Quantitative Imaging in Medicine and Surgery最新文献

Background: The Carotid Plaque Reporting and Data System (Plaque-RADS) standardizes plaque risk assessment, but its ability to predict the prevalence of stroke has not yet been fully investigated. This study aimed to evaluate the association between the Plaque-RADS categories and stroke prevalence in a large prospective cohort using two-dimensional (2D) carotid ultrasound.

Methods: In total, 2,023 patients undergoing carotid ultrasound were enrolled in this prospective cohort study. Stroke risk was stratified using Plaque-RADS categories 1-4. Based on clinical diagnosis, the patients were categorized into stroke and non-stroke groups. The associations between the Plaque-RADS categories and stroke prevalence/recurrence were analyzed by multivariate logistic regression, adjusted for gender, age, and diabetes mellitus. A linear trend test was performed across the ordinal categories.

Results: The stroke group (n=212) had a higher proportion of males (65.1% vs. 56.5%, P<0.05), an older median age (70.5 vs. 63 years, P<0.05), and a higher prevalence of diabetes (42.6% vs. 29.3%, P<0.05) than the non-stroke group (n=1,811). The prevalence of stroke increased significantly across the Plaque-RADS categories: 3.1% (13/412) in category 1, 10.1% (121/1,198) in category 2, 17.2% (65/379) in category 3, and 38.2% (13/34) in category 4 (trend P<0.001). After adjustment for confounders, Plaque-RADS category 4 conferred the highest stroke risk (adjusted odds ratio =8.13, 95% confidence interval: 3.18-20.78), with a significant linear trend in risk across categories (P for trend <0.001). In relation to short-term (0-12 months) stroke recurrence among the 199 stroke patients, the recurrence rates demonstrated a graded association with the Plaque-RADS categories (P=0.008): 10.7% (13/121) in category 2, 23.1% (15/65) in category 3, and 38.5% (5/13) in category 4. This association was particularly strong in the 0-6-month period.

Conclusions: The Plaque-RADS classification, as applied in 2D carotid ultrasound, effectively stratifies both initial and short-term recurrent stroke risk, supporting its integration into routine clinical practice for risk assessment.

Background: Nonarteritic anterior ischemic optic neuropathy (NAION) is among the most common acute optic nerve disorders in individuals over 50 years old. Optical coherence tomography angiography (OCTA) has enabled the quantitative assessment of retinal structure and microcirculation, providing insights into the mechanisms underlying NAION. This study aimed to investigate the longitudinal structure-function (S-F) and vasculature-function (V-F) relationships in the macula across different stages of NAION.

Methods: This study examined 32 eyes from 32 patients with acute NAION. Optical coherence tomography (OCT) and OCTA were used to assess macular ganglion cell-inner plexiform layer (mGCIPL) thickness and macular vessel density (VD) at baseline, 1 month, and 3 months. Correlations between these structural and vascular parameters and visual outcomes, including best-corrected visual acuity (BCVA), visual field index (VFI), and mean deviation (MD), were analyzed.

Results: Compared to the acute stage, the subacute phase exhibited significant reductions in mGCIPL thickness and macular VD. Significant global or regional correlations were found in S-F relationship only in subacute and chronic stage (P<0.05). In the acute stage, perifoveal VD of the superficial vascular plexus (SVP) in nasal section was correlated with the corresponding BCVA (r=-0.372; P=0.036), VFI (r=0.465; P=0.010), and MD (r=0.431; P=0.017) parameters. In the subacute and chronic stages, VD of SVP in most regions was significantly correlated with visual outcomes. None of the VDs in the deep vascular complex (DVC) were significantly correlated with visual outcomes.

Conclusions: Significant V-F correlations were present across all stages of NAION, while S-F correlations were observed only in subacute and chronic stage. Macular VD in the SVP, particularly in the nasal perifoveal region, may serve as an early indicator of visual function decline, highlighting its potential clinical value for monitoring disease progression and evaluating treatment strategies.

Background: After vestibular impairment, the body undergoes static and dynamic compensation. Static compensation is characterized by the resolution of spontaneous nystagmus, indicating rebalanced bilateral vestibular nuclei activity. It remains unclear whether ascending vestibular projections remain abnormal thereafter. This study investigated abnormal functional activity along vestibular nucleus-thalamus-cortex projection pathways in chronic unilateral vestibulopathy (CUVP) after static compensation using degree centrality (DC) and functional connectivity (FC) analyses.

Methods: In total, 25 CUVP patients and 25 age- and sex-matched healthy controls underwent resting-state functional magnetic resonance imaging (fMRI). The DC analysis identified abnormal functional activity in vestibular nucleus-thalamus-cortex pathways, particularly in the subcortical structures. The seed-based FC analyses used eight seeds (the bilateral vestibular nuclei, and the bilateral pulvinar, mediodorsal, and ventrolateral ventral thalamic regions) with 10,000 non-parametric permutations and a cluster-level family-wise error rate (FWER)-corrected threshold of P<0.05. The region-of-interest (ROI)-based FC analyses examined connections among the vestibular nucleus, thalamic subregions (the pulvinar and ventrolateral ventral thalamus), and multisensory vestibular/sensorimotor/visual cortices with a false discovery rate (FDR)-corrected threshold of P<0.05 to confirm the pathway abnormalities. A regression analysis assessed the relationships between the altered brain metrics and Dizziness Handicap Inventory (DHI) scores.

Results: Compared with the healthy controls, the CUVP patients showed reduced DC in the bilateral thalamus (ventrolateral ventral thalamus, and pulvinar), cerebellum, precuneus, postcentral gyrus, and premotor areas (all FDR-corrected P<0.05). Using the vestibular nucleus, pulvinar, and ventrolateral ventral thalamus as the seed regions, similar bilateral patterns of FC change were observed, with notably reduced FC between the pulvinar and visual cortex, as well as between the ventrolateral ventral thalamus and sensorimotor cortex (all FWER-corrected P<0.05). The ROI-based FC analyses confirmed abnormalities along the vestibular nucleus-ventrolateral ventral thalamus/pulvinar-multisensory vestibular/sensorimotor/visual cortices pathways (all FDR-corrected P<0.05). The regression analysis revealed negative associations between thalamic DC/FC changes and DHI scores (all FDR-corrected P<0.05).

Conclusions: Patients with CUVP exhibit persistent abnormalities in the vestibular nucleus-thalamus-cortex pathways even after the establishment of static compensation. The ventrolateral ventral thalamus and pulvinar may serve as key nodes in these abnormalities.

Background: Ultrasound-based assessment of subclinical atherosclerosis improves cardiovascular (CV) risk prediction, but methods to quantify plaque burden vary. We developed a novel method to measure plaque burden entitled weighted total plaque thickness (wTPT) and evaluated its association with CV risk factors and estimated CV risk.

Methods: A total of 5,180 participants (age 45-74 years; 54.2% women) from the Prospective Screening Of Non-invasive Atherosclerosis Risk study, all self-referred for a health check, underwent carotid and femoral ultrasound. Plaque burden was quantified using wTPT, maximal plaque thickness (MPT), plaque count, and number of arteries with plaque. CV risk was estimated using the standardized NORRISK-2 equation.

Results: Prevalence of carotid, femoral, and any plaque was 90.2%, 79.7%, and 96.1%, respectively. Femoral wTPT was significantly higher than carotid wTPT (P<0.001). In multivariable analysis, all CV risk factors except abdominal obesity were independently associated with wTPT. Spearman's correlation between wTPT and CV risk was 0.66 [95% confidence interval (CI): 0.66-0.68], exceeding that of MPT [0.61 (95% CI: 0.59-0.63)], plaque count [0.62 (95% CI: 0.59-0.65)], and number of arteries with plaque [0.59 (95% CI: 0.58-0.62)]. Femoral wTPT showed a stronger correlation with CV risk [0.59 (95% CI: 0.57-0.61)] than carotid wTPT [0.55 (95% CI: 0.53-0.57)].

Conclusions: In this asymptomatic cohort, plaque was present in nearly all individuals, underscoring the importance of stratifying plaque burden rather than prevalence. wTPT demonstrated a stronger association with estimated CV risk than conventional plaque measures, with femoral burden more closely linked to risk than carotid burden.

Background: Alzheimer's disease (AD) is a progressive neurodegenerative disorder characterized by cognitive decline; this decline is closely linked to hippocampal morphological changes observed in structural magnetic resonance imaging (MRI). However, the existing AD prediction models have not fully explored the spatiotemporal correlation of hippocampal morphological features. To address this limitation, this study aims to develop a longitudinal prediction framework that captures both the temporal evolution and spatial distribution of hippocampal morphological alterations.

Methods: In this paper, we propose a novel deep learning framework for predicting the clinical progression of AD, which consists of a multi-view feature fusion convolutional network (M-FCN) and a bidirectional gated recurrent unit (Bi-GRU). The proposed M-FCN is based on the three-dimensional (3D) topological structure features of the hippocampus that introduces thickness features and heat kernel signature (HKS) to encode hippocampal morphological atrophy features. We utilize these features to construct a deep 3D hippocampus features description system for capturing the micro and macro structural changes of hippocampus. Hence, the task driven attention mechanism for prediction can effectively identify significant morphological changes caused by AD. The Bi-GRU module identifies inter sequence patterns and studies the temporal correlation between longitudinal features of hippocampus.

Results: The proposed method was evaluated using longitudinal T1-weighted MRI data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) (n=221). Compared with the existing AD prediction models, the correspondence between AD-related structural changes and clinical neurodegeneration indicators can be more accurately captured by our proposed deep learning model. The predictive performance was evaluated using root mean square error (RMSE), correlation coefficient (CC), and 95% confidence interval (CI). For the prediction of Mini-Mental State Examination (MMSE) scores, the model achieved a RMSE of 2.34 (95% CI: 2.27-2.45, CC =0.72) at M18, 2.58 (95% CI: 2.52-2.64, CC =0.77) at M24, and 2.60 (95% CI: 2.54-2.66, CC =0.83) at M36.

Conclusions: These results highlight the effectiveness of the proposed model in leveraging the spatiotemporal correlation of hippocampal morphology to provide high accuracy and reliable predictions.

Background: Although standard cardiac magnetic resonance (CMR) findings are typically normal in patients with idiopathic premature ventricular contractions (PVCs), it remains unclear whether a comprehensive protocol incorporating left ventricular (LV) strain and tissue characterization can detect subclinical myocardial abnormalities. This study aimed to determine if advanced CMR parameters can reveal a burden-dependent myocardial phenotype in patients with idiopathic PVCs.

Methods: In this retrospective study, 72 patients with idiopathic PVCs and 25 controls underwent comprehensive 3T CMR with feature-tracking (FT) strain analysis and native T1/T2 mapping. Patients were stratified by 24-hour PVC burden into three groups: <500 PVC/24 h (n=24), 500-10,000 PVC/24 h (n=23), and >10,000 PVC/24 h (n=25).

Results: Compared to controls, patients with >10,000 PVC/24 h (high burden) exhibited significantly impaired myocardial strain after multiple-comparison correction, including reductions in two-dimensional (2D) global radial (27.4%±7.0% vs. 33.9%±6.2%; P=0.005), 2D global circumferential (-16.7%±2.8% vs. -19.2%±2.1%; P=0.001), three-dimensional (3D) global circumferential (-17.5%±3.7% vs. -20.6%±2.0%; P=0.003), 2D apical radial (28.1%±11.0% vs. 40.2%±12.2%; P=0.002), and 2D apical circumferential strain (-16.7%±4.4% vs. -20.8%±3.8%; P=0.001). Notably, the most pronounced strain impairment was observed in the apical segments. These strain parameters were also significantly impaired in the high-burden compared to the low-burden (<500 PVC/24 h) group (all P values <0.05). Moreover, the high-burden demonstrated elevated global native T1 (1,344.5±40.0 vs. 1,289.4±16.6 ms in controls; P<0.001) and T2 (54.4±3.7 vs. 49.4±3.1 ms in controls; P<0.001) values, with the most pronounced alterations localized to the basal segments (T1 and T2; P<0.001 for comparison with controls). A significant dose-response relationship was confirmed for both strain and mapping parameters across the burden groups.

Conclusions: As a hypothesis-generating study, our work, based on comprehensive CMR assessment, identified a novel, burden-dependent myocardial phenotype in patients with idiopathic PVC. The discovery of this phenotype, characterized by distinct patterns of systolic dysfunction and interstitial alterations, should be examined in terms of its clinical and prognostic significance in subsequent longitudinal studies.

Background: Coronary atherosclerosis is the primary pathological basis of coronary heart disease, and patients with diabetes face an elevated risk of cardiovascular events. The value of plaque characteristics and derived parameters based on coronary computed tomography angiography (CCTA) in predicting major adverse cardiovascular events (MACEs) may differ between patients with and without diabetes. This study aimed to compare the predicative value of plaque features and computed tomography (CT)-derived parameters from CCTA in forecasting MACEs in between patients with and without diabetes, providing a more accurate reference for clinical management.

Methods: A total of 472 patients with coronary artery disease were retrospectively enrolled, including 132 patients with diabetes and 340 patients without diabetes. Clinical and imaging data were collected, and follow-up was conducted. Multivariate Cox proportional hazards regression was used to identify risk factors, while receiver operating characteristic (ROC) curve analysis assessed the predictive value for MACEs.

Results: In the diabetic group, the independent predictors of MACEs were coronary artery calcium score (CACS) ≥100 [hazard ratio (HR) =1.98; 95% confidence interval (CI): 1.06-3.72; P=0.033] and the presence of low-attenuation plaque (LAP) (HR =2.13; 95% CI: 1.00-4.53; P=0.049). The combination of CACS and LAP predicted 1-, 3-, and 5-year MACEs, with areas under the curve (AUCs) of 0.649, 0.603, and 0.668, respectively. In the nondiabetic group, positive remodeling (PR) was a strong predictor of MACEs (HR =45.00; 95% CI: 22.69-89.28; P<0.001), with AUCs of 0.792, 0.884, and 0.795 at 1, 3, and 5 years, respectively. Significant differences in baseline characteristics such as hypertension, CT-derived fractional flow reserve ≤0.8, and Coronary Artery Disease Reporting and Data System score ≥3 were observed between the two groups (P<0.05).

Conclusions: The value of plaque characteristics and CT-derived parameters in predicting MACEs varies between patients with and without diabetes. The combination of CACS and LAP is effective in assessing MACE risk in patients with diabetes, whereas PR is a stronger predictor of MACEs in patients without diabetes.

Background: Quantitative susceptibility mapping (QSM) has great advantages in evaluating tissue susceptibility across diverse cerebral conditions. However, conventional reconstruction methods are often affected by streaking artifacts and noise amplification, while purely data-driven deep learning approaches frequently lack physical constraints, resulting in deviations from the underlying dipole physics. To address these issues, we aimed to develop a model-driven deep learning approach that explicitly enforces dipole model data fidelity within the network, aiming to enhance quantitative accuracy and suppress artifacts.

Methods: We propose a nonlinear susceptibility inversion deep learning model (NSIDL), which integrates a nonlinear susceptibility inversion (NSI) model into a convolutional neural network and employs the proximal gradient descent (PGD) method to solve the optimization problem. The method was trained and validated using a multi-orientation gradient echo magnetic resonance imaging (MRI) dataset. Quantitative performance of NSIDL was evaluated using the reconstruction challenge (RC-1 and RC-2) datasets and in vivo data, and compared with state-of-the-art iterative methods and deep learning approaches. Clinical feasibility was assessed in patients with hemorrhage, calcification, and multiple sclerosis (MS).

Results: The quantitative evaluation results showed that NSIDL achieved the highest quantitative accuracy on multi-orientation test datasets, with a fitted slope of 0.716 and an R² of 0.6140, significantly outperforming other competing methods (slope range, 0.511-0.676; R² range, 0.3677-0.5714). On the RC-1 dataset, NSIDL demonstrated superior image fidelity, exhibiting the lowest normalized root mean square error (NRMSE) (63.267±0.575) and high-frequency error norm (HFEN) (58.300±0.668), and the highest peak signal-to-noise ratio (PSNR) (42.39±0.311), consistently outperforming other methods across all three metrics (P<0.05). Analysis of deep gray matter regions of interest confirmed that NSIDL estimates most closely matched the susceptibility labels. Clinical evaluation indicated that, compared with baseline reconstructions, NSIDL effectively suppressed artifacts in hemorrhagic lesions and enhanced the clarity of small MS lesions.

Conclusions: By combining a nonlinear physical model with data-driven regularization, NSIDL substantially improves quantitative susceptibility estimation and image quality. This method demonstrates robust artifact suppression and high-fidelity measurements, showing great potential for precise clinical QSM applications.

Background: Central precocious puberty (CPP) results from premature activation of the hypothalamic-pituitary-gonadal (HPG) axis. Although the gonadotropin-releasing hormone (GnRH) stimulation test remains the diagnostic standard for evaluating HPG axis activation, its invasive nature limits clinical utility. Magnetic resonance imaging (MRI)-derived pituitary measurements offer a promising alternative, yet previous studies on two-dimensional measurements have reported limited accuracy. This study aimed to assess the value of adenohypophysis volume (aPV) and height (aPH) precisely measured with the three-dimensional (3D) CUBE T1 sequence (GE HealthCare) in diagnosing HPG axis activation.

Methods: A cohort of 593 children (196 boys and 397 girls; mean age 8.22±2.28 years) who underwent pituitary MRI and GnRH stimulation testing was included. Partial correlation analysis, controlling for sex, age, height, weight, and body mass index (BMI), examined the associations of aPV and aPH with peak luteinizing hormone (LH) and the LH to follicle-stimulating hormone (FSH) ratio (LH/FSH). Multiple linear regression models were constructed, and their diagnostic performance was evaluated via receiver operating characteristic (ROC) analysis.

Results: aPV showed moderate significant correlations with peak LH (r=0.543; P<0.001) and LH/FSH ratio (r=0.480; P<0.001), which remained significant after controlling for confounders (LH: r=0.283, P<0.001; LH/FSH: r=0.207, P<0.001); meanwhile, the aPH correlations were weaker. Multiple linear regression identified age and aPV as significant predictors for LH peak (aPV: B=0.030; P<0.001), while age, weight, and aPV were significant predictors for LH/FSH (aPV: B=0.002; P<0.001). The regression model for predicting LH/FSH >0.6 yielded an area under the curve (AUC) of 0.841 [95% confidence interval (CI): 0.808-0.871], with a sensitivity of 74.5% and a specificity of 84.9% at the optimal threshold.

Conclusions: aPV may have significant potential to be a noninvasive diagnostic biomarker for assessing HPG axis activation in children with CPP. The developed regression models, incorporating aPV, age, and weight, provide promising diagnostic performance and may potentially reduce the reliance on invasive GnRH stimulation tests.

Background: Intravertebral cleft (IVC) is a linear intravertebral imaging characteristic in fractured vertebrae filled with liquid or gas, once considered specific to Kummell's disease but now also observed in acute traumatic fractures, and its risk factors and relationship with bone cement leakage after percutaneous vertebroplasty (PVP)/percutaneous kyphoplasty (PKP) remain to be explored. The study aims to identify IVC risk factors in acute vertebral fractures (AVFs) patients and assess its link to bone cement leakage post-PVP/PKP.

Methods: A total of 1,612 patients diagnosed with AVF at two medical centres were retrospectively included. The diagnostic criterion for IVC was the presence of linear intravertebral lesions filled with liquid or gas in computed tomography (CT) and magnetic resonance imaging (MRI). A total of 126 patients (127 vertebrae) were diagnosed with intravertebral fissures and included in the IVC group, while 125 age- and sex-matched patients (134 vertebrae) without IVC were included in the control group. Clinical and imaging characteristics, including demography, osteoporosis severity, and vertebral fracture line involvement, were compared between the two groups. In patients who underwent PKP or PVP within three months, the incidence and types of bone cement leakage were analysed.

Results: The compression severity and vertebral fracture line involvement of the inferior endplate, posterior wall, and basivertebral foramen significantly differed between the IVC and control groups (P<0.05). Vertebral fracture line involvement of the basivertebral foramen was a risk factor for IVC [95% confidence interval (CI): 2.297 (1.303-4.048), P=0.004]. An interaction effect between vertebral fracture line involvement of the basivertebral foramen and the posterior wall was observed (P<0.05). Within 3 months, 64 vertebrae underwent PKP or PVP in the IVC group, and 15 cases (23.438%) exhibited leakage. Sixty-one vertebrae in the control group underwent PKP or PVP, with 6 cases (10.000%) exhibiting leakage. There was a significant difference between the two groups (P=0.042). C-type leakage was the most common type in the IVC group, accounting for 86.7% (13/15) of all cases.

Conclusions: A basivertebral foramen fracture is an IVC risk factor for AVF. AVF with IVC significantly increases bone cement leakage risk after PVP/PKP.