Clinical Neuroradiology最新文献

Objectives: Determining the involvement of specific peripheral nerves (PNs) in the upper limb associated with signs of muscle denervation can be challenging. This study aims to develop, compare, and validate various large language models (LLMs) to automatically identify and establish potential relationships between denervated muscles and their corresponding PNs.

Materials and methods: We collected 300 retrospective MRI reports in Spanish from upper limb examinations conducted between 2018 and 2024 that showed signs of muscle denervation. An expert radiologist manually annotated these reports based on the affected peripheral nerves (median, ulnar, radial, axillary, and suprascapular). BERT, DistilBERT, mBART, RoBERTa, and Medical-ELECTRA models were fine-tuned and evaluated on the reports. Additionally, an automatic voting system was implemented to consolidate predictions through majority voting.

Results: The voting system achieved the highest F1 scores for the median, ulnar, and radial nerves, with scores of 0.88, 1.00, and 0.90, respectively. Medical-ELECTRA also performed well, achieving F1 scores above 0.82 for the axillary and suprascapular nerves. In contrast, mBART demonstrated lower performance, particularly with an F1 score of 0.38 for the median nerve.

Conclusions: Our voting system generally outperforms the individually tested LLMs in determining the specific PN likely associated with muscle denervation patterns detected in upper limb MRI reports. This system can thereby assist radiologists by suggesting the implicated PN when generating their radiology reports.

Purpose: This study aimed to evaluate the diagnostic performance of 3D T1-weighted black-blood (T1W BB) MRI compared to 3D T1-weighted turbo field echo (T1-TFE) in diagnosing facial neuritis (FN) and to investigate its role in disease monitoring.

Materials and methods: 22 patients with acute idiopathic FN were included in this prospective study. All patients underwent MRI within the first week of clinical presentation including 3D T1W BB and 3D T1-TFE sequences. Two neuroradiologists independently analyzed six facial nerve segments, evaluating contrast enhancement using a three-point grading scale (0-2). Diagnostic accuracy, sensitivity, specificity, and area under the curve (AUC) were compared between the two sequences. Follow-up MRI was performed in 8 patients to monitor temporal changes in nerve enhancement, and these findings were analyzed in relation to House-Brackmann (HB) scores.

Results: The sensitivity, specificity, and accuracy for FN detection were 97.7%, 93.2%, and 95.5% for 3D T1W BB, compared to 86.4%, 97.7%, and 92% for 3D T1-TFE, respectively. Sensitivity was significantly higher with 3D T1W BB (p < 0.05), while AUCs were higher but not significant for both readers. Mean enhancement grades in all affected nerve segments were significantly higher on 3D T1W BB (p < 0.05). Follow-up imaging showed enhancement reduction in 87.5% of patients, correlating with HB score improvements. Enhancement grades significantly correlated with HB scores on T1W BB but not on T1-TFE.

Conclusion: 3D T1W BB has comparable diagnostic performance with 3D T1-TFE for diagnosing FN and can be used as an effective tool in confirming the diagnosis and in follow-up.

Background: Carotid artery stenting (CAS) using the 7F Optimo balloon guide catheter (BGC) allows for smooth navigation and facilitates proximal flow control. However, this method may allow antegrade flow in the internal carotid artery (ICA). This study aims to identify predictors of antegrade flow during CAS with the common carotid artery (CCA) occlusion.

Methods: We retrospectively analyzed 102 lesions treated with CAS using the 7F Optimo BGC and distal filter protection. The ICA flow pattern was assessed via contrast injection during CCA occlusion.

Results: Antegrade flow in the ICA was observed in 22 lesions (22%). Compared with lesions where ICA flow control (stagnation or reverse flow) was achieved, the external carotid artery (ECA) diameter was significantly larger (4.4 ± 0.7 mm vs. 3.6 ± 1.2 mm, p < 0.001), and the minimum lesion diameter was significantly larger (3.2 ± 1.2 mm vs. 2.1 ± 1.0 mm, p < 0.001). Multivariate analysis identified a minimum lesion diameter ≥ 2.1 mm (OR 4.8, 95% CI 1.44-16.1; p = 0.01) and an ECA diameter ≥ 4.2 mm (OR 3.2, 95% CI 1.08-9.09; p = 0.04) as independent predictors of antegrade flow. High-intensity spots in postoperative diffusion-weighted magnetic resonance imaging and the incidence of ischemic events were not significantly different between both groups.

Conclusions: Lesions with a minimum lesion diameter ≥ 2.1 mm or an ECA diameter ≥ 4.2 mm may exhibit antegrade ICA flow with BGC regardless of CCA occlusion, suggesting that an additional distal filter may help reduce embolic risk.

Purpose: The central vein sign (CVS) is a promising imaging biomarker for multiple sclerosis (MS) diagnosis. While isotropic T2* at 3 T and 7 T has demonstrated high diagnostic performance, its utility at 1.5 T remains unclear. This study evaluates the performance of unenhanced FLAIR-T2* fusion at 1.5 T compared to 3 T in MS participants.

Methods: This prospective observational study included 20 MS patients and 20 control subjects. Each participant underwent unenhanced isotropic Epi-T2* and isotropic FLAIR (0.8 mm voxel size) at both 1.5 T and 3 T. Subsequently, the derived isotropic T2* and FLAIR were combined to create the final FLAIR-T2* fusion in both magnetic field strengths. Two independent raters assessed the CVS status of white matter (WM) lesions using NAIMS criteria. WM lesions were classified as CVS+ or CVS-, and two methods-select-n* and CVS+ proportion-were applied. Sensitivity and specificity were computed, and CVS performance was compared across WM lesion locations.

Results: Among eligible WM lesions (MS: 258; controls: 255), the mean CVS+ lesion proportion per participant was 66.9 ± 15.4% for 1.5 T FLAIR-T2* and 77.0 ± 13.6% for 3 T FLAIR-T2* (p < 0.01). At a 40% threshold, 1.5 T FLAIR-T2* achieved 90% sensitivity and 95% specificity, while 3 T FLAIR-T2* achieved 100% sensitivity and 95% specificity. The Select-6* method resulted in only one MS patient being misclassified at both field strengths. 3 T FLAIR-T2* detected more CVS+ lesions in deep WM (87.5% vs. 57.1%, p = 0.05).

Conclusion: 1.5 T FLAIR-T2* fusion demonstrates high performance in CVS assessment, although slightly outperformed by 3 T FLAIR-T2*. The select-6* method may enhance 1.5 T performance, supporting its feasibility for CVS evaluation.

Purpose: Visualizing the culprit perforating artery in subcortical infarction using in vivo imaging is challenging. We aimed to identify the culprit perforating arteries in subcortical infarctions and assess their morphology using an image fusion technique.

Methods: We retrospectively reviewed consecutive patients who had an ischemic stroke in the anterior circulation perforating area (caudate nucleus, lentiform nucleus, internal capsule, corona radiata, or centrum semiovale) and underwent three-dimensional rotational-angiography (3D-RA) and 3D fluid-attenuated inversion recovery MRI. Images were registered using an original fusion software. The spatial relationship between the infarction and culprit perforating artery and its morphological characteristics were analyzed in the fusion images. Stenosis was defined as > 50% luminal narrowing or a focal intraluminal defect in the perforating artery.

Results: Of 118 patients, the culprit perforating artery was identified in 52 patients (44%); They tended to have younger age and had a higher baseline NIHSS score and higher prevalence of infarcts in the lentiform nucleus than did those without identified culprit perforating artery. Among the 44 patients with assessable morphology of the culprit perforating artery, 27 (61%) exhibited stenosis in the proximal segment. Atrial fibrillation was more frequent in patients without stenosis in the proximal segment of the culprit perforating artery than in those with stenosis (29% vs. 4%, P = 0.03).

Conclusion: The 3D-RA and MRI fusion technique enables identification of the culprit perforating arteries in subcortical infarctions, especially in the lentiform nucleus. Morphological features of the culprit perforating artery may be associated with the etiological mechanism of stroke.

Background: The benefits of flow diversion (FD) versus endosaccular coiling (EC) in small and medium-sized proximal saccular aneurysms of anterior circulation, amenable to both methods, remain not well-established. This study aimed to compare the efficacy and safety of FD vs EC and evaluate the triggers of treatment modality choice in the aforementioned setting.

Methods: The study had a retrospective design. Raw sample included 154 FD and 190 EC consecutively enrolled cases. All aneurysms were saccular, sized 4-14 mm, located at intradural ICA or A1/M1, not acutely ruptured and previously untreated. Median lesion size was 6.6 mm (90% < 10 mm), median neck diameter‑3.7 mm. Matched cohorts were 67 cases each, PS difference < 0.1 probit SD, all covariates: size, neck, location, shape, angle, multiplicity, rupture history, age, and incorporated arterial branch. Follow-up DSA available for 94.2% of raw sample at median 9 months.

Results: In both raw and matched samples respectively, FD versus EC demonstrated higher rates of target aneurysm total occlusion (76.4% vs 53.2%, p < 0.001 and 80.3% vs 49.2%, p < 0.001), lower of rates of intraoperative technical adverse events (7.3% vs 21.4%, p < 0.001 and 9% vs 22.4%, p = 0.032) and retreatment (raw‑2.6% vs 15.4%, p < 0.001, matched‑1.5% equal). Rates of neurological morbidity and death were similar. The choice of FD was triggered by neck size and ratio, paraophthalmic location (OR = 2.57), multiplicity (OR = 4.1) and incorporated arterial branch (OR = 4.82), p < 0.001. Incorporated branch was associated with incomplete occlusion in both treatment modalities (p < 0.01).

Conclusion: In this study FD demonstrated higher rates of target aneurysm total occlusion, lower rates of intraoperative technical adverse events and similar rates of neurological morbidity and death, compared to EC.

Purpose: Photon-Counting Detector CT is characterized by enhanced image post-processing capabilities. The diagnostic accuracy of PCD-CT angiography (PCD-CTA) in assessing intracranial stents (ICS) and flow diverters (FD) has yet to be compared with digital subtraction angiography (DSA).

Methods: Retrospective analysis of all consecutive patients who underwent ICS or FD implantation between April 2023 and May 2024. Polyenergetic images, along with virtual monoenergetic imaging (VMI), pure lumen (PL) and iodine (IOD) reconstructions were assessed by three readers using a 5-point Likert scale and defined regions of interest (ROIs). A blinded analysis was performed to identify relevant lumen reduction. The diagnostic accuracy of PCD-CTA was compared to DSA by calculating the area under the receiver operating characteristic curve.

Results: A total of 18 patients (mean age 59 ± 13 years; 14 women) with 14 implanted ICS and 10 FD were analyzed. Across all pairwise comparisons, pooled VMI reconstructions demonstrated higher ratings and signal-to-noise ratios compared to IOD, PL and UHR reconstructions (p < 0.001 for all comparisons). In the pooled assessment of DSA of in-stent vessel lumen 18 (11%) of the 162 vessel segments and 6 (33%) of the 18 patients presented relevant narrowing of the in-stent vessel lumen. The sensitivity of PCD-CTA for detecting stenosis was 100% (18 of 18 in-stent vessel segments), while specificity was 89% (128/144 in-stent vessel segments). All readers reported a 100% negative predictive value (128/128 in-stent vessel segments).

Conclusion: Photon-Counting Detector CT might provide a reliable assessment of intracranial vessels following stent or flow diverter implantation comparable to DSA in many cases.

Background: The anatomy of the proximal anterior cerebral artery (ACA) and its branches, including the recurrent artery of Heubner (RAH) and medial lenticulostriate arteries (MLSAs), is known for frequent variations. Impairment of these branches can result in severe consequences, including neurological deficits or cognitive impairment. This study aimed to analyze these branches and their variations in vivo, using data from 3D rotational angiographies (3D-RA).

Material and methods: We reviewed 3D-RAs of 209 hemispheres from 191 patients with pathologies remote from the region of interest. The presence, origin and course of the RAH were investigated. Delineation, origin and number of perforator vessels originating from the A1 segment (MLSAs) were assessed.

Results: The RAH was observed in 151 hemispheres (72%), including a single RAH in 144 (69%) and a doubled RAH in 7 (3%) hemispheres. In 37% of cases, the RAH originated from the A1 segment, in 56% from the A1/A2 transition, and in 7% from the A2 segment. In the presence of RAH, additional MLSAs originating from the A1 segment were present in 25% of hemispheres. A weak negative correlation was identified between the presence of one or two RAH and the frequency of additional MLSAs.

Conclusion: The study revealed significant differences in the presence and anatomical course of RAH and MLSAs compared to previous research. The findings highlight the importance of 3D-RA in visualizing the complex anatomy of the ACA, particularly to avoid complications during surgical or endovascular procedures.

Purpose: To investigate the shared and distinct alterations in hippocampal subfield atrophy and functional connectivity (FC) in Parkinson's disease (PD) with normal cognition (PD-NC), PD with mild cognitive impairment (PD-MCI) and unspecified MCI (U-MCI).

Methods: A total of 294 participants were included from two centers (30 PD-NC, 86 PD-MCI, 76 U-MCI, and 102 healthy controls (HC)). Comparisons of hippocampal subfield volumes were conducted among groups. Seed-based FC was calculated to assess abnormalities between hippocampal subfields and cortical regions.

Results: Compared to HC, PD-NC group showed volumetric reductions in the right cornu ammonis (CA) 2/3, granule cell layer of the dentate gyrus (GC-DG), and CA4 subfields, while PD-MCI group exhibited bilateral volumetric reductions in the same subfields. PD-MCI patients exhibited increased FC between the bilateral GC-DG/CA4 subfield and the posterior default mode network (pDMN), as well as between the right GC-DG/CA4 subfield and the calcarine, in comparison to PD-NC. U‑MCI patients displayed smaller volume in the right CA4 compared to HC. Decreased FC of the hippocampus with the widespread visual network was observed in the PD-MCI group compared to the U‑MCI group.

Conclusions: PD-MCI is associated with structural vulnerability of hippocampal subfields in the CA2/3, GC-DG, and CA4 subfields, impacting FC with pDMN and visual network. Smaller scope of hippocampal subfields atrophy but weaker hippocampus-visual network FC abnormalities in U‑MCI relative to PD-MCI may suggest distinct progression mechanisms between these two conditions.