http://med-sci.cn/cgzcx/en/contents_list.asp?issue=202610 zh-cn An RSS feed for Chinese Journal of Magnetic Resonance Imaging http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.001 ObjectiveUsing voxel-based morphometry (VBM), this study investigates the characteristics of whole-brain grey matter volume (GMV) changes in children with attention deficit hyperactivity disorder (ADHD) who also have autism spectrum disorder (ASD), to analyse the correlation between grey matter volume in regions of difference and serum vitamin D levels, and to construct a combined diagnostic model to assess its diagnostic value for ADHD co-occurring with ASD. Materials and MethodsThe study included 52 children with ADHD and co-occurring ASD (ADHD+ASD group), 59 children with ADHD (ADHD group), 54 children with ASD (ASD group) and 49 typically developing children [healthy control (HC) group]. High-resolution 3D-T1WI structural images were acquired, and VBM analysis was performed using the CAT12 toolkit in SPM12. Differences in whole-brain GMV across the four groups were compared, with age, sex and total intracranial volume (TIV) as covariates. GMV values from the regions of interest were extracted, and a partial correlation analysis was conducted to investigate their relationship with serum vitamin D levels and clinical scale scores. After standardising all continuous independent variables using Z-scores, Firth-penalised logistic regression was employed to identify independent predictors, and receiver operating characteristic (ROC) curves were constructed to assess diagnostic performance. ResultsCompared with the ADHD group, the ADHD + ASD group showed significantly increased GMV in the right cerebellar CrusⅠ, right fusiform gyrus/right lingual gyrus, right inferior parietal lobule and right precuneus/middle cingulate gyrus (Uncorrected at the voxel level: P < 0.001; cluster-level GRF-FWE correction: P < 0.05); the regions of increased GMV in the ADHD + ASD versus ADHD comparison largely overlapped with those in the ASD versus ADHD comparison, suggesting that the brain structural pattern of the comorbid group more closely resembles ASD. No significant differences in GMV were found between the ADHD + ASD group and the ASD group. Partial correlation analysis revealed that GMV values in the right inferior parietal lobule of the ADHD + ASD group were negatively correlated with serum vitamin D levels (r = -0.648, P_FDR = 0.004); GMV values in multiple regions of difference in the ADHD group were positively correlated with vitamin D levels, whilst GMV values in multiple brain regions of the ASD group were negatively correlated with vitamin D levels. Firth-penalised logistic regression analysis revealed that serum vitamin D levels (OR = 1.523, 95% CI: 1.302 to 1.868, P < 0.001) and right inferior parietal lobule GMV values (OR = 18.532, 95% CI: 6.332 to 77.069, P < 0.001) were independent predictors of ADHD + ASD (OR values represent the change in risk per one standard deviation increase). The area under the ROC curve (AUC) for the combined model was 0.921 (95% CI: 0.870 to 0.971), with a sensitivity of 90.4% and a specificity of 83.1%, significantly superior to those of the individual indicators. ConclusionsNo significant differences were found in the patterns of grey matter volume enlargement between children with ADHD and ASD compared to those with ASD alone, suggesting that the structural changes in the brains of children with comorbid conditions are more similar to those seen in ASD than in ADHD. The combination of GMV values in the right inferior parietal lobule and serum vitamin D levels has some auxiliary screening value in distinguishing whether children with a confirmed diagnosis of ADHD also have ASD; it may also provide some reference value for research into the brain development and neuropathological mechanisms of children with comorbid conditions, as well as for early clinical identification. ]]> Tue,20 Oct 2026 00:00:00 GMT http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.002 ObjectiveTo investigate the alteration patterns of gray matter volume (GMV) in patients with Alzheimer^,s disease (AD) using voxel-based morphometry (VBM), and to identify key genes and biological pathways associated with these structural brain changes. Materials and MethodsA total of 63 subjects from the Alzheimer^,s Disease Neuroimaging Initiative (ADNI) database were enrolled and divided into an Alzheimer^,s disease (AD) group (n = 32) and a healthy control (HC) group (n = 31). The whole brain GMV of the two groups of subjects was calculated by the VBM method, and a stepwise double-correction strategy was adopted. First, perform multiple comparison correction of the voxel level family wise error (FWE) (P < 0.05 after correction) to initially screen the differential voxels, and then set the minimum mass threshold k > 100 voxels. Then, a secondary correction of the false discovery rate (FDR) at the cluster-level was carried out on the continuous voxel clusters obtained through the initial screening. Only the continuous voxel clusters that simultaneously meet the conditions of significant voxel FWE and significant cluster-FDR are defined as brain regions with statistically significant GMV differences. GMV values from identified brain regions were extracted using CAT12 software, and t-values for inter-group differences were obtained via independent samples t-tests. These t-values were then correlated with the Allen Human Brain Atlas using partial least squares (PLS) regression analysis. A total of 15 633 genes were ranked based on their PLS weights. The top 500 positively weighted and top 500 negatively weighted genes were selected for subsequent gene enrichment analysis (FDR correction, P < 0.05), and the top 20 pathways with the highest enrichment significance were further analyzed. ResultsThe AD group exhibited smaller total GMV compared to the HC group (t = 2.405, P < 0.05). Specifically, GMV was reduced in the bilateral hippocampi, right inferior temporal gyrus, and right middle temporal gyrus in the AD group compared to the HC group (peak t value = 5.415 to 8.222, P < 0.05). The first PLS component (PLS1) accounted for 37% of the image-gene association variations and showed positive correlation with the t-values of GMV differences between the two groups (r = 0.61, P = 0.03). The top 20 positively weighted genes from PLS1 were enriched in pathways related to synaptic transmission, neurodevelopment, and neuronal projection. Conversely, the top 20 negatively weighted genes were enriched in pathways related to immune inflammation, Rho GTPase signaling, endocytosis, and apoptosis. The two types of weighting genes were respectively enriched in the biological pathways related to neural structure maintenance and neurodegenerative injury. ConclusionsThe characteristic gray matter atrophy in AD is closely associated with pathways governing synaptic function, neurodevelopment, immune regulation, and cell death. This integrative approach provides insights into the imaging-genetics mechanisms underlying AD, offering candidate evidence for mechanistic research. ]]> Tue,20 Oct 2026 00:00:00 GMT http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.003 ObjectiveTo investigate the correlations between brain glymphatic system functional MRI metrics including diffusion tensor imaging-analysis along the perivascular space (DTI-ALPS), free water (FW), and choroid plexus volume with cognitive function and Alzheimer^,s disease (AD) pathological biomarkers in elderly patients with hypertension, and to analyze the mediating role of glymphatic dysfunction in hypertension-related cognitive impairment. Materials and MethodsThis study was based on the Alzheimer^,s Disease Neuroimaging Initiative (ADNI) database and included 200 elderly participants (aged 60 to 90 years), comprising 94 hypertensive patients and 106 healthy controls. Hypertension was defined based on medical history and medication records [blood pressure ≥ 140/90 mmHg (1 mmHg = 0.133 kPa) or current use of antihypertensive medications]. All participants underwent 3.0 T MRI scanning, including structural imaging and diffusion tensor imaging. Clinical data, AD pathological biomarkers including β-amyloid (Aβ) and tau protein, as well as cognitive function assessments were collected. FW imaging was employed to calculate the whole-brain white matter free water index. Diffusion-related metrics were derived from diffusion tensor imaging data to calculate the DTI-ALPS index. Choroid plexus volume and white matter hyperintensity volume were obtained via segmentation of high-resolution structural images. Multiple linear regression was used to examine the relationship between hypertension and glymphatic metrics, and a mediation model was constructed to explore the mediating role of FW in the associations between hypertension and cognitive function as well as white matter injury. ResultsCompared with the control group, the hypertension group showed significantly higher FW fraction (P = 0.002), and hypertension was an independent predictor of elevated FW fraction (β = 0.010, P = 0.019). The hypertension group also exhibited significantly increased white matter hyperintensity volume and decreased executive function scores (P < 0.05). Mediation analysis revealed that FW fraction had a statistically significant mediating effect on the relationship between hypertension and white matter hyperintensities (β = 0.70, 95% CI: 0.22 to 1.68) and also significantly mediated the relationship between hypertension and executive function (β = -0.06, 95% CI: -0.14 to -0.01). However, no significant differences were observed between the two groups in the DTI-ALPS index, choroid plexus volume or AD pathological biomarkers (Aβ/tau). ConclusionsFW fraction may serve as a potential imaging marker for cognitive impairment in elderly hypertensive patients, and improving glymphatic function may provide a new direction for intervention in hypertension-related cognitive impairment. ]]> Tue,20 Oct 2026 00:00:00 GMT http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.004 ObjectiveTo explore the correlation between atrophy of specific subfields of the hippocampus and amygdala and cognitive impairment in patients with cerebral small vessel disease (CSVD) using artificial intelligence (AI)-based brain region segmentation technology combined with high-resolution 3D T1-weighted MRI, and to identify the relevant imaging markers. Materials and MethodsA total of 76 patients diagnosed with CSVD at Shanxi Cardiovascular Hospital from February 2025 to October 2025 were consecutively enrolled, including 38 cases of cerebral small vessel disease with cognitive impairment (CSVD-CI) and 38 cases of cerebral small vessel disease without cognitive impairment (CSVD-nCI). Thirty-eight normal controls (NC) were also enrolled. The cognitive function of all participants was evaluated using the Montreal Cognitive Assessment (MoCA) and Mini-Mental State Examination (MMSE) scales. All subjects underwent conventional MRI scans and 3D-T1 MRI thin-slice scans. Brain segmentation was performed using uAI (20250730-release). ANCOVA was applied to compare brain regional and white matter hyperintensities (WMH) volumes across the three groups. Main effects were corrected using FDR; significant main effects (P < 0.05) were followed by Bonferroni-adjusted pairwise comparisons (α = 0.05 / 3 ≈ 0.0167). After correcting for FDR, a partial correlation analysis was conducted between the volumes of the three groups of brain regions with differences and the WMH volume, respectively, and the MoCA and MMSE scores. Subsequently, a multiple linear regression analysis was performed. For factors that showed statistically differences in the multiple linear regression analysis, receiver operating characteristic (ROC) curves were calculated, and the area under the curve (AUC) was used to evaluate diagnostic performance. To further evaluate the predictive calibration performance and incremental clinical benefit of the combined model, we performed calibration curves and decision curve analysis (DCA), respectively. ResultsThe significant results of post-hoc comparisons are presented below. Compared with the CSVD-nCI group, the volumes of the right hippocampal CA3 region and the right amygdalar cortical nucleus were decreased in the CSVD-CI group (all P < 0.05). Compared with the HC group, the CSVD-CI group exhibited reduced volumes in the right hippocampal molecular layer, right hippocampal CA3 region, right amygdalar cortical nucleus, right anterior amygdala, right accessory basal amygdala, and right amygdalocortical transition zone (all P < 0.05). The volume of the right amygdalar cortical nucleus and WMH volume remained statistically significant after FDR correction. Partial correlation analysis showed that reduced volume of the right amygdalar cortical nucleus (r = 0.346, P < 0.001; r = 0.330, P < 0.001) and increased WMH volume (r = -0.737, P < 0.001; r = -0.684, P < 0.001) were associated with decreased MoCA and MMSE scores in patients with CSVD. Multi-factor linear regression analysis revealed that after adjusting for confounding variables, the volume of the right amygdalar cortical nucleus was independently positively correlated with the MoCA score (B = 14.78, 95% CI: 6.33 to 23.20, t = 3.47, P < 0.001). It was also independently positively correlated with the MMSE score (B = 10.14, 95% CI: 3.62 to 16.67, t = 3.08, P = 0.003). The total WMH volume was independently negatively correlated with the MoCA score (B = -0.93, 95% CI: -1.10 to -0.75, t = -10.53, P < 0.001). It was also independently negatively correlated with the MMSE score (B = -0.61, 95% CI: -0.74 to -0.47, t = -8.94, P < 0.001). ROC curve analyses based on residual-adjusted volumes revealed that the AUC of the right amygdalar cortical nucleus for discriminating CSVD-CI from CSVD-nCI was 0.662 (95% CI: 0.539 to 0.785), with sensitivity of 29% and specificity of 97%. For raw volume of the right amygdalar cortical nucleus, the AUC, sensitivity and specificity for differentiating CSVD-CI and CSVD-nCI were 0.664 (95% CI: 0.541 to 0.788), 66% and 74%, respectively. The raw WMH volume, the AUC for discriminating CSVD-CI from CSVD-nCI was 0.855 (95% CI: 0.763 to 0.946), with sensitivity of 74% and specificity of 92%. A joint predictive model combining the volume of right amygdalar cortical nucleus and WMH achieved an AUC of 0.884 (95% CI: 0.800 to 0.967); the sensitivity of this combined model increased to 79% while the specificity stayed at 92%. The calibration curves revealed Brier scores of 0.150, 0.142, and 0.230 for the WMH volume model, combined model, and right amygdalar cortical nucleus volume model, respectively. Decision curve analysis (DCA) demonstrated that the combined model yielded higher net benefit than the WMH-only model across the low-to-moderate threshold probability range. ConclusionsPatients with CSVD-CI exhibit atrophy of specific subfields of the hippocampus and amygdala. AI-based automatic subfield segmentation combined with high-resolution 3D T1-weighted MRI can extract neuroimaging biomarkers for CSVD-CI. Clinically, locating and quantifying these biomarkers can help identify high-risk populations and facilitate early personalized interventions. ]]> Tue,20 Oct 2026 00:00:00 GMT http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.005 ObjectiveUsing structural covariance networks (SCN) based on gray matter volume as the metric, and employing network attack analysis in conjunction with degree distribution analysis, we systematically investigated the topological robustness and degree distribution characteristics of brain SCN in patients with frontotemporal dementia (FTD). Materials and MethodsAll data analyzed were obtained from the Neuroimaging in Frontotemporal Dementia (NIFD) database, which included 92 patients with FTD and 89 age- and sex-matched healthy controls (HC). Clinical information and T1-weighted magnetic resonance images were collected for all participants. Image preprocessing was performed using the SPM8 software package, and a SCN was constructed based on the automated anatomical labeling (AAL) atlas. The topological robustness of the networks under targeted and random attack strategies was assessed using the graph analysis toolbox (GAT), and degree distribution fitting was performed. ResultsIn terms of cognitive function assessment, the FTD and HC groups showed statistically significant differences in scores on the Mini-Mental State Examination (Z = -9.470, P < 0.001) and the Clinical Dementia Rating scale(Z = 12.541, P < 0.001), whereas no statistically significant differences were found in baseline variables of age (Z = 1.609, P = 0.108), sex distribution(χ² = 0.533, P = 0.465), and total intracranial volume(Z = -0.297, P = 0.767). Under targeted attack, the difference in the area under the curve (AUC) between groups was not statistically significant (P = 0.148). Point-by-point analysis revealed that only at certain stages was the network relative size in the FTD group lower than that in the HC group (uncorrected P < 0.05; after false discovery rate correction, q > 0.05). Under the random attack scenario, the difference in AUC between groups similarly did not reach statistical significance (P = 0.306). Finally, degree distribution fitting based on the group-level aggregate network yielded the following parameters: for the HC group, a = 1.251 7, b = 11.560 0 (r² = 0.991 1); for the FTD group, a = 1.198 0, b = 17.038 1 (r² = 0.967 8), where a is the power-law exponent and b is the truncation value. The differences in parameters between the two groups are presented descriptively only, as no between-group statistical inference was performed. ConclusionIn this study, we attempted to integrate network attack and degree distribution analyses for the investigation of SCN in FTD. At the group-level aggregated network level, our exploratory analysis revealed a trend of the degree distribution shifting toward a higher truncation degree, suggesting a potential enhancement of topological heterogeneity, whereas overall network robustness showed no significant alterations. these preliminary observations are mainly valuable in demonstrating a multi-dimensional analytical strategy, definitive conclusions warrant further validation in larger cohorts and at the individual network level. ]]> Tue,20 Oct 2026 00:00:00 GMT http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.006 ObjectiveTo explore the value of amide proton transfer (APT) imaging, quantitative susceptibility mapping (QSM) and ultra-high b-values diffusion-weighted imaging (DWI_uh) in the clinical diagnosis of Alzheimer^,s disease (AD). Materials and MethodsA total of 61 participants were enrolled in this study, including 17 in the AD group, 20 in the mild cognitive impairment (MCI) group, and 24 in the normal control (NC) group. Analysis of covariance (ANCOVA) was used to compare the differences in magnetization transfer rate asymmetry (MTR_asym), ultra-high b-value apparent diffusion coefficient (ADC_uh), and QSM values among the three groups. Partial correlation analysis was performed to assess the relationships between the imaging parameters, as well as between each parameter and the Mini-Mental State Examination (MMSE) score. Receiver operating characteristic (ROC) curve analysis was applied to assess the diagnostic performance of MTR_asym, QSM and ADC_uh, using the area under the curve (AUC), sensitivity and specificity as evaluation metrics. The parameter with the highest AUC among the three was selected for binary multivariable logistic regression analysis to compare the diagnostic efficacy of the combined three‑parameter model versus the single‑parameter models. ResultsCompared with the NC and MCI groups, the AD group showed increased MTR_asym, QSM, and ADC_uh values in multiple brain regions (all P < 0.05). The MTR_asym, QSM, and ADC_uh values in multiple brain regions were negatively correlated with MMSE scores. Specifically, MTR_asym values exhibited significant negative correlations in the bilateral frontal white matter, bilateral temporal white matter, bilateral substantia nigra, bilateral amygdala, bilateral hippocampus, right putamen, and right thalamus, with more prominent correlations observed in the right frontal white matter (r = -0.452, P < 0.001), right substantia nigra (r = -0.435, P < 0.001), and right putamen (r = -0.480, P < 0.001). QSM values showed significant negative correlations in the bilateral substantia nigra, bilateral amygdala, bilateral putamen, bilateral thalamus, and bilateral hippocampus. ADC_uh values were negatively correlated in the bilateral hippocampus, left temporal white matter, and left thalamus, with the most significant correlations found in the right hippocampus (r = -0.486, P < 0.001) and left hippocampus (r = -0.512, P < 0.001). ROC curve analysis demonstrated that MTR_asym and QSM parameters could effectively distinguish NC from MCI, NC from AD, and MCI from AD. ADC_uh only differentiated NC from AD and MCI from AD, with no statistically significant difference between the NC and MCI groups. In the differential diagnosis between NC and AD, the QSM value in the left amygdala yielded the best diagnostic performance (AUC = 0.987). Three parameters including MTR_asym of the right frontal white matter, QSM of the left amygdala and ADC_uh of the right hippocampus were incorporated to establish a binary logistic regression combined diagnostic model for differentiating NC from AD. The combined model yielded an AUC of 0.988 (95% CI: 0.961 to 1.000, P < 0.001), with an optimal cut-off value of 0.441 6, a sensitivity of 100.0% and a specificity of 95.8%. Its AUC was close to that of the single left amygdala QSM model (AUC = 0.987), indicating no statistically significant difference in diagnostic performance between the two models. ConclusionsMeasurements of MTR_asym, QSM, and ADC_uh in specific brain regions facilitate the differential diagnosis of AD, with QSM values of the left amygdala exhibiting optimal diagnostic efficiency. The multi‑parameter combined model achieves favorable diagnostic performance for AD, whereas its efficacy is comparable to that of the optimal single‑parameter model without statistically significant difference. Integrating information from multiple pathological dimensions, this model still holds promise as a combined imaging biomarker for AD. The integrated application of multiple quantitative magnetic resonance imaging techniques provides valuable evidence for the clinical auxiliary diagnosis of AD. ]]> Tue,20 Oct 2026 00:00:00 GMT http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.007 ObjectiveTo explore the independent risk factors for cerebral small vessel disease (CSVD)-related cognitive impairment, develop a nomogram prediction model for risk stratification of cognitive impairment in CSVD patients, and evaluate its discriminative ability, calibration performance and clinical decision-making value. Materials and MethodsA total of 108 right-handed patients with CSVD (total imaging burden score ≥ 1) were enrolled in accordance with the 2021 Chinese Expert Consensus on the Diagnosis and Treatment of Cerebral Small Vessel Disease. Brain imaging data were acquired using a 3.0 T MRI scanner, and multidimensional variables including the diffusion tensor image analysis along the perivascular space (DTI-ALPS) index and total imaging burden score were collected. Combined with demographic, clinical, and laboratory parameters, univariate logistic regression was performed to screen potential risk factors, followed by multivariate logistic regression to identify independent risk factors. A nomogram prediction model was established using R software. The model was comprehensively validated in terms of discriminative ability, calibration, and clinical utility via the concordance index (C-index), the area under the receiver operating characteristic (ROC) curve (AUC), calibration curves, and decision curve analysis (DCA). ResultsUnivariate logistic regression analysis demonstrated that homocysteine level, DTI-ALPS index, total imaging burden score, age, sex,systolic blood pressure, low-density lipoprotein cholesterol, triglyceride, serum creatinine, and drinking history were significantly associated with the development of CSVD-related cognitive impairment (all P < 0.05). Multivariate logistic regression further confirmed that homocysteine (OR = 1.507, 95% CI: 1.117 to 2.033, P = 0.007), DTI-ALPS index (OR = 0.008, 95% CI: 0.000 to 0.272, P = 0.007), and total imaging burden score (OR = 19.126, 95% CI: 2.700 to 135.467, P = 0.003) were independent risk factors. The newly constructed nomogram model yielded an AUC of 0.955 (95% CI: 0.920 to 0.990), which was superior to the predictive efficacy of each individual indicator. The calibration curves showed excellent consistency with the ideal curve, and the C-index of the model was 0.955 (95% CI: 0.920 to 0.990), indicating favorable stability and reliability. DCA analysis confirmed that the nomogram has favorable clinical decision-making application value. ConclusionsElevated serum homocysteine, decreased DTI-ALPS index, and increased total imaging burden score are core independent risk factors for CSVD-related cognitive impairment. The established nomogram model integrates laboratory and neuroimaging indicators, exhibiting excellent discrimination, favorable predictive accuracy, and robust clinical effectiveness. This model can rapidly quantify the individual risk of cognitive impairment in CSVD patients, thereby providing a reliable reference for early clinical screening, risk stratification, and individualized intervention. ]]> Tue,20 Oct 2026 00:00:00 GMT http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.008 ObjectiveTo investigate the excitation/inhibition (E/I) balance and brain functional alterations in core nodes of the default mode network (DMN) in patients with Alzheimer^,s disease (AD) and mild cognitive impairment (MCI) using multimodal MRI, and to explore the coupling relationship between neurochemistry and brain function. Materials and MethodsA total of 22 AD patients, 19 MCI patients, and 22 normal control (NC) were prospectively enrolled. Demographic and neuropsychological data were collected, including the Mini-Mental State Examination (MMSE) and Montreal Cognitive Assessment (MoCA). Resting-state functional magnetic resonance imaging (rs-fMRI) data were acquired to calculate the fractional amplitude of low-frequency fluctuation (fALFF) and regional homogeneity (ReHo). Magnetic resonance spectroscopy (MRS) data were acquired from regions of interest (ROI) in the left medial prefrontal cortex (Lm-PFC) and precuneus (PCu) to calculate the levels of glutamate-glutamine complex (Glx) and gamma-aminobutyric acid (GABA) combined signal relative to creatine (Cr) (GABA+), as well as the Glx/GABA+ ratio. It should be noted that the Glx/GABA+ ratio reflects the steady-state concentrations of excitatory and inhibitory metabolites in local brain regions, rather than a direct measurement of dynamic synaptic E/I balance, and serves as an indirect indicator of regional E/I metabolic status in this study. Two-sample t-tests were used for pairwise comparisons of rs-fMRI metrics (GRF correction, voxel-level P < 0.001, cluster-level P < 0.05). MRS data were analyzed using one-way analysis of variance (ANOVA) followed by Bonferroni post-hoc comparisons, including only measurements with fitting error less than 15%. Pearson correlation analysis was performed to examine the relationships between the Glx/GABA+ ratio and fALFF values, as well as cognitive scale (MMSE, MoCA) scores . A P-value < 0.05 was considered statistically significant. ResultsCompared with the NC group, the AD group showed a significantly decreased fALFF value in the left angular gyrus (t = -5.533 4, P < 0.05). No significant differences in ReHo were observed among the three groups (P > 0.05). MRS analysis revealed that the Glx/Cr level and Glx/GABA+ ratio in the Lm-PFC and PCu were significantly higher in the AD group than in the NC and MCI groups (P < 0.001), while no significant difference was found between the NC and MCI groups (P > 0.05). Correlation analysis between whole-brain fALFF and the Glx/GABA+ ratio in the ROI showed that, in the NC group, the Glx/GABA+ ratio in the PCu was significantly negatively correlated with fALFF values in the right superior temporal pole (r = -0.947 1, P < 0.001), left gyrus rectus (r = -0.889 5, P < 0.001), and right cerebellar Crus Ⅲ (r = -0.852 5, P < 0.001). No significant correlations were found in the AD group (P > 0.05). Furthermore, in the NC group, the fALFF values in the right superior temporal pole and left gyrus rectus (which were closely related to cognitive function), as well as the Glx/GABA+ ratios in the ROI in both groups, showed no significant correlations with cognitive scale (MMSE, MoCA) scores (P > 0.05). ConclusionsAlterations in E/I-related metabolite homeostasis, reduced local neural activity, and a loss of the metabolical–functional association observed under physiological conditions exist in core nodes of the DMN in AD patients. The metabolic imbalance reflected by the Glx/GABA+ ratio may serve as a valuable imaging reference indicator for the clinical stage of AD; however, its correlation with cognitive and functional measures was not established in the current study, warranting caution in its interpretation as a proxy for E/I balance. ]]> Tue,20 Oct 2026 00:00:00 GMT http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.009 ObjectiveTo explore the whole brain microstructural changes in the acute phase of mild traumatic brain injury (mTBI) and the value of diffusion tensor imaging (DTI) and diffusion kurtosis imaging (DKI) parameters for assessment. Materials and MethodsThis was a prospective case-control study. From April 2024 to May 2025, 31 patients with mTBI (mTBI group) and 30 healthy volunteers matched for age, sex, and years of education (control group) were recruited. Diagnosis of mTBI was based on a definite history of head trauma, a Glasgow Coma Scale (GCS) score of 13-15, and normal findings on post-injury CT and conventional MRI. All participants underwent DTI and DKI scans within 1-7 days after injury (acute phase). DTI data were preprocessed and metrics were calculated using MRtrix, whereas DKI data were fitted and metrics were calculated using diffusion kurtosis estimation (DKE) software. Whole-brain voxel-wise analysis was conducted using non-parametric permutation testing (Randomise program) in FSL 6.0 to compare intergroup differences, with age, sex, and years of education as covariates. Statistical significance was defined as FDR corrected P < 0.05. Demographic data were analyzed using SPSS 24.0. Intergroup comparisons were performed using the independent-sample t-test, Mann-Whitney U test, or chi-square test, with P < 0.05 considered statistically significant. ResultsNo significant differences in age, sex, or years of education were observed between the two groups (P > 0.05). Compared with the control group, DTI revealed significantly decreased fractional anisotropy (FA) and mean diffusivity (MD) in the bilateral cerebellar lobule Ⅷ and right cerebellar lobule Ⅸ in the mTBI group (t-value range: -6.09 to -5.27, FDR corrected P < 0.05). Increased MD was observed in multiple supratentorial brain regions (including the left anterior thalamic radiation, genu of the corpus callosum, right hippocampus, right middle frontal gyrus, right medial orbitofrontal cortex, right parahippocampal gyrus, left precentral gyrus, splenium of the corpus callosum, right superior fronto-occipital fasciculus, right superior temporal gyrus, and right thalamus) (t-value range: 5.41 to 7.08, FDR corrected P < 0.05). Increased FA was noted in the left external capsule, left fusiform gyrus, and right middle temporal gyrus (t-value range: 5.30 to 5.99, FDR corrected P < 0.05). DKI revealed significantly decreased mean kurtosis (MK), axial kurtosis (AK), and radial kurtosis (RK) in the left cerebellar lobule Ⅷ (t-value range: -8.15 to -8.06, FDR corrected P < 0.05). Reduced MK and AK values were found in the left anterior thalamic radiation, left rectus gyrus, and left inferior fronto-occipital fasciculus (t-value range: -7.31 to -5.42, FDR corrected P < 0.05), whereas RK was increased in the brainstem (t = 5.88, FDR corrected P < 0.05). Decreased AK was observed in the left external capsule (t = -6.81, FDR corrected P < 0.05). Combined DTI and DKI analyses highlighted the most prominent microstructural alterations in cerebellar lobule Ⅷ. ConclusionsmTBI induces widespread group-level microstructural alterations. The cerebellum, particularly cerebellar lobule Ⅷ, may serve as a sensitive target for detecting mTBI-related microstructural changes. Compared with DTI, DKI captures non-Gaussian water diffusion properties, providing more detailed microstructural information, and can serve as an important complementary tool to DTI, which can facilitate the objective evaluation of acute-phase mTBI. ]]> Tue,20 Oct 2026 00:00:00 GMT http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.010 ObjectiveTo evaluate the diagnostic value of quantitative parameters derived from T1-mapping, T2-mapping, and diffusion kurtosis imaging (DKI), as well as their combined models, in differentiating common salivary gland tumor types [pleomorphic adenoma (PA), Warthin tumor (WT), and malignant tumor (MT)], and to explore their optimal diagnostic thresholds. Materials and MethodsThis study was registered in the Medical Research Registration and Filing Information System of the National Health Commission (Filing No. MR-37-26-035352). A retrospective analysis was conducted on 78 patients with pathologically confirmed salivary gland tumors (SGTs) who were treated at Jining First People^,s Hospital between March 2024 and November 2025 and met the eligibility criteria. The patient cohort included 16 cases of MT, 35 cases of PA, and 27 cases of WT. All patients underwent MRI examinations, including T1 mapping, T2 mapping, and DKI sequences. Quantitative parameters [T1, T2, mean diffusivity (MD), and mean kurtosis (MK)] were compared across the three groups, and the discriminatory ability of individual parameters and combined logistic regression models was evaluated using receiver operating characteristic (ROC) curve analysis. ResultsCompared with WT, PA had higher T1, T2, and MD values (t = 4.315, P = 0.002; Z = -6.140, Z = -4.671, respectively; all P < 0.001), but lower MK values (t = -7.850, P < 0.001), the difference was statistically significant. Compared with MT, PA had higher T2 and MD values (Z = -4.304, P < 0.001; Z = -3.167, P = 0.002), while MT had higher MK values than PA (t = -4.754, P < 0.001), the difference was statistically significant. However, the difference in T1 values between PA and MT was not statistically significant (t = -0.116, P = 0.920). Furthermore, the T1 and T2 values of WT were both lower than those of MT (t = -4.651, Z = -3.291, respectively; both P ≤ 0.001), the difference was statistically significant; while the differences in MD and MK values were not statistically significant (Z = -0.892, P = 0.372; t = 1.266, P = 0.213). By combining multiple parameters using logistic regression, the resulting models achieved AUC ranging from 0.732 (95% CI: 0.606 to 0.858) to 0.993 (95% CI: 0.979 to 1.000); among them, the T1_T2_MD_MK combined model demonstrated the best performance in differentiating PA from WT (AUC = 0.993, 95% CI: 0.979 to 1.000). ConclusionsQuantitative parameters derived from T1 mapping, T2 mapping, and DKI show good value in differentiating PA from WT. The combined T1_T2_MD_MK model integrating all three parameters achieved the best performance (AUC = 0.993) in distinguishing PA from WT, which may assist preoperative differentiation between these two types of tumors with markedly different treatment strategies and provide imaging evidence for individualized surgical planning. Findings involving MT are exploratory and warrant further validation. ]]> Tue,20 Oct 2026 00:00:00 GMT http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.011 ObjectiveTo investigate the predictive value of cardiac magnetic resonance (CMR)-derived hemodynamic force (HDF) parameters for major adverse cardiovascular events (MACE) in patients with dilated cardiomyopathy (DCM) at New York Heart Association (NYHA) functional class I or Ⅱ. Materials and MethodsA total of 92 DCM patients with NYHA class I/Ⅱ who underwent CMR examination at the General Hospital of Ningxia Medical University from January 2020 to October 2024 were retrospectively enrolled. Patients were stratified into the MACE group (n = 16) and non-MACE group (n = 76) according to MACE occurrence during follow-up. Image post-processing was performed using Medis software. Clinical baseline characteristics, left ventricular functional parameters, and HDF-derived indices were compared between the two groups. Multivariable Cox regression analysis was performed to identify independent risk factors for MACE. Sensitivity analyses with incremental adjustment for key covariates, including left ventricular ejection fraction (LVEF), left ventricular end-diastolic volume (LVEDV), and late gadolinium enhancement (LGE), were conducted to verify the prognostic robustness of the lateral-septal pressure gradient integral (L-S). Considering the limited number of events, Firth^,s bias-reduced Cox regression was applied to minimize small-sample estimation bias. Internal validation based on 1000 bootstrap resamples was performed to calculate the optimism-corrected Harrell^,s C-index. The Maxstat method was used to determine the optimal cut-off value of L-S for MACE prediction. Kaplan-Meier survival curves were plotted to assess cumulative event-free survival, and the log-rank test was used to compare inter-group survival differences and validate the risk-stratification efficacy of L-S. ResultsThe MACE group had a significantly higher proportion of male patients than the non-MACE group (Fisher^,s exact test, P = 0.002). After adjustment for sex and age, multivariable Cox regression analysis demonstrated that L-S was significantly associated with MACE (HR = 1.726, 95% CI: 1.033 to 2.884, P = 0.037), with a C-index of 0.755. Firth^,s penalized Cox regression further confirmed the stable prognostic effect of L-S (HR = 1.739, P = 0.037), consistent with the standard Cox model. Bootstrap internal validation yielded an optimism-corrected C-index of 0.725 with an optimism of only 0.03, indicating a low risk of overfitting. Sensitivity analyses adjusting sequentially for LVEF, LVEDV, and LGE confirmed no reversal of the effect direction for the prognostic association of L-S (HR range: 1.590 to 1.726). In the male subgroup, the optimal cut-off value of L-S was 1.78. Patients with L-S > 1.78 showed a significantly higher cumulative incidence of MACE compared with those with L-S ≤ 1.78 (log-rank P < 0.05). ConclusionsL-S may serve as a potential imaging biomarker for prognostic evaluation in patients with early-stage DCM. An L-S threshold of 1.78 may help identify high-risk male patients with early-stage DCM, providing a potential imaging reference for individualized risk assessment. However, these findings require further validation in multicenter cohorts. ]]> Tue,20 Oct 2026 00:00:00 GMT http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.012 ObjectiveTo investigate the preoperative value of venous-phase spectral computed tomography (CT) quantitative parameters combined with magnetic resonance imaging (MRI) parameters for predicting Ki-67 expression levels in breast tumors and to compare the predictive performance of single-modality and combined models. Materials and MethodsA total of 104 consecutive patients with suspected malignant breast tumors who underwent venous-phase contrast-enhanced spectral CT and breast MRI examinations (including dynamic contrast-enhanced imaging and diffusion-weighted imaging) before surgery or biopsy between October 2023 and October 2025 were prospectively enrolled. After screening according to the inclusion and exclusion criteria, 100 patients were included in the final analysis. Based on the immunohistochemical Ki‑67 proliferation index, patients were divided into a high‑expression group (≥ 20%, n = 58) and a low‑expression group (< 20%, n = 42). Spectral CT parameters included iodine concentration (IC), normalized iodine concentration (NIC), effective atomic number (Zeff), normalized effective atomic number (nZeff), and spectral curve slope (λHU). MRI features included early enhancement value, apparent diffusion coefficient (ADC), and time–signal intensity curve (TIC) pattern. Variables with significant differences identified by univariate analysis were entered into logistic regression analyses. A spectral CT model incorporating NIC, an MRI model incorporating ADC_mean, and a combined model integrating NIC and ADC_mean were established. Receiver operating characteristic (ROC) curve analysis was performed to evaluate and compare the diagnostic performance of the models using the area under the curve (AUC). ResultsCompared with the low-expression group, the high-expression group showed significantly higher CT_40keV, IC, NIC, Zeff, nZeff, λHU, and MRI lower ADC_mean values (all P < 0.05). The combined model, incorporating NIC and ADC_mean, achieved an AUC of 0.877(95% CI: 0.805 to 0.953), outperforming both the MRI model (AUC = 0.819; 95% CI: 0.725 to 0.913) and the spectral CT model (AUC = 0.713; 95% CI: 0.609 to 0.817). ConclusionsVenous-phase spectral CT quantitative parameters and MRI features provide complementary information for assessing Ki-67 expression in breast tumors. The combined model significantly improves the preoperative prediction of high Ki-67 expression and may serve as a valuable imaging-based tool for clinical risk stratification and treatment decision-making. ]]> Tue,20 Oct 2026 00:00:00 GMT http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.013 ObjectiveTo explore the value of intra-voxel incoherent motion diffusion-weighted imaging (IVIM-DWI) in conjunction with clinical and pathological parameters for predicting 3-year postoperative disease-free survival (DFS) in patients with endometrial carcinoma (EC). Materials and MethodsThis work represents the development and internal validation of a prognostic prediction model. A retrospective analysis was conducted on the clinical and imaging data of 116 patients diagnosed with EC via surgical pathology between 1 Jan 2018 and 1 Jan 2022. Based on the follow-up results, the patients were divided into the remission group (n = 90) and the progression group (n = 26). All patients underwent preoperative standard routine 3.0 T MRI and IVIM-DWI examinations. Differences in parameters such as imaging-based International Federation of Gynecology and Obstetrics (FIGO) staging, IVIM-DWI quantitative parameter values, histological grade, age, maximum tumor diameter, and lymphovascular space invasion (LVSI) between the two groups were compared using independent samples t-test, Mann-Whitney U test, chi-square test, and Fisher^,s exact test. Stratified sampling was used to divide the data into a training set and a validation set at a ratio of 7∶3. Parameters with statistical significance in the univariate analysis were further analyzed by multivariate logistic regression to identify independent risk factors affecting 3-year DFS in patients with EC. A combined prediction model was then constructed based on these factors. The model performance was evaluated through the following methods: (1) The receiver operating characteristic (ROC) curve was adopted to assess the predictive ability of the model; (2) Calibration curves were used to evaluate the model accuracy; (3) Decision curve analysis (DCA) was applied to assess the clinical net benefit of the model for auxiliary decision-making. A two-tailed P < 0.05 was considered statistically significant. ResultsThere were no significant differences in IVIM-DWI quantitative parameters and clinicopathological parameters between the training cohort and validation cohort (P > 0.05). The true diffusion coefficient D value, histological type and LVSI were independent risk factors for 3-year DFS in patients with EC (all P < 0.05). The area under the ROC curve (AUC) of the combined model was 0.972 [95% confidence interval (CI): 0.942 to 1.000] in the training cohort and 0.870 (95% CI: 0.753 to 0.988) in the validation cohort. DeLong test in the training cohort revealed that the discrimination ability of the combined model was superior to that of the single clinicopathological model and the single D value model (all P < 0.05). In the validation cohort, DeLong test showed no significant difference in AUC between the combined model and the single D value model (P = 0.700), while the combined model was significantly better than histological type (P = 0.002) and LVSI (P = 0.003). The calibration curve was close to the ideal diagonal line, and the Hosmer-Lemeshow test yielded P > 0.05, indicating favorable calibration performance of the model. DCA demonstrated that the combined model achieved higher net benefit within the threshold probability range of 0.10 to 0.80, suggesting promising clinical application value. ConclusionsThe predictive model incorporating the IVIM-DWI true diffusion coefficient D value, histological type and LVSI shows satisfactory predictive efficacy for 3-year DFS among patients with EC. With favorable calibration and clinical utility, this model may serve as a reference for individualized prognostic assessment and therapeutic decision-making in clinical practice. ]]> Tue,20 Oct 2026 00:00:00 GMT http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.014 ObjectiveTo investigate the value of a T2-weighted MRI-based habitat radiomics approach for accurate preoperative prediction of extra-pelvic peritoneal metastasis in patients with epithelial ovarian cancer (EOC). Materials and MethodsClinical and MRI data from 301 patients with pathologically confirmed EOC were retrospectively enrolled from two centers. The two centers served as a training cohort (n = 206) and an external test cohort (n = 95), respectively. Based on habitat analysis, tumor subregions were identified on preoperative axial fat-suppressed T2-weighted images, and radiomics features were extracted from each habitat subregion. A stable feature selection strategy based on 100 repeated random samplings was used, and the top 10 highly stable features were retained to construct the habitat radiomics model. The modeling performance of the logistic regression, support vector machine, random forest, and K-nearest neighbor algorithms was compared to determine the optimal model. SHapley Additive exPlanations (SHAP) was used to evaluate the contribution of each radiomics feature. Multivariable logistic regression was used to identify independent clinical predictors, which were combined with the habitat radiomics score to construct an ensemble model. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), calibration curves, and decision curve analysis. Differences in AUCs among models were compared using the DeLong test. ResultsAll tumor regions were divided into three habitat subregions. The support vector machine model was identified as the optimal habitat radiomics model, with a mean AUC of 0.832 ± 0.086 in 10-fold cross-validation. After feature selection and SHAP analysis, H1_wavelet-HHL_glcm_MCC showed the highest reproducibility and made the greatest contribution to the prediction of the habitat radiomics model. Log-transformed serum carbohydrate antigen 125 (CA125) level was an independent clinical predictor of extra-pelvic peritoneal metastasis (P < 0.001). The ensemble model combining the habitat radiomics score and serum CA125 level achieved the best performance, with an AUC of 0.858 (95% CI: 0.765 to 0.935) in the external test cohort, which was significantly higher than that of the clinical model (P = 0.045). ConclusionsThe preoperative MRI-based habitat radiomics can be used to assess the risk of extra-pelvic peritoneal metastasis in EOC, and its combination with serum CA125 further improves predictive performance, potentially providing useful information for preoperative risk stratification, staging assessment, and treatment decision-making. ]]> Tue,20 Oct 2026 00:00:00 GMT http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.015 ObjectiveTo develop a radiomics nomogram based on multi-parameter magnetic resonance imaging (mpMRI) and evaluate its performance in distinguishing benign from malignant ovarian-adnexal reporting and data system (O-RADS) 4 ovarian lesions. Materials and MethodsThis is a retrospective, single-center study on developing a diagnostic prediction model. This study included 178 O-RADS 4 ovarian masses from 178 patients with ovarian space occupying lesions, with 69 benign and 109 malignant. All lesions were randomly divided into a training set and a validation set at a 7∶3 ratio. Radiomics features were extracted from axial T2-weighted imaging (T2WI), T1-weighted imaging (T1WI), diffusion-weighted imaging (DWI), and contrast-enhanced T1WI (CE-T1WI) sequences for each lesion. Relevant features were selected using minimum redundancy maximum relevance (mRMR) and least absolute shrinkage and selection operator (LASSO) regression to build a radiomics model. A radiomics nomogram was then constructed by combining radiomics features with clinical characteristics through multivariate logistic regression. The nomogram^,s performance was assessed using receiver operating characteristic (ROC) curve analysis, with comparisons performed by DeLong test. Calibration curves and decision curve analysis were used to evaluate the nomogram^,s goodness-of-fit and clinical usefulness. ResultsAfter feature selection, 16 radiomics features were retained. The radiomics nomogram combining human epididymis protein 4 (HE4) and radiomics score (Radscore) achieved area under the curve (AUC) values of 0.885 [95% confidence interval (CI): 0.824 to 0.946] in the training set and 0.877 (95% CI: 0.782 to 0.971) in the validation set, with accuracies of 85.7% and 84.6%, respectively. ROC and decision curve analyses showed that the radiomics nomogram outperformed both the clinical model and the radiomics-only model. ConclusionsThe radiomics nomogram integrating mpMRI-based radiomics features with clinical factors can accurately and intuitively differentiate benign from malignant O-RADS 4 ovarian lesions. It can provide a non-invasive imaging reference for individualized preoperative risk assessment. ]]> Tue,20 Oct 2026 00:00:00 GMT http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.016 ObjectiveTo investigate the feasibility of intravoxel incoherent motion (IVIM) imaging for quantitatively characterizing myocardial tissue in a minipig model of acute myocardial infarction (AMI) established by interventional embolization. Materials and MethodsTen healthy Guizhou minipigs were enrolled. IVIM imaging was performed before and after left circumflex artery (LCX) occlusion. The slow diffusion coefficient (D, ADC_slow), fast perfusion coefficient (D^*, ADC_fast) and perfusion fraction (f) were measured in normal, infarcted and peri-infarct myocardium. Differences in these parameters among the three regions were compared, and their diagnostic performance was analyzed. ResultsA total of 48 image slices were acquired, among which 25 slices met the quality criteria, yielding a success rate of 52.1%. Univariate analysis showed that heart rate was significantly lower in the successful group than in the failure group, suggesting heart rate may affect image success rate (P < 0.05). The overall differences in D^* and f among the three myocardial regions were statistically significant (P < 0.05). Post-hoc comparisons revealed that D^* differed significantly between normal and infarcted myocardium, as well as between infarcted and peri-infarct myocardium (P < 0.001). The f in the infarcted region was significantly lower than that in normal and peri-infarct myocardium (P = 0.011), while no significant difference in f was found between normal and peri-infarct myocardium (P = 0.152). No significant intergroup difference in D was observed in overall or pairwise comparisons (P > 0.05). The receiver operating characteristic curve analysis showed that the area under the curve (AUC) of combined D^* and f, D^*, f, and D was 0.827, 0.806, 0.758 and 0.562, respectively. ConclusionsExperiments based on this embolization‑induced minipig model of acute myocardial infarction suggest that IVIM‑derived D^* and f values can reflect early myocardial injury and microcirculatory perfusion alterations. The combination of these parameters exhibits favorable diagnostic performance in distinguishing infarcted myocardium from the peri‑infarct myocardium. However, given the inherent limitations of the animal model, the suboptimal reproducibility of D^*, the small sample size, and the modest image acquisition success rate, these findings should be interpreted with caution and cannot be directly extrapolated to clinical practice. ]]> Tue,20 Oct 2026 00:00:00 GMT http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.017 ObjectiveThis study involved the conjugation of Fe₃O₄ nanoparticles with prostate-specific membrane monoclonal antibody (Anti-PSMA) to fabricate Fe₃O₄-PSMA magnetic resonance molecular imaging probes. Subsequent evaluation of their biosafety and targeted MRI efficacy was performed to establish experimental evidence for the early diagnosis of prostate cancer. Materials and MethodsThe Fe₃O₄ was coupled with PSMA to construct the Fe₃O₄-PSMA molecular probe. The characteristic functional groups of Fe₃O₄-PSMA were characterized using a Fourier Transform Infrared Spectrometer (FT-IR) to verify the coupling efficiency. Its morphology was observed by Transmission Electron Microscopy (TEM). Element analysis was performed via X-ray Photoelectron Spectroscopy (XPS). The in vitro T2 relaxation rate of the probe was measured. Flow cytometry was employed to assess the effects of 10 μg/mL Fe₃O₄-PSMA molecular probe on prostate cancer C4-2 cells after 24 and 48 hours of exposure, evaluating its impact on apoptosis. CCK8 assay was conducted to evaluate the effects of varying concentrations (0 to 500 μg/mL) of Fe₃O₄-PSMA molecular probe on C4-2 cells over 24 hours, assessing its biosafety at the cellular level. We established tumor-bearing mouse models with high PSMA expression in prostate cancer C4-2 cells (experimental group n = 10) and extremely low PSMA expression in prostate cancer PC-3 cells (control group n = 10). After intravenous injection of Fe₃O₄-PSMA molecular probes into the tail veins, changes in tumor T2-weighted imaging (T2WI) signals and tumor-to-normal tissue signal ratios (TNR) were measured at 2 h, 6 h, and 24 h in both groups. Prussian blue staining was used to visualize probe distribution in tumor tissues and key organs across both groups. Hematoxylin and eosin (HE) staining assessed histopathological morphology differences between tumor tissues in both groups. In vivo pharmacokinetic profiles of the probes were also evaluated. At day 7 post-injection, HE staining, TUNEL assays, and Prussian blue immunohistochemical staining were performed on tissue sections from all organs. ResultsFT-IR spectroscopic analysis demonstrated successful conjugation of Fe₃O₄ with the PSMA antibody, with characteristic acylamide absorption peaks observed in the spectrum, including amino skeleton vibration peaks (NH2-st), amino deformation vibration peaks (NH2-δ), carbonyl skeleton vibration peaks (C=O-st), and carbon-nitrogen skeleton vibration peaks (C-N-st). TEM revealed uniform nanoscale size distribution of Fe₃O₄-PSMA molecules. XPS elemental analysis identified Fe₃O₄-PSMA as primarily composed of five elements: C, O, N, S, and Fe. The r2 relaxation rate of Fe₃O₄-PSMA was 15.224 mM^-1S^-1. Flow cytometry results showed that both 10 μg/mL Fe₃O₄-PSMA probes induced apoptosis in C4-2 cells after 24 and 48 hours of exposure, with an apoptosis rate of 8.101% ± 0.332% at 24 hours and rising to 29.321% ± 1.561% at 48 hours, indicating a significant increase with prolonged exposure time. Animal experimental results demonstrated that at 6 hours after injection of the Fe₃O₄-PSMA probe, significant differences were observed in tumor TNR values between the two groups of mice. The TNR values in the experimental group showed progressive decline over time, indicating specific accumulation of the probe in tumor tissue and a marked reduction in T2 signal intensity. Prussian blue staining demonstrated substantial probe enrichment in tumor and renal tissues of the experimental group at 24 hours post-injection, whereas minimal accumulation was observed in the heart, liver, and spleen. ConclusionsThis study successfully constructed the Fe₃O₄-PSMA magnetic resonance molecular imaging probe. Within this experimental system, the probe exhibited acceptable short-term biosafety. It can target PSMA-overexpressing prostate cancer xenografts and reduce the T2 signal intensity of lesions, enabling targeted magnetic resonance imaging of prostate cancer in tumor-bearing mouse models. Renal accumulation of the probe was observed. This work provides experimental evidence for early targeted molecular imaging research of prostate. ]]> Tue,20 Oct 2026 00:00:00 GMT http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.018 ObjectiveTo investigate the effects of artificial intelligence‑assisted compressed sensing (ACS) with different acceleration factors combined with deep learning reconstruction (DLR) of varying strengths on scan time and image quality of high‑resolution 5.0 T breast T2WI. Materials and MethodsFifty‑five patients (109 breasts) were prospectively enrolled. All patients underwent T2‑weighted scans with three acceleration factors (ACS2, ACS5, ACS8). Images acquired with each ACS setting were reconstructed using non‑DLR (Non‑DLR), moderate‑strength DLR (DLR‑3), and high‑strength DLR (DLR‑5). Scan times under different acceleration factors were recorded. Subjective and objective image quality assessments were performed across combinations of ACS and DLR parameters. Objective evaluation included signal‑to‑noise ratio (SNR) of glandular tissue, adipose tissue and pectoralis major muscle, lesion contrast‑to‑noise ratio (CNR), and background standard deviation (SD). Two radiologists performed 5‑point Likert‑scale subjective scoring for overall image sharpness, lesion boundary delineation, adjacent‑organ clarity, image distortion, and motion artifacts. Weighted Kappa test was used to assess inter-observer agreement. The Friedman χ² test was applied to compare differences in SNR and CNR among various ACS‑DLR combinations. ResultsScan time decreased significantly with increasing ACS acceleration factors: 7 min 58 s for ACS2, 4 min 53 s for ACS5, and 3 min 59 s for ACS8. Under identical ACS settings, increasing DLR strength (0, 3, 5) yielded higher signal intensity (SI) and SNR for lesions and glandular tissue as well as reduced background SD (P < 0.001), whereas no significant difference was found in CNR (P > 0.05). With the same DLR strength, lesion SI tended to decrease as the ACS acceleration factor increased; lesion SNR first increased and then decreased. No statistically significant difference in lesion SNR was observed between ACS5 and ACS8 under identical DLR conditions (P > 0.05); background SD decreased (P < 0.001), while CNR showed no significant difference (P > 0.05). The ACS8 combined with DLR‑5 protocol achieved the highest subjective image scores (P < 0.05). Inter‑observer agreement for the five subjective dimensions was good (Kappa: 0.783 to 0.818). The intra‑class correlation coefficient (ICC) for composite subjective scores ranged from 0.831 to 0.912, with the lower bound of the 95% confidence interval (CI) as low as 0.721, indicating moderate‑to‑good agreement between the two radiologists and satisfying the requirements for clinical research. ConclusionsACS can markedly reduce background SD and improve SNR. The ACS8 combined with DLR‑5 protocol shortens scan time by 50% compared with ACS2 while maintaining favorable image quality. High‑resolution T2WI provides a technical foundation for non‑contrast fast breast imaging. ]]> Tue,20 Oct 2026 00:00:00 GMT http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.020 ^,s disease (AD) is the most common neurodegenerative disorder characterized by a complex pathophysiological process involving amyloid-β (Aβ) deposition, tau pathology, neurodegeneration, neuroinflammation, vascular injury, and synaptic dysfunction. With the expansion of ATX(N) diagnostic framework, the X dimension, comprising biomarkers of inflammation (I), vascular injury (V), and α-synucleinopathy (S), has been incorporated into the Aβ pathology (A), tau pathology (T), and neurodegeneration (N) framework, shifting AD diagnosis from single-pathology assessment toward multidimensional biological evaluation. Positron emission tomography/magnetic resonance imaging (PET/MRI), as an advanced imaging approach integrating molecular pathology with structural and functional alterations, plays a critical role in the early diagnosis, disease staging, and therapeutic monitoring of AD. This review summarizes the applications and clinical value of multi-tracer PET and multimodal MRI techniques in AD diagnosis and management within the ATX(N) framework, and further discusses the potential role of PET/MRI in monitoring disease-modifying therapies. The current limitations of PET/MRI research and future directions are also discussed, with the aim of providing a reference for precision diagnosis and treatment of AD and further development of multimodal imaging. ]]> Tue,20 Oct 2026 00:00:00 GMT http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.021 ^,s disease (AD) is a neurodegenerative disorder characterized by progressive cognitive impairment and represents the most common cause of dementia. Acupuncture has been used as a therapeutic approach for AD and has shown potential in alleviating AD-related symptoms and delaying disease progression. Neuroimaging techniques provide objective and quantifiable evidence for elucidating the central mechanisms underlying acupuncture effects in AD. However, existing studies have predominantly focused on functional magnetic resonance imaging, and a systematic synthesis and comprehensive analysis from a multimodal neuroimaging perspective remain lacking regarding the objective effects of acupuncture in the treatment of AD, the key brain regions involved, the patterns of brain network modulation, and the underlying mechanisms. Therefore, we systematically summarize neuroimaging advances in acupuncture intervention for AD, focusing on magnetic resonance imaging modalities, including structural MRI, functional MRI, diffusion tensor imaging, arterial spin labeling, and magnetic resonance spectroscopy, as well as positron emission tomography and functional near-infrared spectroscopy. We synthesize findings related to brain structure, functional activity and connectivity, cerebral metabolism, and cerebral blood flow/perfusion, and discuss AD-related intracerebral alterations and the putative central regulatory mechanisms of acupuncture from a multidimensional perspective. This review aims to provide a reference for future studies and to facilitate further mechanistic and clinical research in this field. ]]> Tue,20 Oct 2026 00:00:00 GMT http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.022 Tue,20 Oct 2026 00:00:00 GMT http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.023 Tue,20 Oct 2026 00:00:00 GMT http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.024 Tue,20 Oct 2026 00:00:00 GMT http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.025 Tue,20 Oct 2026 00:00:00 GMT http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.026 ^, experience, leading to subjectivity, limited inter-observer consistency, and insufficient use of high-dimensional image information. Deep learning and radiomics provide new technical routes for quantitative IBD imaging through automatic segmentation, feature extraction, and predictive modeling. This review summarizes the application progress of deep learning and radiomics in IUS, CTE, and MRE, and compares the performance, advantages, and limitations of different models across key clinical tasks, including disease activity quantification, stricture characterization, complication assessment, treatment response prediction, and prognosis prediction. Current bottlenecks related to data standardization, external validation, interpretability, and clinical system integration are further discussed. Future studies should strengthen multicenter prospective validation, real-time computer-aided diagnosis, multimodal and multi-omics integration, and image-pathology correspondence research, thereby facilitating clinical translation and providing references for clinicians and researchers in selecting appropriate radiomics and deep learning approaches for IBD. ]]> Tue,20 Oct 2026 00:00:00 GMT http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.027 Tue,20 Oct 2026 00:00:00 GMT http://med-sci.cn/cgzcx/en/en_articlexml.asp?doi=10.12015/issn.1674-8034.2026.10.028 Tue,20 Oct 2026 00:00:00 GMT