Neuroradiology最新文献

Purpose: Accurate assessment of midline shift (MLS) is critical in the management of patients with acute intracerebral hemorrhage. Manual estimation of MLS by radiologists is time-consuming and subject to inter-observer variability. We present a lightweight, reproducible deep learning model designed for real-time, automated MLS quantification using non-contrast head CT scans.

Methods: We propose a binarized convolutional neural network based on a residual U-Net architecture that dramatically reduces model parameters (from 31 million to 44 thousand) using XNOR-based binarization and channel-wise scaling. The model segments the cerebral hemispheres and extracts midline features to quantify MLS. Post-processing involves the derivation of the deformed midline (dML) using edge detection and an anatomically guided ideal midline (iML) for reference. MLS is calculated as the maximum horizontal displacement between these lines.

Results: Evaluated on the RSNA 2019 hemorrhage CT dataset (held-out test: 5,000 slices), the model achieved Dice scores of 0.92 (left) and 0.91 (right) and a mean absolute error (MAE) of 0.09 mm for MLS. On an NVIDIA GTX 1650 (4 GB), median inference time was 5.6 ms per slice for the segmentation step; all physical measurements were computed at native DICOM spacing.

Conclusion: This clinically-driven, resource-efficient model enables accurate MLS quantification suitable for emergency neuroimaging settings. Its reproducibility and low computational footprint make it ideal for deployment in low-resource environments, providing radiologists with rapid, decision-support insights. This lightweight model may assist accurate MLS quantification in time-sensitive settings while operating within constrained compute budgets. Further multicenter validation is warranted.

Objective: To compare short (TR = 7000 ms) and long (TR = 10000 ms) repetition time 3D-real inversion recovery (IR) sequences with 3D-FLAIR MRI in terms of image quality and diagnostic performance for evaluating endolymphatic hydrops (EH) after a single intravenous dose of gadobutrol in patients with unilateral Ménière's disease.

Materials and methods: Thirty-seven patients with definite unilateral Ménière's disease underwent inner ear MRI 4 h after administration of gadobutrol (0.1 mmol/kg). Each patient was scanned using 3D-real IR sequences with two different TR values (7000 ms and 10000 ms), as well as a conventional 3D-FLAIR sequence. Quantitative image analysis included contrast-to-noise ratio (CNR) and signal-to-noise ratio (SNR), while qualitative scoring and inter-/intra-observer agreement were also assessed.

Results: The short TR (7000 ms) 3D-real IR sequence failed to demonstrate adequate perilymphatic enhancement in most cases. In contrast, both the long TR (10000 ms) 3D-real IR and 3D-FLAIR sequences consistently visualized perilymphatic enhancement and EH. The TR = 10,000 ms sequence provided significantly higher image quality metrics (CNR and SNR, P < 0.001), particularly improving detection in the cochlear apex. It also showed better inter-observer agreement than 3D-FLAIR.

Conclusion: The 3D-real IR sequence with a TR of 10,000 ms offers superior image quality and enhanced diagnostic confidence for EH evaluation compared to both the shorter TR sequence and conventional 3D-FLAIR. These findings support the use of a longer TR protocol for inner ear MRI following single-dose gadolinium administration in clinical settings.

Purpose: To investigate the magnetic resonance imaging (MRI) characteristics of the trigeminal nerve in Fabry disease (FD) by comparing the morphology of the trigeminal nerve in patients with and without FD.

Methods: This retrospective study included 40 patients with FD and 40 age- and sex-matched controls who underwent 3T MRI with constructive interference in steady-state sequence. Two neuroradiologists measured the short-axis length on axial and oblique coronal images and the long-axis length and cross-sectional area on oblique coronal images of the cisternal segment of the trigeminal nerve bilaterally. Morphological differences were assessed between groups and between sides within each group using appropriate t-tests.

Results: All measurements were significantly smaller in the FD group than in the control group: axial images: short-axis length (3.20 ± 0.52 mm vs. 3.51 ± 0.51 mm, p = 0.0101); oblique coronal images: short-axis length (2.41 ± 0.28 mm vs. 2.61 ± 0.35 mm, p = 0.007); long-axis length (3.75 ± 0.53 mm vs. 4.14 ± 0.44 mm, p < 0.001); and cross-sectional area (7.71 ± 1.68 mm² vs. 8.93 ± 1.30 mm², p < 0.001). Left-right differences were observed in both groups; the left side was generally larger. Receiver operating characteristic curve analysis showed moderate diagnostic performance (area under the curve, 0.66-0.72).

Conclusion: The trigeminal nerves in the cisternal region were significantly smaller in patients with FD than in those without FD. These findings may support its potential utility as an imaging biomarker.

Purpose: To investigate the morphological variability of the posterior inferior cerebellar artery (PICA) origin utilizing computed tomography angiography (CTA) within a substantial adult cohort, and to establish a refined, clinically pertinent classification system for neuroradiologists and neurosurgeons.

Materials and methods: A retrospective review of 500 CTA brain scans, comprising 1,000 arteries, was undertaken. The origin, pathway, and morphological variants of the PICA were systematically documented and classified according to segmental origin and anatomical characteristics.

Results: The PICA most commonly originates from the intradural (V4) segment of the vertebral artery (VA) (63.4%), followed by the extradural (V3) segment (16.4%) and the basilar artery (BA) (15.8%). Bilateral origins from the same segment were seen in 41.6% (V4), 4.4% (V3), and 3.2% (BA). The absence of PICA was found in 4.4% of cases. In these cases, the contralateral PICA was enlarged and compensated for the absent artery. Rare variants included duplicate origin (0.1%), duplication (0.3%), VA terminating in PICA (PICA-VA) (3.4%), and intradural origin with an abnormal course below the foramen magnum (1.2%).

Conclusion: This study provides a thorough radiological classification of the origin and morphology of the PICA, offering significant insights into preoperative planning, endovascular access planning, and neurovascular risk assessment. Recognizing these anatomical variations is crucial for enhancing diagnostic accuracy, surgical safety, and interventional strategies. Future multicenter investigations are recommended to validate and further develop this classification system.

Purpose: Visible perivascular spaces of cerebral white matter at magnetic resonance imaging (MRI) were in several studies proposed to be an integral part of brain-wide perivascular clearance pathways, and their enlargement could therefore serve as markers of perivascular clearance dysfunction. We studied whether MRI-visible perivascular spaces in subcortical white matter communicate with subarachnoid cerebrospinal fluid (CSF).

Methods: MRI-visible perivascular spaces of the basal ganglia served as controls. Intrathecal 0.5 mmol gadobutrol was utilized as CSF tracer, and T1-weighted MRI was performed before, and at multiple time points after (3, 6, 24 and 48 h) injection. Perivascular spaces with diameter ≥ 2 mm were included in the analysis, and a circular region of interest was placed manually within one perivascular space and in adjacent brain parenchyma of each region.

Results: The study included 27 symptomatic individuals undergoing clinical work-up of various CSF circulation disorders, but in whom no treatable condition was found. Perivascular spaces of both white matter and basal ganglia enhanced with intrathecal gadobutrol, confirming CSF exchange with perivascular spaces. While perivascular spaces of basal ganglia enhanced most with peak at 6 h (233.2% [34.9 to 431.5%]) (p < 0.01), coinciding with peak enhancement in subarachnoid CSF, perivascular spaces of white matter enhanced less and more slowly with peak at 48 h (159.1% [37.9 to 280.3%]) (p < 0.01).

Conclusions: The various degrees of CSF exchange with MRI-visible subcortical perivascular spaces suggest these may be dilated for different reasons, therefore questioning their validity as markers of perivascular clearance function.

Background: Aneurysm wall enhancement (AWE) is a potential imaging biomarker that can be used to identify intracranial aneurysms with an increased risk of rupture. Factors and mechanisms causing intracranial aneurysms to rupture or progress differ between women and men. Thus, we performed a cross-sectional and prospective cohort study to identify the risk factors predicting AWE in men and women.

Methods: Consecutive patients with unruptured intracranial aneurysms were prospectively recruited from September 2022 to September 2024. Baseline characteristics were collected through a standard questionnaire. Multivariate logistic regression was performed to assess the predictors of aneurysmal wall enhancement in men and women.

Results: A total of 332 patients with 435 intracranial aneurysms were included. In the multiple logistic regression analysis, obesity (OR = 2.88, 95% CI 1.06-7.81, P = 0.038), the history of hypertension (OR = 2.76, 95% CI 1.16-6.57, P = 0.022) and younger age (OR = 0.96, 95% CI 0.93-1.00, p = 0.035) were independent risk factors for AWE in men. Meanwhile, drinking habit (OR = 2.64, 95% CI 1.09-6.41, P = 0.032), the family history of intracranial aneurysm (OR = 4.01, 95% CI 1.21-13.36, P = 0.024) and older age (OR = 1.04, 95% CI 1.01-1.07, P = 0.009) were independent risk factors for AWE in women.

Conclusion: This study clearly demonstrates a significant sexual dimorphism in the risk factors associated with AWE. These findings underscore the necessity for sex-specific risk stratification models for intracranial aneurysm management, suggesting that men and women have specific lifestyle interventions to prevent the progression or rupture of intracranial aneurysms.

Background and purpose: Imaging findings of otosclerosis have been previously described using traditional energy integrated detector CT scanners. Recent developments in photon counting detector CT allow improved spatial and contrast resolution over previous technology.

Methods: A retrospective observational case series review was carried out in 53 patients (106 temporal bones) who had imaging findings of otosclerosis. Imaging findings of otosclerosis were documented, including specific plaque location, the presence of otic capsule expansion, and IAC diverticula. The Symons and Fanning classification was used to categorize findings. The presence of the cochlear cleft was documented. 50 asymptomatic patients were reviewed to determine the presence of features that could be confounded for otosclerosis.

Results: Per the Symons and Fanning classification, the following types were found in individual temporal bones: type 0-13, type 1-48, type 2a - 12, type 2b - 6, type 2c - 16, type 3-4. Seven patients had imaging findings that did not fit the classification criteria, primarily isolated otospongiotic plaques in other parts of the otic capsule. The fissula ante fenestram was involved in 91 otosclerosis and 4 temporal bones in asymptomatic patients. The cochlear cleft was visible in 3 otosclerotic and 10 normal temporal bones. Otic capsule thickness in symptomatic patients: concave - 52, flat - 27, convex - 24. In asymptomatic patients: concave - 98, flat - 2, convex - 0. IAC diverticula: present in 36% of otosclerosis and 2% of normal. Isolated foci of otospongiosis were identified in several patients without classic fenestral or retrofenestral disease.

Conclusions: Photon counting CT allows excellent visualization of otospongiotic plaques. Due to the high resolution, subclinical findings can be overlooked in asymptomatic patients and plaque may be visualized in the otic capsule that are not part of existing classification systems. The current study coincides with previous studies which demonstrate that otosclerosis may occur in any part of the otic capsule.

Background and purpose: Internal carotid artery stenosis (ICS) is a major cause of ischemic stroke, requiring accurate evaluation of stenosis severity and cerebral hemodynamics to guide treatment strategies. This study aimed to assess the clinical utility of a novel four-dimensional carotid blood flow laterality (4D-CBL) scale, derived from four-dimensional pseudo-continuous arterial spin labeling MRA with Keyhole and View-sharing (4D-PACK), in evaluating the severity of ICS compared with DSA and SPECT.

Materials and methods: Forty patients with unilateral cervical ICS were prospectively enrolled. Patients were classified into three stages according to the 4D-CBL scale: stage 0 (no blood-flow delay), stage 1 (delayed ipsilateral perfusion), and stage 2 (collateral-dependent circulation). The severity of stenosis was determined by NASCET criteria on DSA and compared with 4D-CBL scale. Hemodynamic status was assessed by interhemispheric CBF ratios derived from SPECT. Interobserver agreement for 4D-CBL scale was evaluated using weighted Cohen's κ statistics.

Results: Among the 40 patients, 10 were stage 0, 23 stage 1, and 7 stage 2. The median degree of stenosis significantly increased with higher 4D-CBL stages (58.3% in stage 0, 71.0% in stage 1, and 95% in stage 2; p < 0.001). The median CBF ratios decreased significantly across stages (0.98, 0.97, and 0.93; p = 0.037). Interobserver agreement was excellent (concordance rate = 0.85, κ = 0.82; 95% CI 0.68-0.96).

Conclusions: The 4D-CBL scale derived from 4D-PACK enables noninvasive and reliable assessment of ICS severity and hemodynamic compromise without the need for contrast medium or radiation exposure. It demonstrates correlation with DSA and SPECT findings, suggesting its potential role as a practical alternative or adjunct for evaluating ICS severity.