Background:Pathological risk stratification of prostate cancer (PCa) guides treatment decisions. Preoperative noninvasive assessment of PCa risk stratification holds promise for reducing unnecessary invasive biopsies. Elevated levels of iron and fat, along with metabolic disorders in PCa significantly correlate with tumor proliferation and aggressiveness, yet its predictive value in risk stratification remains unclear. We aimed to noninvasively measure fat content as well as iron deposition of PCa lesions by multiparametric magnetic resonance imaging (mpMRI) and investigate their effectiveness in predicting PCa risk. Methods:We prospectively collected patients suspected of PCa with preoperative MRI from 2019 to 2022, and ultimately included 109 pathologically confirmed PCa patients. The Gleason score (GS) and International Society of Urological Pathology grade group (ISUP GG) were determined by two uropathologists who evaluated independently and reached a consensus. Patients were stratified based on the ISUP GG, with 42 in the pathological low-risk (PL) group (ISUP GG ≤2; 69.9±6.08 years), 67 in the pathological high-risk (PH) group (ISUP GG ≥3; 71.82±5.86 years). We also collected clinical, pathologic, and imaging data from the patients. Based on the variables screened by Least Absolute Shrinkage and Selection Operator (LASSO) regression analysis, an improved fusion (IF) model was established and visualized with a nomogram plot. The conventional fusion (CF) model was constructed by removing the non-conventional image variables in the IF model. Model performance was evaluated using 10-fold cross-validation, receiver operating characteristic (ROC) analysis, DeLong test, and decision curve analysis (DCA). P<0.05 was considered statistically significant. Results:Significant differences were observed in the prostate-specific antigen (PSA), prostate volume (PV), Prostate Imaging Reporting and Data System (PI-RADS) scores, fat fraction (FF), T2*, and average apparent diffusion coefficient (ADC) values of the lesions between the two groups. These variables were selected to construct the IF model. The CF model was developed by removing FF and T2* values. The IF model demonstrated higher accuracy than the CF model [IF model: sensitivity =0.952, specificity =0.761, area under the curve (AUC) =0.920; CF model: sensitivity =0.762, specificity =0.761, AUC =0.819; DeLong test: P=0.002, <0.05]. Conclusions:mpMRI‑derived FF and T2* values were significantly associated with ISUP GG in PCa. Intergrating FF and T2* values with ADC, PI-RADS, PSA and PV may predict pathological risk classification more effectively.

This study aims to develop a 2.5D deep learning model with shape and margin as auxiliary tasks to improve the diagnostic performance of benign–malignant classification of breast lesions in automated breast ultrasound system (ABUS) images. In this retrospective study, 387 breast lesions (106 malignant and 281 benign) from 313 patients were enrolled from two centers between 2021 and 2024. Lesions from Center 1 (315 lesions, 85 malignant, 230 benign) were used as the training cohort, and lesions from Center 2 (72 lesions, 21 malignant, 51 benign) were used as the held-out external testing cohort, with no patient overlap between the two centers. A deep learning classification algorithm, combined with a ResMask Fusion module, was used to extract and integrate morphological features, particularly shape and margin, from two-dimensional ultrasound images. ABUS-ResMask-Net uses Swin Transformer V2-T as the backbone, with shape and margin classification as auxiliary tasks. In the final lesion-level evaluation, ABUS-ResMask-Net achieved an AUC of 0.91 (95

INTRODUCTION:This study aimed to establish and validate CT-based radiomics models combined with clinical data to identify Fungal Co-Infections (FCI) in immunocompromised patients with Viral Pneumonia (VP). MATERIALS AND METHODS:A total of 406 patients (VP: 283; FCI: 123) from two hospitals were retrospectively enrolled and divided into training (n = 218), testing (n = 96), and external validation (n = 92) cohorts. Radiomics features were extracted from chest CT images. Feature selection was performed using the Least Absolute Shrinkage And Selection Operator (LASSO), and logistic regression models were built with clinical, radiomics, and combined inputs. Model performance was assessed using the Area Under the Receiver Operating Characteristic Curve (AUC), calibration, and Decision Curve Analysis (DCA). RESULTS:The combined model achieved AUCs of 0.981 (95% CI: 0.959 - 0.992), 0.845 (95% CI: 0.762 - 0.950), and 0.835 (95% CI: 0.715 - 0.937) in the training, testing, and external validation cohorts, respectively, and consistently outperformed clinical-only and radiomics-only models. DISCUSSION:The model identified characteristic clinical and imaging differences between VP and FCI, including higher neutrophil counts, lower lymphocyte counts, and imaging markers such as reversed halo sign and solid nodules in FCI. These findings support the potential of radiomics as a noninvasive tool for early detection and risk stratification. CONCLUSION:CT-based radiomics provides an effective approach for differentiating VP and FCI in immunocompromised patients, with potential to improve diagnosis and clinical management.

BACKGROUND:Tinnitus affects 10-15% of adults globally, yet there are still no effective treatments for this major health condition. Repetitive transcranial magnetic stimulation (rTMS), a noninvasive neuromodulation technique, allows modulation of pathologically altered functional activities to promote symptom remission. However, its efficacy critically depends on the selection of stimulation targets, and substantial interindividual variability has been observed in clinical trials. Here, we aimed to identify potential target regions that are causally involved in alleviating distinct functional abnormalities using the digital twin brain (DTB). METHODS:A cohort of 89 participants was used to characterize whole-brain neural activity patterns. Multimodal neuroimaging data were used to develop the tinnitus-specific DTB and to generate causal response maps based on more than 1.64 million virtual stimulations. Whole-brain gene expression data were further integrated to examine the neurobiological plausibility of the DTB-derived causal response maps. Finally, we validated the predictive capacity of such response maps using an independent rTMS dataset. RESULTS:We identified two aberrant brain states that emerged sequentially with disease progression, predominantly overlapping with the somatomotor and default mode networks, respectively. DTB-derived causal response maps revealed that the modulation of sensory and cognitive states requires stimulation of distinct, functionally specialized regions. Specifically, parieto-occipital regions play a crucial role in sensory modulation, while the dorsolateral prefrontal cortex exerts a causal influence on cognitive modulation. Moreover, these causal response maps correlate with the expression of tinnitus risk genes. By incorporating individual connectivity profiles of target regions, DTB-derived causal response maps accurately predicted rTMS effects on both sensory state (r > 0.85, Ppermutation < 0.01) and cognitive state (r > 0.78, Ppermutation < 0.05). Particularly, the predictive capacity exhibited a state-specific nature. CONCLUSIONS:This work suggests that brain functional alterations in tinnitus evolve with disease progression, and DTB has the potential to predict rTMS effects on distinct brain states, thereby informing more precise and targeted noninvasive brain stimulation interventions for tinnitus. TRIAL REGISTRATION:Trial registered with https://www.chictr.org.cn/indexEN.html , Explore the mechanism of repetitive transcranial magnetic stimulation intervention in tinnitus based on multi-modal functional magnetic resonance imaging (ChiCTR2100047989), Submitted June 2021, First Patient Enrolled July 2021.

Hepatocellular carcinoma (HCC) shows marked spatial heterogeneity, limiting biopsy-based Edmondson-Steiner (ES) grading. We developed a multicenter radiogenomic framework to noninvasively predict ES grade and explore underlying molecular mechanisms. Arterial-phase DCE-MRI from 295 patients and The Cancer Imaging Archive (TCIA) cases were analyzed using three tumor regions (body, edge, and out). An integrated volume-of-interest (VOI) random forest (RF) model was trained with selected features and externally validated. Radiogenomic analysis correlated radscore with TCIA transcriptomic profiles using weighted gene co-expression network analysis (WGCNA). The model achieved high discrimination (area under the curve [AUC] 0.959 internally; 0.860 externally). Radscore-associated modules revealed ribosomal dysregulation and immune exhaustion. A derived prognostic signature stratified and The Cancer Genome Atlas (TCGA) patients into distinct risk groups and independently predicted survival (hazard ratio [HR] 3.95, p < 0.0001; C index 0.643). This integrated radiogenomic approach enables noninvasive ES grading and provides insight into biologically relevant tumor heterogeneity.

INTRODUCTION:Diffusion- and perfusion-based imaging, including Apparent Diffusion Coefficient (ADC) and CT Perfusion (CTP), are standard tools for evaluating ischemic stroke but primarily reflect structural and hemodynamic changes. They provide limited insight into tissue metabolism. Amide Proton Transfer-weighted (APTw) imaging enables noninvasive assessment of pH-related metabolic alterations and may offer complementary diagnostic and prognostic information. MATERIALS AND METHODS:In this prospective study, 54 patients with hyperacute or acute ischemic stroke (mean age 64.39 ± 10.91 years; 41 males) were enrolled. Correlations between APTw and conventional imaging parameters, including ADC, Cerebral Blood Volume (CBV), Cerebral Blood Flow (CBF), and Time to Peak (TTP), were analyzed. Kaplan-Meier analysis assessed 18-month outcomes. Time-dependent Receiver Operating Characteristic (ROC) analysis, decision curve analysis (DCA), calibration curves, and Clinical Impact Curves (CIC) were used to evaluate prognostic performance. RESULTS:APTw values were significantly correlated with CBV (r = 0.6886), CBF (r = 0.6702), ADC (r = 0.6565), and TTP (r = -0.6519) (all p < 0.05). Kaplan-Meier analysis demonstrated significant differences in 18-month outcomes between APTw-based subgroups (log-rank p < 0.05). APTw achieved higher AUCs at 6, 12, and 18 months (0.821, 0.831, and 0.877) than ADC and CTP parameters. DCA, calibration, and CIC analyses confirmed stable predictive performance. DISCUSSION:APTw showed strong concordance with diffusion- and perfusion-based biomarkers while providing complementary metabolic information, suggesting exploratory comparative prognostic relevance in ischemic stroke. Conclusions APTw imaging offers complementary metabolic insight and may enhance diagnostic and prognostic evaluation in ischemic stroke.

Background:Magnetic resonance imaging (MRI) provides excellent soft-tissue contrast and enables multi-parametric assessment of tumor biology. Longitudinal relaxation time (T1) mapping has emerged as a quantitative method capable of measuring the intrinsic T1 value of tissues, reflecting microscopic structural and compositional changes in the tumor microenvironment. This study aimed to evaluate the utility of magnetic resonance T1 mapping, alone and in combination with diffusion-weighted imaging (DWI) and dynamic contrast-enhanced MRI (DCE-MRI), in differentiating histologic subtypes and assessing tumor differentiation in non-small cell lung cancer (NSCLC). Methods:A total of 76 patients with pathologically confirmed NSCLC [48 adenocarcinoma (AD), 28 squamous cell carcinoma (SCC)] were prospectively enrolled. Patients were further stratified into poorly differentiated (n=32) and moderately/highly differentiated (n=44) groups. All underwent conventional MRI, DWI, DCE-MRI, and native/post-contrast T1 mapping. Quantitative parameters included apparent diffusion coefficient (ADC), Ktrans, Kep, Ve, T1pre, T1post, absolute T1 reduction (T1d), and percentage T1 reduction (T1d%). For parameters showing statistically significant differences between groups, receiver operating characteristic (ROC) curve analysis was performed to evaluate diagnostic performance. The area under the curve (AUC), optimal cutoff values, sensitivity, specificity, and Youden index were calculated. Results:The agreement between the two readers was reasonably good with intraclass coefficient (ICC) values of 0.938 for T1pre, 0.922 for T1post, and 0.814 for ADC. AD demonstrated significantly higher ADC values (1,159.01 vs. 1,041.75)×10-6 mm2/s and lower T1pre (1,440 vs. 1,576.83) ms, T1post (549.07 vs. 607.44) ms, and T1d (890.93 vs. 969.39) ms values compared with SCC (P<0.05). The four-parameter model (ADC + T1pre + T1post + T1d) achieved the highest performance for differentiating AD from SCC (AUC =0.805, with 75% sensitivity and 79.2% specificity). Poorly differentiated tumors showed significantly lower ADC (985.69 vs. 1,210.44)×10-6 mm2/s and higher T1pre (1,553.4 vs. 1,444.61) ms values than moderately/highly differentiated tumors (P<0.05), with the combination of ADC + T1pre yielding the best diagnostic accuracy (AUC =0.866, with 77.3% sensitivity and 84.4% specificity). No DCE parameters showed significant differences between groups (All P>0.05). Conclusions:Multi-parametric MRI centered on T1 mapping, particularly when combined with ADC, provides a reproducible and non-invasive tool for subtyping and grading NSCLC, underscoring its potential as a clinically useful imaging biomarker.

Some patients with myasthenia gravis (MG) present with cognitive impairment (CI), and dysfunction of the cerebral glymphatic system (GS) is considered a key contributor. We hypothesized that dysfunction of the GS contributes to cognitive impairment in MG and aimed to identify potential neuroimaging biomarkers. This study included 41 patients with MG (29 patients without CI, 12 patients with CI) and 35 healthy controls (HC). GS function was assessed by calculating the diffusion tensor image analysis along the perivascular space (DTI-ALPS) and gBOLD-CSF coupling strength. Concurrently, neuropsychological testing (including MMSE, MoCA, CDT, etc.) was performed to examine the relationship between GS function and cognitive performance. Single factor analysis of variance and Kruskal-Wallis test were used for intergroup comparison. Spearman correlation analysis was used to observe the relationship between DTI-ALPS and gBOLD-CSF coupling and clinical scales. Compared with HC, patients with MG exhibited significantly reduced DTI-ALPS indices and gBOLD-CSF coupling strength (p < 0.05). Additionally, MG patients with CI demonstrated a decrease in DTI-ALPS index compared with MG patients without CI (p < 0.05). Furthermore, DTI-ALPS index demonstrated significant correlations with overall cognitive function in all participants. Patients with MG demonstrate impaired GS function, which correlates with their overall cognitive performance. These findings suggest that GS dysfunction may represent a characteristic feature of cognitive decline in MG.

The novel titanium implant coating based on the principle of immune osteogenesis can better mimic the natural post-implantation biological processes within the body. However, most current studies are limited to solely inducing M2 polarization of macrophages to promote bone regeneration, neglecting the critical role of M1 macrophages during the early implantation phase. This approach does not fully align with the natural macrophage polarization pattern observed in the human body. Building upon prior research, we designed a magnetoelectric titanium dioxide nanotube array coating to achieve programmed regulation of macrophage polarization. This coating employs hydrogenation to create high-density oxygen vacancies, significantly enhancing surface conductivity. In vitro experiments demonstrate that currents induced by magnetic fields promote M1 polarization of macrophages, effectively recruiting bone marrow-derived mesenchymal stem cells (BMMSCs)—a critical process during early implantation. Conversely, under non-magnetic conditions, the coating sustainably induces M2 polarization and enhances osteogenic differentiation of BMMSCs. In vivo experiments further confirmed that this coating significantly enhances integration between titanium implants and surrounding bone tissue. Thus, this smart coating, combining magnetoelectric properties with nanostructures, successfully mimics the natural immune-osteogenic coupling process in vivo, offering a promising new strategy for enhancing the long-term stability and clinical efficacy of implants.

Bone and bone-related disorders remain difficult to diagnose and treat because of the dense mineralized matrix, limited vascular perfusion, complex pathological microenvironment, and insufficient site-specific accumulation of conventional agents. Liposomes, as biocompatible vesicular nanocarriers capable of loading therapeutic, genetic, and imaging cargos, have emerged as versatile platforms for bone-oriented theranostics. In contrast to previous reviews mainly focusing on general liposomal drug delivery or individual skeletal diseases, this review provides an integrated perspective on liposomal multimodal theranostics for bone disorders, emphasizing responsive delivery, diagnostic imaging, biomimetic engineering, and clinical translation. We first summarize the structural features, classification, and functional modification of liposomes, including bone-targeting ligands, stimuli-responsive release, membrane-penetrating delivery, and biomimetic designs. We then discuss their diagnostic and therapeutic applications in osteomyelitis, joint tuberculosis, rheumatoid arthritis, osteoarthritis, osteoporosis, bone defects, fracture nonunion, bone tumors, and spinal cord injury, with particular attention to radiolabeled liposomes, fluorescent probes, imaging-guided therapy, and microenvironment-responsive systems. Moreover, biomimetic liposomes, such as cell membrane-camouflaged, macrophage- or erythrocyte membrane-engineered, membrane-fusion-inspired, and organelle-delivering liposomal systems, are highlighted for their roles in immune evasion, lesion targeting, barrier penetration, and intercellular delivery. Finally, we discuss key translational challenges, including manufacturing reproducibility, stability, biosafety, immune responses, disease-model relevance, and clinical evaluation. This review aims to provide a rational framework for developing responsive and biomimetic liposomal theranostic systems for bone-related diseases.

Diabetic foot ulcer (DFU) is a clinically challenging complication characterized by poor healing outcomes, and conventional therapies provide limited benefit. Mesenchymal stem cell (MSC) transplantation offers a promising strategy for DFU repair. However, the low survival of transplanted MSCs in the hostile wound microenvironment, coupled with the lack of real-time, non-invasive methods to track these cells in vivo, severely hampers their therapeutic efficacy and clinical translation. We engineered MSCs to co-express a dual reporter system comprising near-infrared fluorescent protein (iRFP) and ferritin heavy chain (FTH1). These modified cells were then integrated with a fibrin glue (FG) scaffold to create a unified platform that supports both multimodal imaging and therapeutic function within skin wounds. First, FTH1 overexpression enhances the antioxidant capacity of MSCs, while the FG scaffold provides structural support; this combination enhances cell survival and retention. Second, the iRFP/FTH1 dual reporter enables near-infrared fluorescence imaging and MRI-based localization, establishing a multimodal platform for real-time cell tracking. In a full-thickness skin defect model in diabetic mice, multimodal imaging revealed that transplanted cells persisted in the wound area for approximately seven days. Treatment with iRFP/FTH1-MSCs/FG significantly accelerated wound closure and promoted hair follicle regeneration and angiogenesis. Additionally, local iron deposition resulting from FTH1 expression enhanced fibroblast migration and collagen synthesis, further facilitating extracellular matrix remodeling. Mechanistic studies demonstrated that this therapy drives macrophage polarization toward the anti-inflammatory M2 phenotype and activates the PI3K–AKT–VEGF signaling pathway. These complementary effects synergistically enhance tissue regeneration and systematically improve diabetic wound healing. Collectively, this multimodal stem cell–scaffold system effectively integrates dynamic cell tracking with stem cell therapy during skin wound repair. It addresses a critical technical gap in visualizing stem cells within the wound microenvironment and provides valuable methodological and theoretical foundations for optimizing regenerative strategies for diabetic skin wounds.

To address the technical bottlenecks in traditional mineral sorting processes, this paper proposes a SegNet-based industrial visual inspection method for molybdenum ore integrated with a multi-scale attention mechanism. Targeted optimizations are achieved through the construction of a multi-level feature enhancement mechanism: (1) Aiming at the classification confusion caused by the highly similar texture features between low-grade molybdenum ore and tailings, this study proposes integrating a Multi-Scale Adaptive Spatial Attention Gate (MASAG) to enhance the encoder's ability to capture the texture features specific to molybdenum ore; (2) To address the feature transmission inefficiency in low-contrast and high-density overlapping regions of X-ray images, this study integrates a Convolutional Additive Self-Attention (CAS) module across the encoder-decoder interface, aiming to refine the feature propagation pathway and enhance information interaction between these two components; (3) For the missed detection problem caused by sparse features of small particles after molybdenum sulfide crushing, a Multi-Order Gated Aggregation Network (MogaNet) is integrated to adjust the decoder output, enhance local feature processing, and optimize the segmentation results; (4) To improve the segmentation accuracy of molybdenum ore targets, the hybrid loss of Binary Cross-Entropy and Dice (BCEDiceLoss) is adopted to refine the optimization direction during model training and enhance segmentation accuracy. Validation using a professional dataset of high-resolution molybdenum ore images constructed by X-ray equipment shows that the improved Segmentation Network(SegNet) model achieves comprehensive performance upgrades in detection and segmentation, demonstrating excellent robustness and generalization ability in complex ore scenarios. This delivers reliable technical backing to facilitate the intelligent transformation of mineral processing operations.

BACKGROUND:Percutaneous mesh-container-plasty (PMCP), a modified traditional percutaneous kyphoplasty (PKP) technique, is increasingly being used to treat osteoporotic vertebral compression fractures with up-endplate injury. This retrospective study aimed to compare the clinical and radiological results of PKP and PMCP for the treatment of this disease. METHODS:We retrospectively analyzed the medical records of patients with osteoporotic compression fractures and upper endplate injuries treated at our hospital between January 2019 and December 2021. A total of 192 patients who met the inclusion and exclusion criteria were enrolled. Of these, 103 underwent PKP and 89 underwent PMCP. Key outcome measures included surgical safety, clinical efficacy, and radiological results. RESULTS:Both the PKP and PMCP groups showed significant improvements in visual analog scale and Oswestry Disability Index scores postoperatively. Additionally, anterior vertebral body height ratio and Cobb's angle improved in both groups, though no statistically significant difference was observed between them. The hospital stay duration was similar between the 2 cohorts. Notably, the PMCP group required a larger volume of bone cement injection yet exhibited a significantly lower incidence of cement leakage and adjacent vertebral fractures (9/89 and 2/89, respectively) compared to the PKP group (24/103 and 11/103, respectively) (P < 0.05). Moreover, the PMCP group had shorter operation times (34.64 ± 9.88 minutes) and reduced fluoroscopy frequency (35.43 ± 5.46 instances) compared to the PKP group (27.23 ± 8.54 minutes and 23.87 ± 5.59 instances, respectively) (P < 0.05). CONCLUSIONS:PMCP provided superior clinical outcomes for the management of osteoporotic compression fractures with upper endplate injuries. It was associated with reduced operation and fluoroscopy times, as well as lower risks of adjacent vertebral fractures and cement leakage, compared to PKP.

In the diabetic milieu, fluctuations in blood glucose levels, elevated reactive oxygen species (ROS), and abnormal macrophage polarization exacerbate the imbalance of the osteoblast–osteoclast axis, posing significant challenges for the repair of critical‐sized bone defects. Multifunctional conductive biomaterials based on electrical stimulation (ES) therapy present a potential strategy to modulate the adverse inflammatory microenvironment and promote bone regeneration under diabetic conditions. However, traditional complex endogenous implantable battery devices are often bulky and difficult to seamlessly integrate with the body's natural biological processes. Herein, a novel implantable smart bio‐battery—comprising GelMA, tetrafluorophenylboronic acid (FPBA), osteostatin, and graphene oxide (GF‐Os G )—is developed for bone defect regeneration in diabetic inflammatory environments. GF‐Os G bio‐microbatteries can generate microcurrents in high‐glucose environments, reprogramming macrophages to the M2 phenotype and modulating immune responses. A favorable immune microenvironment is a crucial prerequisite for vascular regeneration and bone differentiation. ES can also directly stimulate osteogenic differentiation of bone marrow mesenchymal stem cells and synergistically modulate the osteoblast‐osteoclast axis with Osteostatin to promote bone regeneration. Furthermore, the underlying therapeutic mechanism is elucidated, demonstrating that GF‐Os G promotes osteogenesis via the ERK/P38‐GPX4 axis, effectively enhancing osteogenic differentiation. In vivo experiments revealed that the GF‐Os G can modulate immune responses and facilitate the repair of diabetic bone defects. This innovative approach combines immune regulation with a bio‐microbattery ES system, offering a novel material platform for microcurrent‐enhanced tissue regeneration in diabetic microenvironments.

Dense video captioning aims to locate multiple events from untrimmed videos and generate corresponding captions for each meaningful event. The application of multimodal information(e.g., video, audio) for dense video captioning has recently achieved great success. However, learning the information interactions between different modalities while achieving cross-modal feature alignment is highly challenging for an encoder. Recent studies of several multimodal tasks have shown that multimodal models benefit from shared and individual representations. Thus, in this paper, we propose a novel feature fusion module, which uses shared and individual modality representations to capture commonalities and complementary relationships between modalities. Moreover, the proposed model bridges the gap between shared modality representations, which helps to obtain deeper cross-modal associations for better feature interaction and alignment. Furthermore, to compensate for the limitation that different level proposal heads do not interact sufficiently during event detection, we propose a multilevel information interaction mechanism to dynamically adjust and fuse the information among different level proposal heads in the event detection module. Based on the ActivityNet Captions, subdatasets of ActivityNet Captions and YouCook2, we conducted comprehensive experiments to evaluate the performance of our proposed model. The experimental results show that our model achieves impressive performance compared with state-of-the-art methods.

Background:The extracellular volume fraction (fECV) based on equilibrium phase iodine density images (IDIs) of dual-layer spectral detector computed tomography (DLCT) can be used in the assessment of gastric cancer (GC). However, obtaining the equilibrium phase images requires a higher radiation dose. The purpose of our study was to evaluate the feasibility of low-dose equilibrium phase scans on DLCT for fECV acquisition in histological grading assessment of GC. Methods:A total of 86 gastric adenocarcinoma patients confirmed by surgical pathology were divided into two groups that underwent contrast-enhanced DLCT with routine-dose (120 kV/129 refmAs) and low-dose (120 kV/90 refmAs) equilibrium phases, respectively. The fECV values of GC lesions were measured from IDIs in the equilibrium phase. The radiation dose, image quality of the equilibrium phase images, and fECV values were compared between the low- and routine-dose groups. Then, the performance of the fECV in the two groups to distinguish histological grades of GC lesions was evaluated using a receiver operating characteristic (ROC) curve and the DeLong test. The fECV maps were reconstructed from the IDIs of the equilibrium phase. Results:The radiation dose of the equilibrium phase and the accumulated dose in the low-dose group decreased by 54% and 34%, respectively, compared to the routine-dose group (both P<0.001). The image noise of equilibrium phase images was higher in the low-dose group than that in the routine-dose group (P<0.001) and the noise scores of the low-dose group were lower than those of the routine-dose group (P=0.003), whereas no significant differences were detected in the signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), detail score, and fECV values between the two groups (P=0.243, 0.607, 0.861, and 0.301, respectively). The fECV values of high-grade GC lesions were higher than those of the low-grade lesions in the two groups (52.98%±8.06% vs. 38.31%±5.24%, P<0.001, and 51.94%±9.11% vs. 36.91%±5.26%, P=0.002). The fECV obtained in the low-dose group had a similar performance compared to the routine-dose group in histological grading assessment of GC [area under the curve (AUC): 0.871 vs. 0.879, Z=-0.148, P=0.882]. Conclusions:Contrast-enhanced DLCT with low-dose equilibrium phase scans in GC reduced the radiation dose while providing comparable image quality and performance of fECV in histological grading assessment to those of routine-dose scans.

Predicting repetitive transcranial magnetic stimulation (rTMS) effects on whole-brain dynamics in clinical populations is crucial for developing personalized therapies and advancing precision medicine in brain disorders. This study provides the first proof-of-concept demonstrating that the Digital Twin Brain (DTB) can forecast rTMS effects on brain state dynamics in individuals with brain disorders (chronic tinnitus). First, we identified two aberrant brain states that predominantly overlapped with the somatomotor and default mode networks, respectively. Subsequently, we developed DTB for patients and derived regional responses for each brain region, revealing distinct roles of the parieto-occipital and frontal regions. Mechanistically, we examined the biological plausibility using tinnitus-specific risk genes and investigated the multi-scale neurobiological relevance. Clinically, we found that DTB can predict rTMS effects in an independent, longitudinal dataset (all r > 0.78). Particularly, the predictive capacity exhibits a state-specific nature. Overall, this work proposes a novel DTB-based framework for predicting rTMS effects in clinical populations and provides the first empirical evidence supporting its clinical utility. This approach may be generalizable to other brain disorders and neuromodulation techniques, promoting broader advancements in brain health. ### Competing Interest Statement The authors have declared no competing interest. STI2030-Major ProjectsSTI2030-Major Projects, , 2021ZD0200201

To develop and validate a machine learning-based prediction model to predict axillary lymph node (ALN) metastasis in triple negative breast cancer (TNBC) patients using magnetic resonance imaging (MRI) and clinical characteristics. This retrospective study included TNBC patients from the First Affiliated Hospital of Soochow University and Jiangsu Province Hospital (2016-2023). We analyzed clinical characteristics and radiomic features from T2-weighted MRI. Using LASSO regression for feature selection, we applied Logistic Regression (LR), Random Forest (RF), and Support Vector Machine (SVM) to build prediction models. A total of 163 patients, with a median age of 53 years (range: 24-73), were divided into a training group (n = 115) and a validation group (n = 48). Among them, 54 (33.13%) had ALN metastasis, and 109 (66.87%) were non-metastasis. Nottingham grade (P = 0.005), tumor size (P = 0.016) were significant difference between non-metastasis cases and metastasis cases. In the validation set, the LR-based combined model achieved the highest AUC (0.828, 95%CI: 0.706-0.950) with excellent sensitivity (0.813) and accuracy (0.812). Although the RF-based model had the highest AUC in the training set and the highest specificity (0.906) in the validation set, its performance was less consistent compared to the LR model. MRI-T2WI radiomic features predict ALN metastasis in TNBC, with integration into clinical models enhancing preoperative predictions and personalizing management.

Glypican-3 (GPC3) is frequently overexpressed in hepatocellular carcinoma (HCC) and plays a key role in immune and metabolic remodeling of the tumor microenvironment. Reliable noninvasive biomarkers for predicting GPC3 status could improve patient stratification and support precision immunotherapy. This multicenter retrospective study included 274 patients with pathologically confirmed hepatocellular carcinoma from three institutions, 34 external cases with MRI from The Cancer Imaging Archive, and 363 transcriptomic profiles from The Cancer Genome Atlas. Contrast-enhanced T1-weighted imaging and diffusion-weighted imaging were analyzed. Tumor and peritumoral regions were segmented manually and radiomic features extracted using PyRadiomics. Feature selection was performed with correlation filtering and least absolute shrinkage and selection operator regression. Machine learning classifiers including logistic regression, random forest, support vector machine, k-nearest neighbor, and decision tree were trained with 10-fold cross-validation and tested on independent external cohorts. A radiomics score was calculated for each patient. Radiogenomic analysis correlated radiomics scores with transcriptomic data using weighted gene co-expression network analysis. Hub genes and enriched pathways were identified, and immune infiltration and predicted immunotherapy response were assessed using computational methods. The random forest model using contrast-enhanced T1-weighted imaging achieved an area under the curve of 0.966 in training and 0.935 in internal validation. The integrated contrast-enhanced T1-weighted imaging plus diffusion-weighted imaging model reached an internal validation area under the curve of 0.979. In external testing, the best performance was obtained with a support vector machine model (area under the curve 0.756). Radiomics scores were significantly correlated with GPC3 expression (R = 0.78, p < 0.05). Transcriptomic analysis identified a 10-gene signature enriched in hypoxia and lipid metabolism pathways that stratified patients into prognostic subgroups (concordance index 0.720, hazard ratio 4.07, p < 0.0001). High-risk patients had greater immune infiltration and a lower predicted immune evasion score, suggesting a potential benefit from immunotherapy. MRI-based radiomics models can noninvasively predict GPC3 expression in hepatocellular carcinoma. Radiomics scores reflect underlying hypoxia and lipid metabolism pathways and stratify patients by prognosis and predicted immunotherapy response. These findings support radiogenomics as a translational approach to imaging-guided precision treatment in hepatocellular carcinoma.

BackgroundTinnitus persists as a significant public health challenge with elusive neurochemical underpinnings. Emerging evidence implicates dysregulated excitatory-inhibitory neurotransmission in the anterior cingulate cortex (ACC), a limbic-auditory hub governing tinnitus salience. This study investigates dynamic ACC neurochemical changes during tinnitus progression.MethodsUsing single-voxel magnetic resonance spectroscopy (MRS), GABA+/creatine (Cr) and Glx (glutamate+glutamine)/Cr ratios were measured in the ACC of 16 recent-onset (RO; <6 months), 22 chronic (CH; ≥6 months) tinnitus patients, and 26 healthy controls (HC). Tinnitus severity was assessed via tinnitometry and Tinnitus Functional Index (TFI).ResultsRO patients exhibited significantly reduced ACC GABA+/Cr compared to CH and HC groups (p < 0.05), while CH and HC showed no differences. GABA+/Cr positively correlated with tinnitus duration across patients (r = 0.364, p = 0.025). Although Glx/Cr did not differ between groups, elevated Glx/Cr associated with higher tinnitus pitch-matching frequencies (r = 0.421, p = 0.008) and emotional distress (TFI-E; r = 0.370, p = 0.022), though these findings did not survive multiple comparison correction.ConclusionEarly tinnitus is characterized by ACC GABAergic deficits, while chronicity features normalized GABA+/Cr levels—suggesting compensatory neuroplastic restoration of inhibition over time. Glutamatergic activity may modulate perceptual and emotional dimensions of tinnitus. These phase-specific ACC neurochemical shifts highlight potential therapeutic targets for arresting tinnitus progression. Longitudinal studies are warranted to validate temporal dynamics.