AJNR. American journal of neuroradiology最新文献

Background and purpose: Predicting intracranial aneurysm growth remains difficult, with relatively few widely accepted predictors beyond aneurysm size, location, and patient age. This study investigated the relationship of vascular morphology and hemodynamics to aneurysm location and growth using a matched pairs study design that limited the influence of other predictors.

Materials and methods: Longitudinally followed ICA aneurysms identified as growing or stable were matched by size and location, based on their proximity to either the ophthalmic arteries, superior hypophyseal arteries, or posterior communicating arteries (PComA). The 34 aneurysms (17 pairs) were monitored with 3D CTA imaging for an average of 4.10 (1.98) years. Three longitudinal image studies for each aneurysm were collected and used to reconstruct models of arterial geometry. These models were used to analyze morphology and hemodynamics over time with computational fluid dynamics simulation. Results were analyzed with the Wilcoxon signed-rank test and longitudinal multivariate models were built with generalized estimating equations clustered according to each aneurysm.

Results: Higher average flow (P < .01) was observed at the aneurysm inlet and lower average pulsatility index at the outlet (P < .05) of aneurysms near the PComA. Over time, morphologic metrics related to size differed between paired aneurysms, but shape metrics did not. Pulsatility index at the aneurysm body and outlet differed between paired aneurysms (P < .01). Modeling identified correlates of aneurysm size including aneurysm flow and wall shear stress pulsatility index and generally demonstrated high correlation between different time points (0.94). The models showed a strong association between the attachment of the nearby artery to the aneurysm and growth.

Conclusions: Significant differences in hemodynamics were observed between nearby aneurysm locations. Despite the relatively small sample size, longitudinally measured hemodynamics and general aneurysm morphology were significant correlates for aneurysm growth and size, as well as possible predictors of future aneurysm growth.

Background and purpose: Localizing a CSF leak in a patient with spontaneous intracranial hypotension (SIH) is of utmost importance when pursuing a targeted therapy. The study aimed to evaluate the accuracy of dynamic myelography techniques in localizing spinal CSF leaks and report the number of examinations required, stratified by leak type and spinal level.

Materials and methods: Consecutive patients with SIH with a spinal longitudinal extradural CSF collection (SLEC) investigated at our department from January 2013 to February 2025 were screened. All included patients underwent a dynamic myelography work-up to localize the level of spinal CSF leak using conventional dynamic myelography (CDM) and/or dynamic CT myelography (DCTM).

Results: In total, 198 SLEC-positive patients with SIH (mean age: 50 ± 12 years; 67% women, 133/198) were included. In total, 147 patients had a ventral (74%), 49 patients had lateral (25%), and 2 patients had a primary dorsal (1%) leak. The spinal CSF leak was localized with the first, second, third, or fourth dynamic myelography in 97 (49%), 70 (35%), 16 (8%), and 11 patients (6%), respectively. The median number of myelography examinations (CDM + DCTM) per patient to localize a CSF leak was 2 (interquartile range [IQR] 1-2; range 1-8), 1 (IQR 1-2; range 1-5) for ventral, 2 (IQR 1-2; range 1-6) for lateral, and 6 (IQR 5-7; range 4-8) for dorsal leaks. In total, 160 patients (81%) were referred for microsurgical closure. The dural leak was identified intraoperatively on the indicated vertebral level in 153 patients (96%); in 2 patients (1.3%), spontaneous sealing occurred, and in 5 patients (3%), wrong level surgery occurred.

Conclusions: Dynamic myelography examinations accurately and reliably localize spinal CSF leaks in patients with SIH with SLEC. In about one-half of the patients, the level of the leak can be localized with the first CDM. In case of a repeat dynamic myelography, the technique and patient positioning can be adopted according to the results of the previous examination. Primary dorsal leaks are rare, but because of the low level of suspicion, they pose a diagnostic challenge.

Background and purpose: Aneurysmal SAH (aSAH) is a life-threatening condition associated with angiographic vasospasm, delayed cerebral ischemia (DCI), and other complications that may lead to significant morbidity and mortality. Our study evaluated associations of vessel wall enhancement (VWE) on intracranial vessel wall MRI with rupture status, and predictive value of VWE, conventional imaging, and clinical features for angiographic vasospasm, DCI, which encompasses delayed infarction and symptomatic vasospasm.

Materials and methods: A retrospective cohort study included patients who underwent endovascular treatment for ruptured or unruptured intracranial aneurysms with immediate postintervention vessel wall MRI between November 2015 and August 2022. Logistic regression models were used to assess the relationship of VWE with SAH, angiographic vasospasm, and a composite of DCI, delayed infarction, and symptomatic vasospasm after adjustment for clinical and traditional imaging factors. We used adjusted VWE segments as a qualitative variable, defined as the proportion of enhanced vessel segments relative to the total number of evaluable segments per patient.

Results: Among 128 patients (79 ruptured, 49 unruptured), adjusted VWE was significantly higher among patients with ruptured than unruptured aneurysms (OR: 1.62; 95% CI: 1.39-1.99; P < .001). In the rupture cohort, adjusted VWE was the only independent predictor of angiographic vasospasm (OR: 1.12; 95% CI: 1.03-1.23; P = .01). In contrast, the modified Fisher grade and hypertension were independent predictors of DCI (P < .05), whereas adjusted VWE was not.

Conclusions: Qualitative adjusted VWE is associated with SAH and is a reliable, independent predictor of angiographic vasospasm, but is not predictive of DCI. Given the multifactorial nature of SAH-related complications, integrating VWE with conventional imaging and clinical factors may improve risk stratification and predictive assessment.

Purpose: This multi-institutional study investigates how preoperative functional MRI (fMRI) and diffusion tensor imaging (DTI) influence surgical decision-making and clinical outcomes in patients undergoing brain tumor resection.

Methods and materials: Seventy patients from four academic centers: Thomas Jefferson University (TJU), n=51, University Hospital Basel, n=11; University of Pennsylvania (UPenn), n=4, and Johns Hopkins University (JHU) n=3, underwent preoperative task-based fMRI and DTI. Six neurosurgeons completed structured pre- and post-imaging surveys evaluating changes in surgical approach, craniotomy planning, extent of resection, operative duration, and diagnostic confidence.

Results: Integration of fMRI and DTI into surgical planning resulted in a significant shift from awake to asleep craniotomies, especially at TJU (P = 0.01), with "asleep craniotomy" increasing to 51% overall (Chi-square P < .0001). fMRI led to a "much more aggressive" surgical plan in 39% of cases globally, most prominently at TJU (74%) and UPenn (50%), while JHU reported a decrease in aggressiveness in 33.3% of cases. DTI had a similar but slightly reduced impact, with "much more aggressive" being the top response (34%). fMRI was rated as more clinically valuable than DTI in 53.4% of cases overall with TJU having the highest rate (72%). Postoperatively, a larger extent of resection was reported in 61% of cases, with shorter-than-expected surgical durations in 51%. Overall, combined fMRI/DTI had a significant "strong positive" influence on surgery in 71% and clinical care in 68% of cases, with significant inter-institutional differences (P < 0.001).

Conclusion: Preoperative fMRI and DTI significantly reshape neurosurgical planning by optimizing resection strategies. Most notably, preoperative mapping facilitated a significant shift from awake to asleep craniotomies, contributing to shorter than expected surgical durations without compromising the extent of resection.

Background and purpose: Pediatric low-grade gliomas (pLGGs) are the most common brain tumors in children and frequently harbor BRAF alterations, most commonly KIAA1549-BRAF fusions and BRAF V600E mutations, which have distinct therapeutic implications. The aim of our study was to assess multiclass radiomics-based prediction of BRAF mutation status in children with pLGG using multi-sequence MRI. This study follows TRIPOD-AI guidelines.

Materials and methods: This retrospective bi-institutional study included pediatric patients with pLGG and known BRAF mutation status who underwent pre-surgical MRI between January 2009 and January 2023. Tumors were manually segmented, and radiomics features were extracted using PyRadiomics. Random Forest classifiers were trained for three-class classification (BRAF fusion vs BRAF V600E vs non-BRAF) using clinical-only, radiomics-only, and combined models. Performance was evaluated with leave-one-out cross-validation, and results were compared across single-sequence and multisequence approaches. Single-sequence models were trained using all available patients for each MRI sequence, whereas multisequence models were restricted to the subset of 180 patients with all four sequences available.

Results: 511 children were included (mean age 8.5 ± 5.1 years; 45% female). Molecular subtypes included BRAF fusion (223/511, 44.6%), BRAF V600E (105/511, 21.0%), and non-BRAF tumors (172/511, 34.4%). FLAIR sequences were available for 495, T2WI for 454, contrast-enhanced T1WI (CE-T1WI) for 285 and ADC maps for 252 children. All sequences were available for 180 children. FLAIR was the best-performing single sequence (AUC 0.82), followed by T2WI (0.80), ADC (0.77), and CE-T1WI (0.75). Reported AUC values represent macro-average one-vs-rest performance across the three molecular classes. Combined clinical-radiomics models consistently outperformed single-source models. In the 180-patient multisequence cohort, radiomics feature concatenation (macro-AUC 0.79) and ensemble modeling (0.79) both outperformed single-sequence approaches (p < 0.001). Feature analysis showed FLAIR-derived features dominated, but adding T2, ADC, and CE-T1WI improved balanced classification across subtypes.

Conclusion: MRI-based machine learning models may support noninvasive prediction of BRAF mutation status in pLGG. FLAIR is the best-predicting single sequence, but multisequence integration was associated with improved and more balanced performance. These findings support multisequence radiomics as a promising tool to guide precision treatment in pLGG, particularly when tissue sampling is not feasible.

Background and purpose: Aicardi-Goutières syndrome (AGS) is a rare, genetically-determined spectrum of neurodegenerative disorders that remains poorly understood. Owing to the paucity of data from Middle-Eastern population, we aimed to delineate the clinical, radiological, and genetic features of AGS in an under-represented Middle-Eastern cohort.

Materials and methods: A retrospective case-series review was performed of all genetically-confirmed AGS cases managed at a tertiary pediatric hospital in Qatar between November 2016 and December 2024. Demographic, clinical, radiologic, and genetic data were extracted; white-matter (WM) disease severity and imaging course were graded, and associations with genotype were explored.

Results: Fifteen individuals (73.3% male; 80% consanguinity) were identified. Symptom onset occurred in the infantile age in 86.7%, with developmental delay or regression (100%), intellectual impairment (76.9%), and impaired motor function (69.2%) predominating. Two siblings with SAMHD1-homozygous variant were neurologically normal but had chronic arthritis. Genetic variants were identified in RNASEH2B (26.7%), RNASEH2A and TREX1 (each 20%), followed by ADAR1 and SAMHD1 (each 13.3%), and RNASEH2C (6.7%); 80% of variants were homozygous. Radiological assessment was suggestive of the disease in 40% of the cases. Initial neuroimaging revealed focal WM disease in 76.9%, calcifications in 53.8%, hypomyelination in 38.5%, and basal-ganglia involvement in 30.8%. WM disease was absent (13.3%), mild (40.0%), moderate (40%), or severe (6.7%); follow-up imaging showed stable (10%), regressive (30%), progressive (40%), or progressive-then-regressive (20%) course of imaging findings, independent of the genotype and clinical course. Genotype correlated significantly with WM disease severity at presentation (p < 0.05), but not with longitudinal imaging trajectory. Intrafamilial radiological discordance was observed in half of the family clusters.

Conclusions: This AGS series from a Middle-Eastern population broadens the phenotypic spectrum, highlighting high homozygosity and RNASEH2 predominance in a highly consanguineous population, a significant genotype-WM disease severity link, and radiologic variability independent of genotype or clinical course. Awareness of these patterns may inform population-based testing and management strategies.

Abbreviations: AGS ＝Aicardi-Goutières syndrome; VUS ＝ variant of uncertain significance.

Background and purpose: Sacral dural tears are an underrecognized cause of spontaneous intracranial hypotension, and their clinical behavior and response to treatment remain incompletely defined. We hypothesized that outcomes following epidural patching in sacral dural tears are driven primarily by disease chronicity and baseline imaging features, mirroring patterns observed in other SIH leak subtypes, and are less dependent on procedural variables.

Materials and methods: We performed a multicenter retrospective cohort study of patients with SIH attributed to sacral dural tears who underwent epidural patching and had clinical and/or imaging follow-up. Of 61 identified patients, 54 met inclusion criteria. Clinical outcomes, brain and spine imaging findings, and procedural variables were analyzed. Univariate and multivariable logistic regression models were used to identify predictors of clinical and imaging outcomes. Myelographic techniques were compared for rates of precise leak localization.

Results: The mean age was 37.8 years, and 85% of patients were women. Complete clinical improvement following patching occurred in 28 of 50 patients (56%), and complete resolution of sacral extradural CSF on follow-up spine MRI occurred in 17 of 37 patients (46%). Greater improvement in Bern score was independently associated with complete clinical improvement (OR 0.73, 95% CI 0.54-0.99, p = 0.04) and showed a pattern toward imaging resolution. Unorganized baseline extradural CSF morphology predicted better clinical (p = 0.03) and imaging (p = 0.03) outcomes. Procedural variables, including injectate type, volume, and needle approach were not associated with outcome. Dynamic CT myelography precisely localized the leak more frequently than digital subtraction myelography (75% vs 31%, p = 0.02). Eight patients underwent surgery, with mixed clinical and imaging outcomes.

Conclusions: Epidural patching outcomes in sacral dural tears causing SIH are driven primarily by baseline imaging phenotype, disease stage, and intracranial imaging response rather than procedural technique. These findings support a morphology- and chronicity-aware approach to diagnosis and treatment and suggest that sacral dural tears represent a distinct SIH subtype with outcome patterns similar to other leak mechanisms.

Background and purpose: Paramagnetic rim lesions (PRL) are a highly specific imaging biomarker for multiple sclerosis (MS) that are now integrated into the 2024 McDonald criteria. PRL can be detected on standard SWI, which combines data from the homodyne-filtered phase (phase) and enhanced magnitude (SWI) images, as well as on the phase images alone. However, the relative visibility of PRL on SWI versus phase images is unknown and it remains unclear which of these contrasts should be used for PRL evaluation in the clinical setting.

Materials and methods: The study sample (n=40) consisted of two cross-sectional cohorts-a primary cohort including both MS and non-MS cases (n=20) and a distinct secondary cohort consisting of MS participants only (n=20)-both selected from a prospective, multicenter observational study (CAVS-MS) of participants presenting for initial diagnostic evaluation of suspected MS. Standardized brain MRI sequences (3D T1WI with and without contrast, T2-FLAIR, and vendor-provided SWI) were acquired at 3T. T2-hyperintense lesions were evaluated on SWI and phase images for classification as PRL or non-PRL in separate trials by trained, blinded raters (two raters in the primary cohort, one rater in the secondary cohort). Differences in PRL frequency between SWI and phase images were analyzed with McNemar test in the primary, secondary, and pooled cohorts.

Results: 155 PRL were identified in the pooled cohort, of which 66 (42.6%) were detected only on phase images. The number of lesions classified as PRL was 2.7- and 1.4-fold greater when evaluated on phase images compared to SWI in the primary (43 vs. 16, p<0.001) and secondary (109 vs. 73, p<0.001) cohorts, respectively, and 1.7-fold greater in the pooled cohort (152 vs. 89, p<0.001). The main reasons for non-PRL classification on SWI were nodular appearance (37.9%) and discontinuity of the rim (28.8%).

Conclusions: Filtered phase images derived from standard SWI outperform enhanced magnitude images for PRL identification and should be primarily used for PRL evaluation in clinical practice.

Background and purpose: Collateral circulation is a key determinant of treatment response and outcomes in acute ischemic stroke (AIS), yet its assessment in clinical practice remains limited and subjective. While CT perfusion (CTP) offers insight into tissue viability, its restricted availability and susceptibility to artifacts reduce its practical utility, particularly in smaller centers. As an accessible alternative, we developed and validated an automated quantitative collateral index (qCI) derived from CTA using a deep learning U-Net segmentation framework, and evaluated the ability of CTA-based features to predict post-stroke recovery and functional outcomes.

Materials and methods: We retrospectively analyzed prospectively collected data from 230 AIS patients who underwent endovascular thrombectomy (EVT) between 2019-2023. CTA scans were segmented using a validated neural network-based vascular extraction model to generate 3D vessel networks and compute morphology metrics (vessel length, branching, fractal dimension, tortuosity). A fully automated qCI was derived through hemispheric comparison of vascular features following spatial registration. Agreement of qCI with clinician grading was quantified. Gradient-boosted decision tree models were trained to predict early neurological deterioration (END), early neurological improvement (ENI), and 90-day modified Rankin Scale (mRS) using (i) CTP-only (core, penumbra, mismatch), (ii) CTA-only (qCI + morphology), and (iii) combined CTA+CTP features.

Results: Automated qCI (graded 0-3) showed strong concordance with expert scoring (accuracy 0.863; Pearson R = 0.880; Cohen's κ = 0.786). Dichotomized collateral status achieved 0.938 accuracy (AUROC = 0.945). For 90-day mRS prediction, the CTA-only model outperformed the CTP-only model (AUROC 0.730 vs 0.645) with better calibration (Brier score 0.178 vs 0.295). The combined CTA+CTP model performed best overall (AUROC 0.781), with similar improvements observed for END. CTA-derived features led to significant reclassification gains when added to perfusion-based models.

Conclusions: Automated CTA-derived qCI and cerebrovascular morphology provide rapid, objective, and reproducible collateral assessment with high agreement to expert grading. These features outperform perfusion metrics in several predictive tasks and further enhance prognostic accuracy when combined with CTP. Because CTA is widely available, qCI offers a scalable, clinically practical tool for improving stroke outcome prediction, particularly in settings where CTP is unavailable.

Background: Contrast-enhanced 3D T1-weighted MRI is the imaging reference for detection and follow-up of brain metastases. Volumetric GRE-based sequences, such as MPRAGE, are widely used but remain prone to susceptibility and lower lesion conspicuity. 3D black-blood TSE-based sequences, such as Sampling Perfection with Application-Optimized Contrasts by using different flip angle Evolutions (SPACE), have been increasingly embedded into routine workflow and are thought to improve lesion detection in part through vessel signal suppression.

Purpose: We aimed to investigate the comparative diagnostic performance of 3D T1 TSE versus GRE sequences for the detection of brain metastases.

Data sources: Studies comparing the diagnostic performance of postcontrast 3D T1 SE and GRE sequences in adults with brain metastases were searched on MEDLINE, EMBASE, Cochrane Central, Google Scholar, and PROSPERO, from inception through April 2025.

Study selection: Fifteen studies encompassing 544 patients with 4338 metastases were included.

Data analysis: Data on diagnostic accuracy parameters, image quality, and inter-rater agreement were extracted. Random-effects models were applied to compute pooled sensitivity and comparative OR for lesion detection. Risk of bias was assessed using QUADAS-2 and QUADAS-C tools.

Data synthesis: Pooled sensitivities for detection of brain metastases were 97.4% (95%CI, 93.2%-99.0%) for TSE and 76.1% (95%CI, 69.3-81.9) for GRE-based sequences, with a comparative OR of 12.0 (95%CI, 5.45-26.6, P <.0001). Detectability of small lesions (<5 mm) was significantly better on TSE (96.1%; 95%CI, 87.7-98.8) than GRE (58.4%; 95%CI, 47.9-68.2), while both techniques performed comparably for larger (≥5 mm) lesions (98.2% for TSE, 94.4% for GRE). OR estimates were 17.2 (95%CI, 4.50-66.1) for small and 2.81 (95%CI, 0.92-8.56) for large lesions. Contrast-to-noise-ratio and inter-rater agreements were slightly higher on TSE than GRE. False positives were more common with TSE, mostly related to incomplete vessel suppression (49 FP counts in TSE, 35 in GRE).

Limitations: Our meta-analysis is limited by high heterogeneity, case-only studies, possible small-study effects, and high risk of bias for the reference standard domain.

Conclusions: Postcontrast 3D T1 TSE sequences provide higher sensitivity and improved lesion conspicuity compared with GRE sequence, particularly for small metastases, though at the cost of slightly higher false positives.