Background:Point-of-care ultrasound (PoCUS) is now more commonly being utilized in cardiopulmonary resuscitation to offer immediate evaluation of cardiac motion and to direct management. However, the evidence is still heterogeneous. Therefore, the purpose of this systematic literature review and meta-analysis is to assess the clinical utility of PoCUS in terms of its effects on resuscitation outcomes. Methods:A systematic review and meta-analysis were performed by searching MEDLINE, EMBASE, Cochrane Central, Web of Science, and Scopus databases. Studies on adult patients with in-hospital cardiac arrest or out-of-hospital cardiac arrest (OHCA) were included following predefined Population, Intervention, Comparison, Outcomes, and Study Design (PICOS) criteria. Two reviewers independently screened, extracted data, and appraised quality using Risk of Bias 2 (RoB 2) tool for randomized trials and the Risk of Bias in Non-randomized Studies of Interventions (ROBINS-I) tools. Subgroup and sensitivity analyses were performed to assess robustness and the presence of publication bias was evaluated. Results:Pooled results of all included studies involving patient populations showed that PoCUS during cardiopulmonary resuscitation (CPR) increased the rate of the return of spontaneous circulation (ROSC) (43% vs. 28%). Survival to hospital discharge was also greater in the PoCUS group (19 % vs. 12 %). Subgroup and sensitivity analyses confirmed the robustness of the results, and no significant publication bias was found. Conclusions:This systematic review and meta-analysis indicates that performing PoCUS during CPR is associated with better clinical outcomes, including increased ROSC and survival to hospital discharge, further highlighting the potential advantages of PoCUS to assist in cardiac arrest management as well as making clinical decisions.
Background:Parotid gland tumors are the most common type of salivary gland neoplasms. The accurate determination of tumor nature is essential for clinical treatment decision-making. Existing deep learning (DL)-based methods for image segmentation and differential diagnosis often yield blurred boundaries when segmenting parotid tumors on computed tomography (CT) images and face difficulties in distinguishing benign from malignant lesions. This study aimed to develop a solution based on an improved TransUNet to achieve precise tumor segmentation and to enable automatic benign-malignant differentiation of parotid gland tumors by fusing CT images with clinical text information. Methods:First, we propose an enhanced PT-TransUNet segmentation model. Built upon the classic TransUNet architecture, it incorporates a learnable Difference of Gaussians (DoG) edge enhancement module at the front end to adaptively sharpen tumor boundary features. In addition, a PSA-CBAM module-which integrates multi-scale convolution with a convolutional block attention module-is embedded in the decoder to improve the model's capability to capture multi-scale features. Second, we construct a multimodal diagnostic model that uses PT-TransUNet as the image branch to extract CT image features, employs the Chinese BERT Whole Word Masking (WWM) model as the text branch to extract clinical text features, and performs deep feature fusion via a bidirectional cross-modal attention mechanism to achieve binary classification of benign versus malignant tumors. Results:On a dataset of 158 CT images, PT-TransUNet achieved Dice coefficients of 0.8212±0.0269 for parotid gland segmentation and 0.8075±0.0245 for parotid tumor segmentation, compared with 0.8118±0.0324 and 0.7928±0.0350 by nnUNet, respectively (P<0.05). The 95% Hausdorff distance (HD95) was 16.14±0.58 mm for benign tumors and 6.28±0.58 mm for malignant tumors. In the multimodal diagnostic task, the fusion model using bidirectional cross-modal attention achieved an accuracy of 0.8909±0.0249 and a sensitivity of 0.9000±0.1369, whereas the CT-only model achieved 0.6091±0.0249 accuracy and 0.1500±0.1369 sensitivity, and the text-only model achieved 0.7273±0.0321 accuracy and 0.4000±0.1369 sensitivity (P<0.001 for both comparisons). Compared with feature-level concatenation, which achieved an accuracy of 0.8455±0.0518, and decision-level weighted average, which achieved an accuracy of 0.7545±0.0249, the cross-modal attention mechanism yielded higher accuracy. Conclusions:This study confirms the superiority of the proposed PT-TransUNet model for CT image segmentation of parotid tumors and highlights the great potential of the cross-modal fusion strategy that integrates CT images with clinical text information for benign-malignant differentiation. This approach provides a new technical pathway for automated and accurate assisted diagnosis of parotid gland tumors and holds important value for clinical application.
Background:Size-based risk stratification often overlooks small but unstable intracranial aneurysms (IAs). Aneurysm wall enhancement (AWE) on vessel wall imaging (VWI) is a validated marker of wall instability, yet the local hemodynamic drivers of this pathology, particularly in complex anterior communicating artery (ACoA) aneurysms, remain incompletely characterized. This study leverages a combined computational fluid dynamics (CFD)-VWI approach to characterize the mechanobiological coupling between local hemodynamics and quantitative wall remodeling in ACoA aneurysms. Methods:We retrospectively analyzed 24 patients harboring 25 ACoA aneurysms. A Vector-Integrated Surface Parametrization (VISP) pipeline achieved sub-voxel sampling density [through adaptive interpolation rather than imaging resolution beyond the native 0.6 mm magnetic resonance imaging (MRI) voxel] for co-registration of CFD and 3T-VWI, with wall enhancement defined at a contrast ratio (CR) ≥0.6. To identify hemodynamic drivers of enhanced wall thickness (EWT) while explicitly accounting for within-patient hierarchical clustering, four complementary analytical frameworks were applied in parallel: (I) intra-patient paired bootstrap tests (2,000 resamples) comparing enhanced and non-enhanced wall segments within each of the 13 AWE-positive patients; (II) a multivariate linear mixed model (LMM) with patient-level random intercepts for EWT severity (n=12,473 enhanced segments); (III) generalized estimating equations (GEEs) with cluster-robust variance for AWE presence (n=157,284 segments); and (IV) ensemble machine-learning models (Random Forest and XGBoost) interpreted via Shapley Additive exPlanations (SHAP) values across segment-level, patient-centered, and patient-level GroupKFold cross-validation (CV). Cross-patient generalization of EWT prediction was disclosed separately as an out-of-sample analysis. Results:Focal AWE was identified in 14 of 25 aneurysms (13 patients), spatially coinciding with hemodynamic stagnation zones. Enhanced segments exhibited significantly lower local Pressurepeak (∆ =-35.25 Pa, PFDR =0.02) and wall shear stress (WSS)peak (∆ =-3.97 Pa, PFDR <0.001) compared to non-enhanced segments under intra-patient paired bootstrap testing. Three further frameworks converged on Pressurepeak as the dominant independent driver of wall thickness among enhanced segments: multivariate LMM β=-0.181 (P=2.31×10-11); GEE β=-0.5452 (robust P=0.0499); and a Random Forest model, in which Pressurepeak ranked first by SHAP feature importance at the segment level and remained among the top three across every CV regime. Conclusions:We present a facet-level CFD-VWI pipeline that achieves sub-voxel sampling density for spatially mapping local hemodynamics onto quantitative wall remodeling in ACoA aneurysms on a clinical 3T platform. Across four independent hierarchical analyses, local Pressurepeak consistently emerged as the dominant independent hemodynamic driver of wall thickening among enhanced segments, complementing the established low-WSS association. This framework is intended as a mechanistic explanatory tool for local hemodynamic-AWE coupling; broader clinical translation will require larger, externally validated cohorts.
Background:Robert's uterus (also known as Robert's uterine malformation) is a rare congenital anomaly characterized by a complete, asymmetric muscular septum that divides the uterine cavity into two separate hemi-cavities, one of which is often blind or severely hypoplastic, while the cervix is single and exhibits a normal morphological configuration with a single normal cervix. It is associated with poor reproductive outcomes, including infertility and recurrent pregnancy loss. Accurate diagnosis with conventional two-dimensional (2D) ultrasound is challenging. Three-dimensional (3D) ultrasound enables coronal-plane reconstruction of the uterus, which is crucial for a definitive diagnosis. This study aimed to summarize the 3D ultrasound imaging features of Robert's uterus and evaluate its diagnostic value and impact on reproductive outcomes. Case Description:This retrospective case series included 10 patients with Robert's uterus confirmed by hysteroscopy and laparoscopy. All patients underwent preoperative transvaginal 2D and 3D ultrasound. 3D ultrasound clearly demonstrated the characteristic features in all cases: a broad fundus, a complete uterine septum extending to just above the internal cervical os, resulting in two completely separate cavities with visible asymmetry in size and shape, and a single normal cervical canal. The diagnostic accuracy of 3D ultrasound compared with surgical findings was 100%. The main clinical presentations were primary infertility (four cases) and recurrent miscarriage (four cases). All patients underwent hysteroscopic septoplasty under laparoscopic guidance based on 3D ultrasound findings. Postoperative 3D ultrasound at three months showed a largely restored uterine cavity with a residual septum <1 cm in all cases. Conclusions:3D ultrasound provides a non-invasive, intuitive, and accurate method for diagnosing Robert's uterus by clearly displaying its characteristic anatomy in the coronal plane. It is an ideal tool for preoperative diagnosis, surgical planning, and postoperative assessment. 3D ultrasound should be considered a first-line imaging modality for the evaluation of suspected uterine anomalies, especially in women with unexplained infertility or recurrent pregnancy loss.
Background:The application of artificial intelligence (AI) in ultrasound (US)-guided regional anesthesia has expanded, particularly in enhancing the accuracy, safety, and training effectiveness of procedures through deep learning-based anatomical segmentation. This study aimed to develop and validate an automatic segmentation model for supraclavicular-to-interscalene brachial plexus block (ISB) using the You Only Look At CoefficienTs (YOLACT) algorithm and to compare its performance with that of a U-Net model. Methods:A total of 1,100 patients scheduled to undergo ISB were enrolled. US images encompassing the anatomical range from the supraclavicular fossa to the C7 vertebral level were acquired by experienced anesthesiologists. Images were annotated to identify the brachial plexus nerve, anterior scalene muscle (ASM), middle scalene muscle (MSM), and subclavian artery (SA). YOLACT and U-Net models were trained for automatic segmentation. Model performance was assessed using Intersection over Union (IoU), Dice similarity coefficient (DSC), Hausdorff distance (HD), the proportion of images with a brachial plexus nerve IoU > 0.5, and segmentation accuracy metrics. Results:A total of 6,600 US images were analyzed. The YOLACT model demonstrated significantly higher IoU and DSC values and lower HD values compared with the U-Net model for segmentation of the brachial plexus nerve, ASM, MSM, and SA (P<0.001). The proportion of images with a brachial plexus nerve IoU greater than 0.5 was also significantly higher with YOLACT (P<0.001). Conclusions:An automatic segmentation model for ISB, spanning the supraclavicular region to the C7 level, was developed using the YOLACT algorithm. Although quantitative performance metrics favored YOLACT over U-Net, subjective accuracy assessments were comparable between models. Further studies using larger datasets are required to clarify the potential clinical applicability of this approach.
Background:Accurate segmentation of glioma subregions from multimodal magnetic resonance imaging (MRI) is essential for diagnosis, treatment planning, and response assessment, but remains challenging because of boundary ambiguity, heterogeneous appearance, and small enhancing tumor (ET) components. This study aimed to develop and evaluate a controlled nnU-Net v2-based framework for three-dimensional (3D) brain tumor segmentation (BraTS) by improving adaptive feature representation and boundary-aware learning while preserving the reproducibility of the self-configuring nnU-Net pipeline. Methods:We propose a conditional convolution and squeeze-and-excitation with boundary-aware learning network (CondSEB-Net), a controlled enhancement of the nnU-Net v2 framework for 3D medical image segmentation. Conditional convolution (CondConv) is introduced to improve sample-specific feature adaptation, multi-level 3D squeeze-and-excitation (SE) attention is used to recalibrate channel responses, and a boundary-aware loss based on signed distance maps (SDMs) is incorporated to strengthen contour-level supervision. The proposed method was evaluated on BraTS2020 and kidney tumor segmentation (KiTS)2019 using fixed five-fold cross-validation and compared with representative 3D segmentation baselines. Results:On BraTS2020, CondSEB-Net improved the average Dice score from 0.8525±0.0217 to 0.8659±0.0184 and reduced the average 95% Hausdorff distance (HD95) from 10.57±4.01 to 9.08±3.64 mm compared with nnU-Net v2. The average symmetric surface distance (ASSD) also decreased from 3.08±0.37 to 2.65±0.30 mm. Fold-wise paired statistical analysis showed statistically supported improvements in both region overlap and boundary-related metrics. On KiTS2019, CondSEB-Net also achieved superior or competitive performance with limited additional computational overhead. Conclusions:CondSEB-Net improves adaptive feature representation and boundary-related segmentation performance while preserving the self-configuring pipeline of nnU-Net v2. The current results provide benchmark-level evidence for the proposed framework, while external multi-center validation remains necessary before clinical deployment.
Background:Although the peak value of the renal blood perfusion time-activity curve obtained by renal dynamic imaging (RDI) represents the real-time maximum renal blood perfusion (MAX-RP), it is easily affected by the radioactive activity of the injection and is difficult to quantitatively measure. This study proposes a renal blood perfusion peak ratio (RBPPR) model to eliminate injection dose differences and explore its clinical value in quantitatively evaluating MAX-RP. Methods:A retrospective analysis was conducted on 151 cases (a total of 205 kidneys) who underwent RDI examination from August 2020 to June 2025. According to the 30% limit of the peak difference rate of renal graph and the normal lower limit value of total glomerular filtration rate (GFR), the participants were divided into a control group and an experimental group (which included: the completely compensated group and the incompletely compensated group). Analysis was conducted of distribution of RBPPR and the differences among groups, its correlation with GFR, and its diagnostic efficacy. Results:RBPPR values were concentrated within each group [standard deviation (SD) ≤1.35%]. RBPPR strongly correlated with each kidney GFR (ekGFR; r=0.778, P<0.001). Across the three groups, both ekGFR (F=94.123, P<0.001) and RBPPR (F=43.336, P<0.001) differed significantly. Compared with the control group (RBPPR 3.67%±0.90%, ekGFR 49.12±9.11 mL/min/1.73 m2), the complete compensation group showed significantly higher RBPPR (5.20%±1.35%, P<0.001) and ekGFR (76.00±17.75, P<0.001), whereas the incomplete compensation group exhibited elevated ekGFR (58.12±8.50, P<0.001) but no significant RBPPR increase (3.92%±0.66%, P=0.145 vs. control). For diagnosing complete renal compensation, the optimal RBPPR threshold was 3.620%, yielding an area under the curve (AUC) of 0.842 [95% confidence interval (CI): 0.782-0.902], sensitivity 91.6%, and specificity 62.0%. Conclusions:The RBPPR model can stably and quantitatively evaluate MAX-RP with good accuracy. It provides a simple and reliable new quantitative method for evaluating whether renal function can be completely compensated and the assessment of renal reserve function.
This retrospective multicenter study evaluated whether deep learning-based super-resolution (SR) reconstruction can enhance structural conspicuity in thick-section chest computed tomography (CT) and improve the detection performance of an artificial intelligence (AI)-based computer-aided detection (CAD) system for lung nodules in 96 patients with colorectal cancer (CRC) undergoing chest CT for metastatic surveillance. Pulmonary nodules <10 mm and ≤3 per patient were analyzed. Three image sets were evaluated using a commercial AI-CAD system: original thin-section images (reference), original 5-mm thick-section images, and SR-converted thin-section images. Nodule- and patient-level detection performances were compared using the Cochran-Mantel-Haenszel test and aligned rank transform analysis of variance (ART-ANOVA). Quantitative nodule metrics, image noise, and morphologic consistency were assessed between original and SR-converted thin-section images. Among 105 reference nodules, AI sensitivity increased from 31.4% with thick-section images to 61.0% with SR-converted images, and positive predictive value (PPV) increased from 42.9% to 80.0%. Patient-level sensitivity improved from 41.5% to 67.1%. SR reconstruction reduced image noise (P<0.001) and preserved nodule morphology, with 95.3% anatomic concordance and 80% solid-feature consistency. Nodule size remained comparable; however, density was lower in converted images. SR reconstruction generated 20% hallucinated nodules (16/80), predominantly benign or artifactual structures. In conclusion, SR reconstruction enhances AI-based pulmonary nodule detection by compensating for structural detail lost in thick-section CT. Despite hallucinated nodules, SR reconstruction may provide a feasible harmonization strategy for retrospective multicenter AI research using heterogeneous CT datasets, although further validation in larger populations is required.
Background:Ischemia with non-obstructive coronary arteries (INOCA) is associated with elevated cardiovascular risk, yet early subclinical myocardial changes often go undetected by conventional imaging. This study aimed to determine whether patients with INOCA exhibit subclinical tissue alterations identified by T1 mapping and left ventricular wall motion abnormalities assessed by cardiac magnetic resonance feature tracking (CMR-FT), in comparison with matched healthy controls (HC). Methods:A total of 44 patients with INOCA and 22 matched HC were prospectively enrolled from August 2024 to February 2026, and all underwent hybrid cardiac positron emission tomography/magnetic resonance imaging (PET/MRI) examination. CMR-derived parameters were obtained and compared across the study groups. Results:Native T1 values increased progressively across three groups: HC, INOCA with myocardial flow reserve (MFR) >2, and INOCA with MFR <2 (1,196±19 vs. 1,213±8 vs. 1,244±20 ms; P<0.001). In contrast, the absolute values of global longitudinal strain (GLS) showed a gradual reduction (-19.21%±1.84% vs. -17.83%±0.92% vs. -16.25%±1.59%; P<0.001), whereas no significant intergroup differences were observed in left ventricular ejection fraction (LVEF) or extracellular volume (ECV) (both P>0.05). Both native T1 and GLS were negatively correlated with MFR (r=-0.739, P<0.001; r=-0.638, P<0.001, respectively). Similarly, native T1 and ECV were positively correlated with GLS (P<0.001 and P=0.004). Conclusions:Native T1 and GLS correlate with impaired MFR in patients with INOCA, consistent with subclinical left ventricular (LV) systolic dysfunction related to myocardial ischemia. Incorporating CMR into clinical practice may aid in the monitoring and treatment of patients with INOCA, and further studies exploring the underlying mechanisms are warranted.
Background:Conventional computed tomography (CT)-based liver fat assessment relies mainly on mean attenuation or liver-to-spleen ratios, whereas a novel voxel-based approach enables whole-liver, voxel-level quantification of hepatic fat. However, its value for predicting incident type 2 diabetes (T2D) remains unclear. Therefore, this study aimed to validate whether voxel-based liver fat quantification derived from routine CT can be used to assess the risk of incident T2D in a health check-up population. Methods:In this retrospective cohort study, individuals who underwent routine noncontrast abdominal CT examinations between January 2013 and December 2023 and were free of diabetes at baseline were included. Liver fat was quantified via an automated voxel-based method incorporating spleen-referenced attenuation criteria on the basis of whole-liver segmentation. The average liver fat fraction was defined as the proportion of liver voxels classified as fat. Incident T2D was identified through longitudinal review of outpatient, inpatient, and health check-up records. Associations between liver fat and incident T2D were evaluated via Cox proportional hazards models adjusted for demographic, anthropometric, and metabolic covariates. Restricted cubic spline (RCS) analysis was used to explore dose-response relationships. Predictive performance was assessed via time-dependent receiver operating characteristic (ROC) analysis at 3 and 5 years, calibration plots, and decision curve analysis (DCA). Incremental predictive value was evaluated by comparing liver fat-based models with hemoglobin A1c (HbA1c)-based and clinical models. Results:Among 356 participants [median age, 57 years, interquartile range (IQR), 53-63 years; 56.5% male], 32 individuals (9.0%) developed incident T2D during a median follow-up of 4.5 years. A higher average liver fat fraction was significantly associated with an increased risk of incident T2D after multivariable adjustment [hazard ratio (HR) per standard deviation (SD), 1.65; 95% confidence interval (CI): 1.05-2.60, P=0.031]. RCS analysis demonstrated a monotonic increase in diabetes risk with increasing liver fat fraction. The liver fat fraction showed moderate discrimination for incident T2D, with time-dependent area under the curve (tAUC) values of 0.71 (95% CI: 0.59-0.84) at 3 years and 0.72 (95% CI: 0.60-0.82) at 5 years. Compared with clinical variables or HbA1c alone, the addition of liver fat improved predictive performance for incident T2D, increasing the C-index from 0.574 (0.453-0.680) to 0.740 (95% CI: 0.663-0.803). Conclusions:Voxel-based CT quantification of liver fat is independently associated with incident T2D and provides incremental value for diabetes risk assessment in a health check-up population.
Background:Current artificial intelligence (AI) models for Parkinson's disease (PD) diagnosis via magnetic resonance imaging (MRI) are significantly impeded by domain shift, and performance often decreases due to heterogeneity in imaging protocols and scanners across hospitals. While existing unsupervised domain adaptation (UDA) methods combining self-training and adversarial training enhance model generalizability, their dependence on pseudo-labels often introduces confirmation bias from noisy predictions. This study aimed to mitigate pseudo-label noise, enhance domain-invariant feature learning and develop a novel UDA framework for robust cross-center PD diagnosis. Methods:We propose a curriculum-guided unified UDA (CGU-UDA) framework that integrates self-training and adversarial training. Its core innovation is a feedback loop between adaptive pseudo-label refinement and contextual feature regularization. First, a curriculum learning scheduler dynamically adjusts confidence thresholds per class based on real-time learning progress and progressively filters high-quality pseudo-labels. Second, a consistency constraint module enforces prediction agreement between original target images and their randomly masked image and leverages these refined labels to promote robust, context-aware feature learning. Finally, an adversarial domain discriminator conditioned on a randomized multilinear map is applied to align feature distributions across domains. Results:Evaluations on two independent multi-cohort PD MRI datasets show that CGU-UDA consistently surpasses leading UDA benchmarks. It achieves an average increase of over 2% in classification accuracy and over 3% in area under the curve (AUC) across varied cross-domain tasks. On Hospital→Parkinson's Progression Markers Initiative (PPMI), accuracy improved from 65.81% of baseline to 67.95%, AUC from 0.6474 to 0.7148; on PPMI→Hospital, accuracy improved from 63.83% to 65.96%, AUC from 0.6324 to 0.6687. Ablation studies confirm that both the dynamic thresholding mechanism and the masked consistency constraint are crucial to this performance gain. Conclusions:This work advances robust medical AI model by directly tackling the pseudo-label noise problem in domain adaptation. The CGU-UDA framework demonstrates strong potential for deploying reliable diagnostic models across diverse clinical settings and leading to effective clinical application.
Background:In robotic endodontic navigation, preoperative cone-beam computed tomography (CBCT) provides root canal information, whereas intraoperative vision usually provides only single-view tooth surface images. This study aimed to reconstruct the target tooth occlusal surface from a single visible-light image as a geometric interface for CBCT registration and root canal information mapping. Methods:A statistical shape model (SSM) was constructed from 50 homologous tooth intraoral scan (IOS) models using curvature-adaptive non-rigid iterative closest point (NICP) registration. For each single-view image, the target tooth region was segmented, illumination interference was suppressed, and fissure features were enhanced. The extracted texture cues were converted into a bounded pseudo-height field and injected into the statistical template under top-surface gating and vertex stability weighting. Results:On the occlusal region of interest (ROI), the proposed method achieved a root mean square distance (RMSD) of 0.432 mm, average symmetric surface distance (ASSD) of 0.308 mm, Hausdorff distance (HD) of 2.540 mm, Chamfer distance (CD) of 0.466 mm2, and Dice similarity coefficient (DSC) of 0.340. It outperformed the selected single-view comparison method in surface error metrics. Ablation experiments further showed that illumination normalization, stability weighting, and accurate segmentation contributed to improved reconstruction stability and detail recovery. Conclusions:The proposed method reconstructs an anatomically constrained occlusal surface from a single-view image and provides a practical geometric interface for subsequent CBCT registration and root canal information mapping. It does not directly reconstruct the root canal anatomy or replace CBCT-based canal identification. Current validation remains limited to dental models, limited sample diversity, and intact or nearly intact occlusal morphology; altered clinical crowns require further evaluation.
Background:Non-valvular atrial fibrillation (NVAF) carries a high risk of left atrial appendage thrombus (LAAT) and dense spontaneous echo contrast (dense SEC), the primary triggers of cardioembolic stroke. Conventional CHADS2 [congestive heart failure, hypertension, age ≥75 years, diabetes mellitus, prior stroke/transient ischemic attack (TIA) score] and CHA2DS2‑VASc (congestive heart failure, hypertension, age ≥75 years, diabetes mellitus, prior stroke/TIA, vascular disease, age 65-74 years, Sex category score) scores lack left atrial appendage (LAA) morphological features, yielding limited predictive accuracy for dense SEC/LAAT. This study constructed a nomogram based on quantitative LAA parameters derived from three-dimensional transesophageal echocardiography (3D-TEE) to predict dense SEC/LAAT in NVAF patients and compared its performance with the two conventional clinical risk scores. Methods:We retrospectively enrolled 159 NVAF patients who underwent 3D-TEE from July 2024 to December 2025, stratified into a dense SEC/LAAT positive group (n=50) and a negative group (n=109). Variables with severe multicollinearity [variance inflation factor (VIF) ≥10] were excluded. Univariate logistic regression (P<0.10) screened candidate predictors, followed by forward stepwise multivariate logistic regression to identify independent risk factors and construct a nomogram. Model discrimination, calibration and clinical utility were assessed via receiver operating characteristic (ROC) curves, calibration curves, 10-fold cross-validation, decision curve analysis (DCA) and clinical impact curves; inter-model area under the curve (AUC) comparisons used P<0.05 as the statistical significance threshold. Results:Four independent predictors of dense SEC/LAAT were identified: D-dimer >0.550 mg/L [odds ratio (OR) =7.805, 95% confidence interval (CI): 2.044-29.795, P=0.002]; European Heart Rhythm Association (EHRA) score ≥ IIb (OR =6.255, 95% CI: 1.458-26.833, P=0.013); LAA poor echogenicity (OR =23.037, 95% CI: 5.651-93.909, P<0.001); LAA orifice morphology (OR =0.537, 95% CI: 0.296-0.975, P=0.041). The nomogram achieved an original AUC of 0.892 (95% CI: 0.839-0.945) at an optimal cutoff value of 0.29, with sensitivity 0.82, specificity 0.83, accuracy 0.82, and negative predictive value 0.91. It significantly outperformed CHADS2 (AUC =0.629) and CHA2DS2-VASc (AUC =0.606) (all P<0.05). Ten-fold cross-validation yielded an optimism-corrected AUC of 0.868 and a mean Brier score of 0.127; however, marked overfitting was observed (calibration slope =4.175, mean maximum calibration error =0.385). DCA confirmed sustained positive net clinical benefit within the 10-50% threshold probability range. Conclusions:The 3D-TEE-based nomogram shows acceptable discrimination for dense SEC/LAAT in NVAF patients and addresses the limitations of traditional risk scoring systems. Nevertheless, prominent overfitting prevents its direct clinical use without external validation and recalibration; it can serve as an auxiliary research tool for LAAT risk stratification.
Background:Ultrasound characteristics of the Giacomini vein (GV) in Chinese populations remain poorly defined. The aim of this study was to analyze the anatomical and hemodynamic characteristics of the GV using color Doppler ultrasound and to provide evidence to support clinical diagnosis and treatment. Methods:Prospectively, 121 patients with lower extremity discomfort and varicose veins were continuously collected from July 2025 to January 2026, involving a total of 238 limbs. Patients were categorized into a primary group (C0) and a varicose group (C1-C4) according to the CEAP (Clinical, Etiological, Anatomical, and Pathophysiological) classification. All patients underwent ultrasound examination, and 38 cases underwent digital subtraction angiography (DSA). The GV was classified into four types. Agreement between ultrasound and DSA findings was assessed using Kappa statistics. Differences in GV detection rate, inner diameter, and reflux-related parameters were compared between groups, among GV types, and according to the severity of saphenous vein reflux. Results:Ultrasound and DSA demonstrated high concordance in GV classification (kappa =0.854). The overall GV detection rate was 73.95%, with a significantly higher rate in the right lower limb than in the left lower limb (P<0.05) and no significant difference between sexes. The great saphenous vein (GSV) confluence type was the most prevalent (73.3%). Detection rates of GV presence and reflux were significantly higher in the varicose group than in the primary group (P<0.05), whereas no significant difference in inner diameter was observed. Reflux detection rates showed an increasing trend across varicose subgroups; however, this trend did not reach statistical significance. The incidence of GV reflux was significantly higher in the small saphenous vein (SSV) reflux group and in the group with both GSV and SSV reflux than in the GSV reflux group (P<0.05). The severity of saphenous vein reflux had no significant effect on GV-related parameters. Conclusions:Ultrasound demonstrated significant clinical utility in the assessment of the GV. GV reflux was closely associated with the severity of lower extremity varicose veins and with the type of saphenous vein reflux. Prioritization of preoperative GV assessment may reduce postoperative recurrence rates and minimize intraoperative injury.
Background:Gestational hypertension (GH) is associated with maternal vascular remodeling; however, the biomechanical and hemodynamic features related to increased carotid intima-media thickness (CIMT) remain insufficiently characterized. This study aimed to evaluate carotid remodeling in women with GH using radiofrequency data-based quantitative vessel stiffness (R-QVS) analysis and vector flow imaging (VFI). Methods:This prospective observational study included 347 pregnant women, comprising 108 normotensive controls and 239 women with GH. The GH group was further divided into normal-CIMT and increased-CIMT subgroups based on a mean CIMT threshold of 1.0 mm. All participants underwent standardized carotid ultrasonography. Stiffness-related parameters, including the hardness coefficient (HC) and pulse wave velocity (PWV), and flow-related parameters, including maximum wall shear stress (WSSmax) and mean wall shear stress (WSSmean), were measured. Diagnostic performance was assessed using receiver operating characteristic (ROC) analysis, and the factors associated with increased CIMT were evaluated using logistic and linear regression analyses. Results:Carotid stiffness increased and wall shear stress (WSS) decreased progressively from the control group to the GH normal-CIMT and GH increased-CIMT groups. The GH increased-CIMT group had the highest PWV and the lowest WSSmean. The PWV values were 5.87±0.68, 7.05±0.77, and 8.11±0.85 m/s in the control, GH normal-CIMT, and GH increased-CIMT groups, respectively, while the WSSmean values were 1.30±0.38, 0.89±0.32, and 0.65±0.27 Pa, respectively; all group differences were statistically significant. The HC and PWV showed strong diagnostic performance in the detection of increased CIMT, with areas under the curve (AUCs) of 0.909 and 0.885, respectively. In the multivariable logistic regression analysis, pre-pregnancy body mass index (BMI), systolic blood pressure (SBP), triglycerides (TG), and PWV were independently associated with increased CIMT, while WSSmean was higher WSSmean was independently associated with lower odds of increased CIMT. In the linear regression analysis, the same key predictors were associated with CIMT as a continuous outcome, and the model explained 56.8% of variance in CIMT, with an adjusted R2 of 0.553. Conclusions:Women with GH exhibited a progressive high-stiffness and low-shear carotid phenotype, particularly in the presence of increased CIMT. Integrating R-QVS analysis and VFI may improve vascular risk stratification in hypertensive pregnancies.
Background:Differentiating cardiac amyloidosis (CA) from hypertrophic cardiomyopathy (HCM) is clinically challenging due to overlapping phenotypes of left ventricular hypertrophy. This study aimed to evaluate the diagnostic value of cardiac magnetic resonance (CMR)-derived left atrial strain parameters in differentiating CA from HCM. Methods:A retrospective analysis was performed on CMR data obtained from 37 patients with HCM, 41 patients with CA, and 29 individuals in the normal control group. Left ventricular and left atrial functions were quantified using CVI42 software (version 5.14.2). Left atrial strain parameters were assessed using the tissue tracking module, including left atrial reservoir strain (εs) and strain rate (SRs), conduit strain (εe) and strain rate (SRe), and booster strain (εa) and strain rate (SRa). Differences in left ventricular and left atrial functional parameters, as well as left atrial strain parameters among the three groups, were analyzed using the Kruskal-Wallis test. Pairwise comparisons were conducted using the Mann-Whitney U test. Receiver operating characteristic (ROC) curve analysis was performed using MedCalc (version 15.2.2) to determine the diagnostic efficacy of left atrial strain parameters in differentiating CA from HCM. Results:The absolute values of εs, εe, εa, SRs, SRe, and SRa were significantly lower in both the HCM and CA groups compared with the normal control group (P<0.05). Furthermore, the absolute values of these parameters were markedly lower in CA compared with HCM (P<0.05). Among these parameters, εs demonstrated the highest diagnostic performance for distinguishing CA from HCM, with an area under the ROC curve of 0.921. Conclusions:CMR-derived left atrial strain parameters, particularly εs, might serve as a reliable means for differentiating between HCM and CA, which need further prospective study including patients with all HCM types and conducting internal and external validation to confirm credibility and expand generalizability.
Background:Quantitative liver T1 mapping is a promising noninvasive biomarker for diffuse liver disease, but it is influenced by physiological variability. Hydration alters tissue water content and perfusion, yet its effect on liver water-specific T1 (wT1) remains poorly characterized, particularly across liver segments. Prior studies relied on single-slice acquisitions, limiting assessment of spatial heterogeneity. The objective is to assess hydration-related changes in liver wT1 across all Couinaud segments using an accelerated multi-slice technique and to compare results with single-slice T1-modified Look-Locker inversion recovery (T1-MOLLI) and vibration-controlled transient elastography (VCTE). Methods:Twenty-nine healthy adults underwent blood sampling, VCTE, and liver magnetic resonance imaging (MRI) after an ≥8-hour fast without fluid intake and again after ingestion of 1 L of water followed by a 1-hour equilibration period. Multi-slice wT1 mapping was performed using a Dixon-based continuous inversion-recovery Look-Locker (CIR-LL) sequence covering the entire liver in a single breath-hold. Single-slice T1-MOLLI and whole-liver proton density fat fraction (PDFF) and T2* mapping were acquired for comparison. Regions of interest (ROIs) were placed on PDFF, T2*, and wT1 maps in all liver segments. Statistical analysis included intraclass correlation coefficient (ICC; [2, 1]), linear regression, Pearson correlation coefficients Bland-Altman plots and paired two-sided t-tests. Results:Hydration resulted in a higher global increase in mean liver wT1 (before: 773.4±63.2 ms, after: 800.3±66 ms) than T1-MOLLI (before: 853±68.8 ms, after: 861.6±55.9 ms). PDFF, T2*, and liver stiffness varied modestly within the physiological range. For liver stiffness, T1-MOLLI, and wT1, measurements obtained before and after hydration showed good agreement, with wT1 (y=0.9x+80.5, r=0.89, P<0.001) exhibiting a regression coefficient closer to that of liver stiffness (y=0.9x+0.7, r=0.63, P<0.001) than T1-MOLLI (y=0.6x+372.4, r=0.71, P<0.001). T1-MOLLI and wT1 showed good agreement under both hydration conditions (before: y=0.5x+355.7, r=0.64, after: y=0.8x+98.9, r=0.78), with T1-MOLLI yielding systematically higher values than wT1. Paired analysis demonstrated significant hydration-related changes in wT1 (P<0.001), with segment-wise wT1 increases remaining significant after correction for multiple testing. Furthermore, wT1 revealed spatial heterogeneity across liver segments. Conclusions:Liver wT1 is physiologically modulated by hydration and exhibits spatial heterogeneity across liver segments, even in the healthy liver. These findings underscore the importance of whole-liver, segment-resolved wT1 mapping and physiological standardization when interpreting quantitative liver MRI.
