-tuning. To further improve clinical applicability, we develop a deep-supervised knowledge distillation (DSKD) technique that transfers knowledge from the large image encoder of the fine-tuned SAM2 to a super lightweight encoder, substantially reducing computational requirements without compromising performance. Extensive experiments demonstrate that UniUltra outperforms state-of-the-arts with superior generalization capabilities. Notably, our framework achieves competitive performance using only 8.91

Our prognostic model and mobile application enable multi-time-point prognostic evaluation for patients with acute-on-chronic hepatitis B liver failure, thereby improving patient care and facilitating clinical decision-making for liver transplantation.

Joint learning on diverse ultrasonic source scenarios presents a challenge in preserving stable gen-eralization due to the combination of heterogeneity of different sources and the inconsistency of joint learning features. Previous joint learning studies, which are not source-resilient frameworks, may not preserve stable generalization when trained on diverse source scenarios. Furthermore, the limited variations insingle-source data and the interference from ultrasound imaging, which are common in ultrasonic source scenarios, further decrease generalization. To address these problems, we pro posed a source-resilient joint learning framework consisting of three stages: 1) Source transforming, where our 1-to-N transformation unifies diverse source scenarios for source-resiliency. 2) Our feature enhancement modules model the source-resilient joint learning network, including a manifold-constraint normalization module (MCNM) for addressing heterogeneity by minimizing manifold-based loss, a task-consistent attention module (TCAM) shares the multi-scale features with self-attention to address inconsistency, and an adaptive feature-shifting module (AFSM) for feature-level augmentation to overcome single-source data.3) Our ultrasound-hybrid linear mapping (USmapping) cascades speckle randomization and mask-guiding Monge-Kantorovitch linear mapping to achieve ultrasonic style randomization for addressing the interference of ultrasonic data. Our framework was evaluated on eight ultrasound datasets from various scanners at multiple center sand surpassed previous comparable studies in both segmentation (DSCWAvg of 75.7%) and classification (AUROCWAvg of 68.8%) tasks. Our framework has the potential to serve as a general framework for enhancing the performance of joint learning under diverse ultrasonic source scenarios.

Plane-wave (PW) imaging, which provides high temporal resolution, has gained a significant attention. However, it is constrained by inherent lack of focus, requiring support from beamforming methods, such as Coherent Plane-Wave Compounding (CPWC), which can diminish the temporal advantages. Therefore, it is crucial to develop an advanced approach capable of addressing the trade-off between quality and frame rate. This paper proposes, for the first time, a hybrid progressive cross-domain fusion CNN-Transformer network (Hy-PCF), which is designed to simultaneously utilize both channel data and single-angle PW image for image reconstruction. Further, we design a novel Sparse Cross-Attention Guided Information Perception (SCAIP) module and use the single-angle PW image as guidance to extract information from the channel data. The qualitative and quantitative experimental findings demonstrate that Hy-PCF not only outperforms DAS significantly but also achieves performance comparable to the target CPWC. Hy-PCF represents a promising hybrid-CNN-Transformer network for PW imaging, offering a novel solution to enhance image quality, a critical advancement for clinical practice.

Sonodynamic therapy (SDT), as a promising noninvasive therapeutic modality with superior penetration depth, receives tremendous attention. To date, the widely accepted mechanism for reactive oxygen species (ROS) generation in SDT involves acoustic cavitation-triggered sonoluminescence (SL), followed by the SL-activation of sonosensitizers. However, current research on sonosensitizer development primarily focuses on promoting SL-to-ROS conversion, overlooking the essential role of the cavitation process. To fully unleash the potential of SDT, herein, a dual-enhanced strategy that harnesses the enhanced cavitation for SL generation and efficient SL-to-ROS conversion is developed for the first time to realize an all-around enhancement of SDT. Specifically, the proposed nano-sonosensitizer, namely MeTTh-PAE NPs, is released as hydrophobic aggregates with a rough surface in response to an acidic environment, allowing for highly enhanced cavitation-triggered SL under ultrasound. Meanwhile, as a typical aggregation-induced emission molecule, MeTTh demonstrates a highly promoted intersystem crossing process at its aggregated state, facilitating efficient SL-to-ROS conversion. Notably, combining these two fascinating attributes in MeTTh-PAE NPs results in an excellent sonodynamic antitumor effect in both in vitro and in vivo. This work proposes a novel strategy to fully exploit the potential of SDT and provides valuable insights for advancing the design of nano-sonosensitizers.

Objective: Current methods for assessing corneal mechanical properties are limited, particularly in their ability to provide localized information. This study proposes a novel approach based on the spectroscopic magnetomotive optical coherence elastography (MM-OCE) technique, aiming to enable non-invasive, localized evaluation of corneal mechanical properties. Methods: Magnetic nanoparticles (MNPs) were distributed on sample surfaces to induce vibrations via magnetic excitation. A spectral-domain OCT system combined with phase-sensitive OCT analysis tracked mechanical responses. Gelatin phantoms (varying stiffness) and ex vivo porcine corneas (untreated vs. crosslinked [CXL] regions) were tested. MB-mode validated MNP-induced vibrations, while M-mode scans and spectral analysis determined resonance frequencies. Histology assessed tissue integrity post-MNP application. Results: Gelatin resonance frequencies increased with concentration, confirming sensitivity to mechanical variations. In corneas, MM-OCE detected significant differences between untreated and CXL-treated regions: resonance frequencies rose from 74.48 ± 6.23 Hz (untreated) to 83.42 ± 4.97 Hz (1-min UV), 110.92 ± 2.40 Hz (3-min UV), and 121.23 ± 3.02 Hz (6-min UV). Histology confirmed no MNP-induced tissue damage. Conclusion: MM-OCE effectively differentiates localized biomechanical changes in corneal tissue, demonstrating feasibility for quantifying stiffness variations induced by CXL. Significance: Although further improvements are needed for potential clinical applications, this study demonstrated that MM-OCE may offer a promising, non-invasive method for assessing local mechanical properties of a cornea, with the potential to enhance early diagnosis, treatment planning, and monitoring in ophthalmology.

Objective: The aim of this study was to evaluate the consistency and reproducibility of attenuation coefficient (AC) measurements using different commercial ultrasound (US) across via a phantom experiment to investigate the relationship between the AC and MRI-derived proton density fat fraction (MRI-PDFF) values and the conversion equation. Methods: Twelve phantoms containing varying fat proportions (0–100%) were constructed. Phantom ACs were estimated via three US attenuation systems, including attenuation imaging (ATI), ultrasound attenuation analysis (USAT), and the US-guided attenuation parameter (UGAP), along with MRI-PDFF. Agreement among the AC values from the three ultrasonic attenuation systems was evaluated. Linear correlation analysis was used to explore the ACs, fat concentrations of the phantom, and MRI-PDFF measurements, from which a linear conversion formula between the ultrasonic attenuation parameters and the MRI-PDFF was derived. Results: MRI-PDFF and phantom fat concentration measurements appeared with a strong linear correlation (R2 = 0.996, p < 0.001). For the three US attenuation parameters, both inter-operator and intra-operator intraclass correlation coefficients (ICCs) ranged from 0.990 to 0.995 and 0.989 to 0.995, respectively. Bland–Altman analysis revealed no significant differences between the above three (all p > 0.05). Significant linear relationships were demonstrated between ultrasound attenuation parameters and phantom fat concentration (r = 0.938–0.986; all p < 0.001), as well as between ultrasound attenuation parameters and MRI-PDFF values (r = 0.922–0.982; all p < 0.001). A conversion formula (fat proportions ≤ 50%) was derived: US (dB/cm/MHz) = 0.501 + 0.012 MRI-PDFF (%). Conclusions: AC across different commercial ultrasound devices demonstrated significant diagnostic value in fat concentrations that appeared good consistency in measuring phantom fat concentration both between and within groups. The linear relationship between AC and MRI-PDFF enables the application of a conversion formula.