BackgroundNon-cardioembolic non-disabling minor ischemic stroke carries a considerable risk of early recurrence that is not fully captured by clinical severity scales. Hemodynamic impairment and collateral status may improve risk stratification.PurposeTo assess the prognostic value of four-dimensional computed tomography angiography (4D-CTA)-derived collateral status and CT perfusion parameters for predicting recurrence in patients with minor stroke and moderate middle cerebral artery (MCA) stenosis.Material and MethodsIn this prospective study (July 2024 to June 2025), 183 patients with minor ischemic stroke (National Institutes of Health Stroke Scale (NIHSS) ≤ 5) and symptomatic moderate (50%-69%) MCA stenosis were enrolled. All underwent multimodal CT imaging and received standardized dual antiplatelet therapy. Collateral status was graded using 4D-CTA (modified American Society of Interventional and Therapeutic Neuroradiology/Society of Interventional Radiology (ASITN/SIR)), and perfusion parameters, including relative cerebral blood flow (rCBF) and Tmax, were quantified. The primary endpoint was recurrent ischemic stroke or transient ischemic attack. Cox regression and subgroup analyses were performed.ResultsOver a median follow-up of approximately 10 months, rCBF < 30% volume was the only independent predictor of recurrence (hazard ratio (HR) = 1.168, 95% confidence interval (CI) 1.039-1.314; P = 0.010). Poor collateral status showed a non-significant trend overall (HR = 2.079; P = 0.093) but was strongly associated with recurrence in patients without limb weakness (HR = 9.875; P = 0.030).ConclusionMultimodal CT-derived hemodynamic parameters outperform clinical scores in predicting recurrence. rCBF <30% is a robust predictor, while 4D-CTA collateral assessment is particularly valuable in clinically mild patients without motor deficits.

BackgroundResidual tibial tubercle morphology compatible with previous Osgood-Schlatter disease (OSD) may coexist with patellofemoral malalignment; however, its relationship with patellar cartilage damage is unclear.PurposeTo compare magnetic resonance imaging (MRI)-based patellofemoral morphology between adults with and without an OSD-compatible MRI phenotype and evaluate imaging factors associated with lateral patellar cartilage degeneration.Material and MethodsThis retrospective study included 199 adults (209 knees) with an OSD-compatible MRI phenotype and 238 age- and sex-frequency-matched symptomatic comparison participants. Patellofemoral measurements and lateral patellar cartilage were assessed on 1.5-T MRI. A patient-clustered parsimonious generalized estimating equation model included age, sulcus angle, patellar tilt, and Insall-Salvati index; performance was evaluated by patient-level bootstrap.ResultsOSD-compatible knees showed greater patellofemoral malalignment and more frequent lateral patellar cartilage degeneration than symptomatic comparison knees (47.8% vs. 10.1%; P < 0.001). Older age (adjusted odds ratio (OR) = 1.12 per year, 95% confidence interval (CI) = 1.08-1.16), larger sulcus angle (OR = 1.07 per degree, 95% CI = 1.01-1.13), greater patellar tilt (OR = 1.19 per degree, 95% CI = 1.09-1.30), and higher Insall-Salvati index (OR = 1.64 per 0.10 increase, 95% CI = 1.11-2.44) were associated with chondromalacia. Apparent and optimism-corrected areas under the receiver operating characteristic curve were 0.893 and 0.884, respectively.ConclusionAdults with an OSD-compatible MRI phenotype had more patellofemoral malalignment and lateral patellar cartilage degeneration than symptomatic comparison participants. The cross-sectional design and unmeasured confounding limit causal and population-level inference.

BackgroundSeromucinous borderline tumor (SMBT) and endometriosis-associated ovarian cancer (EAOC) may show overlapping magnetic resonance imaging (MRI) features. Quantitative diffusion-weighted imaging may improve their preoperative differentiation.PurposeTo evaluate the diagnostic performance of the cystic-solid apparent diffusion coefficient difference (ΔADC) for differentiating SMBT from EAOC.Material and MethodsThis retrospective single-center study included 49 patients with pathologically confirmed SMBT (n = 14) or EAOC (n = 35) who underwent preoperative pelvic MRI. Apparent diffusion coefficient (ADC) values were measured in the cystic and solid components, and ΔADC was calculated as cystic ADC minus solid ADC. Receiver operating characteristic (ROC) analysis was performed with EAOC as the positive outcome.ResultsSMBT showed lower cystic ADC values and higher solid ADC values than EAOC, resulting in a smaller ΔADC. ΔADC and solid ADC yielded areas under the ROC curve of 0.924 (95% confidence interval (CI) 0.817-1.000) and 0.918 (95% CI 0.837-1.000), respectively. The difference was not statistically significant (P = 0.862). At the internally derived ΔADC cutoff of 0.529 ×10[-3] mm[2]/s, sensitivity was 94.3% (95% CI 80.8-99.3) and specificity was 92.9% (95% CI 66.1-99.8); because the cutoff was derived and evaluated in the same cohort, these performance estimates may be optimistic.ConclusionΔADC showed high diagnostic performance for differentiating SMBT from EAOC, but incremental diagnostic performance over solid ADC was not demonstrated. The internally derived cutoff and associated performance estimates require independent multicenter validation before clinical application.

BackgroundComputed tomography pulmonary angiography (CTPA) is essential for diagnosing pulmonary embolism (PE); however, the clinical and prognostic relevance of clot burden remains uncertain.PurposeTo evaluate the clinical significance and prognostic value of the refined Modified Miller Score (rMMS) in patients with intermediate- to high-risk PE.Material and MethodsIn total, 204 patients with intermediate- to high-risk PE who underwent baseline CTPA were included. Data on demographics, comorbidities, and biomarkers (troponin I/T, lactate, and D-dimer) were collected at admission. Patients were stratified by rMMS quartiles (Q1-Q4). Trends were analyzed using Cochran-Armitage and Jonckheere-Terpstra tests, while associations were assessed using Pearson's correlation and multivariable logistic regression.ResultsBaseline characteristics were comparable across quartiles. Higher rMMS was associated with alcohol use and prior arrhythmia. Increasing rMMS correlated with higher right ventricle:left ventricle ratio, increased heart rate, lower systolic blood pressure, and greater oxygen requirement. Weak correlations were observed between rMMS and lactate and troponin T, whereas no significant associations were found with troponin I or D-dimer. Crucially, rMMS was not associated with in-hospital outcomes or 1-year mortality.ConclusionThe rMMS reflects right ventricular strain in PE but does not appear suitable for prognostic risk stratification or guiding management decisions.

BackgroundDifferentiating ovarian endometriomas from hemorrhagic cysts matters because their management differs.PurposeTo evaluate whether a multiparametric magnetic resonance imaging (MRI) approach improves differentiation of endometriomas from hemorrhagic cysts over conventional subjective assessment.Material and MethodsIn total, 83 hemorrhagic cystic ovarian lesions (45 endometriomas, 38 hemorrhagic cysts) on 1.5-T pelvic MRI were retrospectively evaluated by two independent blinded radiologists (with 20 and 7 years of experience in abdominopelvic and gynecological MRI) for quantitative T1 and T2 signal intensities (quantitative shading), intralesional and perilesional apparent diffusion coefficient (ADC), lesion size, and the T2 shading and dark-spot signs. ROC analysis, logistic regression, and inter-/intra-observer agreement were assessed, and models were compared with the DeLong test.ResultsEndometriomas were larger (55.0 vs. 40.7 mm; P = 0.007), had lower intralesional ADC (1.054 vs. 1.563 ×10[-3] mm[2]/s; P = 0.002), and a greater T1-T2 signal difference (P < 0.001), which had the highest single-parameter accuracy (AUC = 0.799). There was no statistically significant difference in the T2 shadowing sign between the groups (91.1% vs. 81.6%; P = 0.342). The T2 dark-spot sign was specific (100%) but insensitive (24.4%). Perilesional ADC did not differ (P = 0.112) and showed poor inter-observer agreement (ICC = 0.064). A multiparametric model (quantitative shading, intralesional ADC, lesion size) outperformed conventional MRI (AUC = 0.894 vs. 0.670; DeLong P < 0.001), increasing sensitivity from 24.4% to 88.9% at comparable specificity.ConclusionA multiparametric model combining quantitative shading, intralesional ADC and lesion size significantly improved differentiation over conventional MRI, whereas perilesional ADC showed no diagnostic utility at 1.5 T.

BackgroundCentral bronchial deformity (CBD) has been proposed as a radiological marker of a fibrotic upper-lobe phenotype in pulmonary sarcoidosis.PurposeTo evaluate whether CBD and predefined high-resolution computed tomography (HRCT) findings at diagnosis predict the upper-lobe shrinkage (SUL) within 1 year in patients with pulmonary sarcoidosis with upper-lobe involvement.Material and MethodsThis retrospective cohort included adults with histopathologically confirmed pulmonary sarcoidosis who underwent HRCT and chest radiography at diagnosis and approximately 1 year later. Patients with baseline SUL were excluded. Two radiologists independently assessed CBD and predefined HRCT findings. SUL was evaluated on chest radiographs using prespecified signs of upper-lobe volume loss.ResultsAmong 79 patients, baseline CBD was present in 33 (42%). Compared with patients without CBD, those with CBD were older (54 ± 12 vs. 46 ± 14 years; P = 0.009), had greater smoking exposure (25 (12-35) vs. 15 (6-20) pack-years; P = 0.046), more frequently received treatment (67% vs. 39%; P = 0.029), and had lower spirometric values. Neither baseline CBD (P = 0.966) nor other baseline HRCT findings were associated with incident SUL. At 1 year, SUL was more frequent among patients with CBD (28% vs. 3%; P = 0.007) and ground-glass attenuation (29% vs. 10%; P = 0.031).ConclusionCBD was associated with lung remodeling and lower spirometric values but did not predict SUL within 1 year. Its associaton with SUL at 1 year suggests that CBD may be a contemporaneous marker of advanced parenchymal involvement and airway distortion rather than a short-term predictor.

BackgroundVolumetric interpolated breath-hold examination (VIBE) is widely used in upper abdominal MRI but can be limited by low signal-to-noise ratio (SNR), especially when rapid acquisition is prioritized. Deep learning (DL)-enhanced reconstruction may improve image quality without extending acquisition time.PurposeTo compare subjective image quality, artifacts, noise, and estimated SNR and contrast-to-noise ratio (CNR) between DL-supported and standard (ST) VIBE sequences of the upper abdomen in women undergoing pelvic MRI, including the effect of contrast enhancement.Material and MethodsThis prospective study included 60 women (mean age 42.1 ± 14.7 years). Four axial sequence types were evaluated: ST and DL VIBE, both non-contrast (NC) and contrast-enhanced (CE). Three radiologists rated image quality, artifacts, and noise using a standardized 4-point Likert scale. Interobserver agreement and the effects of age and body mass index (BMI) were assessed. A quantitative region-of-interest (ROI)-based SNR/CNR analysis was also performed.ResultsDL VIBE yielded better image quality, fewer artifacts, and less noise than ST VIBE. DL-by-CE interactions were significant for image quality and artifacts. Interobserver agreement was moderate for image quality and artifacts but low for noise. Quantitative analysis showed no significant differences in estimated SNR or CNR between ST and DL VIBE before or after contrast administration (all p ≥ 0.35).ConclusionDL VIBE improves image quality and reduces artifacts, particularly in NC imaging.

BackgroundLeft ventricular non-compaction (LVNC) is increasingly recognized as a morphological phenotype rather than a distinct cardiomyopathy. Current cardiovascular magnetic resonance (CMR) criteria rely predominantly on static morphologic thresholds and may not adequately reflect dynamic myocardial adaptation.PurposeTo investigate whether absolute myocardial mass difference (AD) represents a marker of structural-functional remodeling and to assess its relationship with hemodynamic and flow-related parameters in morphologically defined LVNC.Material and MethodsIn this retrospective study, 99 patients fulfilling Petersen and Grothoff CMR criteria for LVNC and 66 frequency-matched controls underwent multiparametric 1.5-T CMR. Volumetric indices, left atrial volume, cine-derived functional parameters, and aortic phase-contrast MRI flow timing indices, including time-to-peak velocity (TPV), ejection time (ET), and TPV/ET ratio, were analyzed. AD was defined as the indexed difference between end-diastolic myocardial mass and cycle-averaged myocardial mass. Between-group comparisons, effect size analysis, correlation testing, and ROC analysis were performed.ResultsAD was significantly higher in LVNC patients than in controls (3.71 ± 2.43 vs. 2.03 ± 1.49 g/m[2]; P < 0.001) and demonstrated a large effect size (Cohen's d = 0.81). LVNC patients exhibited increased end-diastolic and end-systolic volume indices (both P < 0.001), preserved cardiac index, and altered systolic flow timing characteristics. AD correlated positively with left atrial volume (r = 0.39; P < 0.001) and cardiac index (r = 0.27; P = 0.007), whereas no significant association was observed with TPV/ET. ROC analysis demonstrated modest discriminatory performance (AUC = 0.71).ConclusionIn morphologically defined LVNC, AD is increased and associated with atrial remodeling and preserved hemodynamic adaptation. These findings suggest that AD may represent a marker of dynamic structural-functional remodeling rather than static trabecular excess. Incorporation of AD into multiparametric CMR assessment may improve phenotypic characterization and help distinguish compensated hypertrabecularization from clinically relevant myocardial disease.

BackgroundBronchial artery embolization (BAE) is the first-line treatment for hemoptysis with high success rates. However, recurrent hemoptysis remains a significant clinical problem. The role of non-bronchial systemic arteries (NBSAs) and pulmonary-bronchial arterial anastomoses in predicting recurrence remains incompletely understood.PurposeTo evaluate clinical and angiographic predictors of recurrent hemoptysis after BAE.Material and MethodsThis retrospective single-center study included 148 patients who underwent technically and clinically successful BAE for hemoptysis between 2022 and 2024. Univariate and multivariate logistic regression analyses were performed to identify independent predictors of recurrence. ROC analysis was used to evaluate the predictive value of hemoptysis volume.ResultsRecurrent hemoptysis occurred in 42 (28.4%) patients. Multivariate analysis identified NBSAs (odds ratio (OR) = 10.11, 95% confidence interval (CI) = 1.50-68.02; P = 0.017), pulmonary-bronchial arterial anastomoses (OR = 4.43, 95% CI = 1.53-12.85; P = 0.006), and tuberculosis etiology (OR = 6.81, 95% CI = 2.20-21.05; P < 0.001) as independent predictors. Hemoptysis volume showed limited predictive performance (AUC = 0.591; P = 0.039).ConclusionNBSAs and pulmonary-bronchial anastomoses are strong independent predictors of recurrent hemoptysis after BAE. Notably, this association persisted even when NBSAs were successfully embolized, suggesting a biological rather than technical basis for recurrence. Patients with complex vascular anatomy and tuberculosis require closer post-procedural surveillance and may benefit from more intensive follow-up strategies after embolization.

BackgroundAccurate staging of cervical cancer is essential for treatment planning, with Magnetic Resonance Imaging (MRI) being the modality of choice. Cervical cancer is suspected to have variations in its presentation and etiology depending on the age of patients.PurposeTo explore the association between patient age and specific findings on contrast-enhanced MRI in women with newly diagnosed cervical cancer.Material and MethodsThis study included 146 consecutive patients with pathologically confirmed cervical cancer who underwent 3.0T contrast-enhanced MRI for initial staging. MRI scans were evaluated for tumor size, parametrium, uterine, bladder, and intestinal invasion, among other features. Ordinal, binary, and multinomial logistic regression models were used to assess age as a predictor for MRI stage and specific invasive findings. Receiver Operating Characteristic (ROC) curve analysis was performed for significant findings.ResultsThe mean patient age was 49.1 ± 12.2 years. Age was a significant predictor of a higher MRI stage (OR: 1.029, p = 0.025). Increasing age emerged as a significant predictor for uterine invasion (OR: 1.050, 95% CI: 1.022-1.085, p = 0.002), an association that remained significant even after controlling for disease stage. Age predicted higher odds of bladder mucosal (OR: 1.052, p = 0.011) and non-mucosal intestinal invasion (OR: 1.038, p = 0.041). ROC identified an age cutoff of 45.5 for uterine invasion (AUC: 65.2%).ConclusionRadiologists should maintain a heightened index of suspicion for more extensive local invasion when interpreting cervical cancer staging MRIs in older women.

BackgroundSeveral commercial artificial intelligence (Al) algorithms are available for detecting intracranial hemorrhage (ICH), but independent clinical validation remains limited.PurposeTo compare three commercially available Al algorithms for ICH detection on non- contrast head computed tomography (NCHCT).Material and MethodsIn this retrospective study, 4027 consecutive NCHCT examinations from a large emergency hospital in southwest Sweden were analyzed. Three Al algorithms were applied, with one vendor disclosing participation. Reports from two radiologists and all Al outputs were evaluated. Human-AI performance was assessed using an idealized logical OR model, assuming radiologists perfectly dismissed all false-positive Al flags to calculate system specificity. All positive or discrepant cases underwent expert manual review using two-tier consensus adjudication as the reference standard.ResultsOf 3902 evaluable examinations, 3517 were consistently negative by all readers. The remaining 385 cases underwent manual review, confirming ICH in 176 cases (4.5% prevalence) and excluding it in 209. Eight ICH cases missed by both radiologists were detected by at least one Al system. Aidoc performed best, with 90.3% sensitivity and 99.0% specificity. A simulated mathematical combination of Aidoc with a human reader increased sensitivity to 96.0% while maintaining 99.4% specificity (P < 0.001), comparable to two radiologists.ConclusionAl performance varied substantially, with only one system demonstrating clinically relevant accuracy. Combining Al with human interpretation improved ICH detection and shows promise for future clinical implementation.

BackgroundNon-invasive monitoring of split renal function (SRF) impairment in the context of urinary tract obstruction remains challenging. Renal scintigraphy (RS), the clinical standard, involves ionizing radiation, whereas magnetic resonance imaging (MRI)-based alternatives are underexplored.PurposeTo evaluate arterial spin-labeled (ASL) and structural MRI for SRF assessment in obstructive nephropathy.Material and MethodsConsecutive patients underwent 3.0-T MRI (ASL/T1-weighted). Kidneys (n = 92) were stratified by scintigraphy-derived glomerular filtration rate (RS-GFR) into normal renal function, mild to moderate renal impairment, severe impairment, or no function groups. Cortical renal blood flow (RBF), parenchymal thickness (RPT), and hydronephrosis width (HW) were measured. Group differences (ANOVA/Kruskal-Wallis) and correlations with RS-GFR (Pearson/Spearman) were analyzed. Receiver operating characteristic curves assessed diagnostic performance.ResultsRBF, RPT, and HW differed significantly among groups (P < 0.001), correlating strongly with RS-GFR (r = 0.767, 0.684, -0.589). They discriminated SRF subgroups (P < 0.001); RBF + RPT + HW achieved areas under the curve of 0.869 (normal vs. mild), 0.883 (mild vs. severe), and 0.994 (normal vs. severe).ConclusionASL-derived RBF and structural markers (RPT and HW) correlate robustly with RS-GFR, offering a feasible MRI-based, non-invasive approach for SRF monitoring.

BackgroundBreath-hold triggered functional magnetic resonance imaging (bh-fMRI) is a widely available, easily implementable and inexpensive method to quantify cerebrovascular reactivity. Estimation of patients' respiratory rate is relevant, as it might influence bh-fMRI data quality, but respiratory sensors integrated in the scanner table are not always available.PurposeTo propose an approach to estimate patients' respiratory rate using realignment data of a 1-min high-temporal-resolution cerebral MRI sequence.Material and MethodsIn a prospective study, 15 patients with Moyamoya angiopathy underwent a 1-min MRI sequence during routine clinical MRI. The respiratory rate was obtained by analyzing patients' head motion from image realignment data. To obtain robust results, the respiratory rate was calculated using a multi-method approach, incorporating power spectral density and fast Fourier transform and weighted data from multiparametric rigid body transformation. The respiratory rate detected from respiratory sensors integrated in the scanner table was used as reference. Bland-Altman analysis was performed and intra-class correlation coefficients (ICCs) were calculated to compare the respiratory rates.ResultsBland-Altman analysis revealed a mean bias of 0.5 breaths/min and limits of agreement (LoA) ranging from -5.3 to 4.4 breaths/min, with no outliers. The overall agreement between the respiratory rate calculated from realignment data and from scanner sensor data was excellent (ICC = 0.866, 95% confidence interval = 0.607-0.955; P < 0.001).ConclusionThe calculated mean bias and LoA appear to be clinically acceptable. Estimating respiratory rate from a 1-min high-temporal-resolution MRI sequence may be a feasible method when respiratory sensors integrated in the scanner are not available.

BackgroundMagnetic resonance imaging (MRI) is currently the primary method for evaluating liver iron concentration (LIC); however, most techniques necessitate intricate data post-processing. Consequently, the development of simple quantitative methods is highly desirable.PurposeTo investigate the value of diffusion-weighted imaging (DWI) for quantifying LIC in patients with transfusion-dependent thalassemia (TDT).Material and MethodsData were collected on patients with TDT, including conventional MRI iron quantification sequences such as FerriScan R2, T2*, and DWI data. The liver's apparent diffusion coefficient (ADC) and T2* values were measured, and these measurements were compared to the LIC reports obtained from MRI-R2/FerriScan, which are U.S. Food and Drug Administration certified, to evaluate the utility of ADC values in quantitatively assessing LIC.ResultsAmong the 58 patients with TDT, the linear regression analysis indicated R[2] values in the range of approximately 0.4765-0.6022 for ADC values and LIC, with non-linear regression R[2] in the range of 0.6263-0.6479. The T2* model demonstrated excellent diagnostic accuracy for extremely severe liver iron overload with an area under the receiver operating characteristic curve (AUC) of 0.996, whereas the ADC model exhibited a similarly high AUC in the range of 0.891-0.941.ConclusionIn patients with known hepatic iron overload due to TDT, ADC values show a correlation with liver iron content and have the potential to serve as a supplementary method for quantitative measurement of liver iron by T2*/R2*.

BackgroundPercutaneous transhepatic cholangiography (PTC) is essential when endoscopic retrograde cholangiopancreatography (ERCP) fails or is contraindicated. Despite emerging evidence favoring primary stenting, data directly comparing it with the traditional two-staged drainage approach remain scarce.PurposeTo evaluate the efficacy, complication rates, and reintervention burden of initial stent placement versus drainage.Material and MethodsA retrospective single-center analysis of 96 patients undergoing PTC (2021-2024) was conducted; 95 received intervention (primary stent n = 30, drain n = 65). Outcomes (bilirubin kinetics, clinical success, complications, reinterventions) were compared using Welch's t-test, chi-square test, or Fisher's exact test, with multivariable adjustment for baseline differences.ResultsMalignant obstruction accounted for 96.8% of cases; technical success was 97.9% (CIRSE standards). Bilirubin reduction was greater with stents (45.7% vs. 35.5%; mean difference 10.2 percentage points, 95% confidence interval = 0.3-20.0; P = 0.043), attenuated after adjustment (β = 9.8 pp; P = 0.093). Complications, all reinterventions, and combined adverse events were lower with stents (23.3% vs. 55.4%, 16.7% vs. 73.8%, and 30.0% vs. 86.2%; all P ≤0.004); adjusted analyses confirmed independently reduced odds (adjusted odds ratio = 0.26, 0.08, and 0.07, respectively). Clinical success trended higher with stents (46.7% vs. 38.5%) but was not statistically significant. Overall, 30-day mortality was 17.7% with 2% procedure-related. Median post-intervention survival was 101 days.ConclusionIn this retrospective single-center cohort, a stent-first approach was associated with greater bilirubin reduction and fewer complications and reinterventions than drainage-first, without compromising technical success or survival. Findings are hypothesis-generating and require prospective multicenter validation.