Clinical Neuroradiology最新文献

Purpose: To compare the diagnostic performance among Generative Pre-trained Transformer (GPT)-4-based ChatGPT, GPT‑4 with vision (GPT-4V) based ChatGPT, and radiologists in challenging neuroradiology cases.

Methods: We collected 32 consecutive "Freiburg Neuropathology Case Conference" cases from the journal Clinical Neuroradiology between March 2016 and December 2023. We input the medical history and imaging findings into GPT-4-based ChatGPT and the medical history and images into GPT-4V-based ChatGPT, then both generated a diagnosis for each case. Six radiologists (three radiology residents and three board-certified radiologists) independently reviewed all cases and provided diagnoses. ChatGPT and radiologists' diagnostic accuracy rates were evaluated based on the published ground truth. Chi-square tests were performed to compare the diagnostic accuracy of GPT-4-based ChatGPT, GPT-4V-based ChatGPT, and radiologists.

Results: GPT‑4 and GPT-4V-based ChatGPTs achieved accuracy rates of 22% (7/32) and 16% (5/32), respectively. Radiologists achieved the following accuracy rates: three radiology residents 28% (9/32), 31% (10/32), and 28% (9/32); and three board-certified radiologists 38% (12/32), 47% (15/32), and 44% (14/32). GPT-4-based ChatGPT's diagnostic accuracy was lower than each radiologist, although not significantly (all p > 0.07). GPT-4V-based ChatGPT's diagnostic accuracy was also lower than each radiologist and significantly lower than two board-certified radiologists (p = 0.02 and 0.03) (not significant for radiology residents and one board-certified radiologist [all p > 0.09]).

Conclusion: While GPT-4-based ChatGPT demonstrated relatively higher diagnostic performance than GPT-4V-based ChatGPT, the diagnostic performance of GPT‑4 and GPT-4V-based ChatGPTs did not reach the performance level of either radiology residents or board-certified radiologists in challenging neuroradiology cases.

Background: Endovascular thrombectomy (EVT) has shown promise in randomized controlled trials (RCTs) for large ischemic core stroke patients, yet variability in core definition and onset-to-imaging time creates heterogeneity in outcomes. This study aims to clarify the prevalence and implications of core-perfusion mismatch (MM) versus no mismatch (No MM) in such patients, utilizing established imaging criteria.

Methods: A retrospective cohort study was conducted including patients from 7/29/2019 to 1/29/2023, with data extracted from a continuously maintained database. Patients were eligible if they met criteria including multimodal CT imaging performed within 24 h from last known well (LKW), AIS-LVO diagnosis, and ischemic core size defined by specific rCBF thresholds. Mismatch was assessed based on different operational definitions from the EXTEND and DEFUSE 3 trials.

Results: Fifty-two patients were included, with various time windows from LKW. Using EXTEND criteria, a significant portion of early window patients exhibited MM; however, fewer patients met MM criteria in the late window. Defining MM using DEFUSE 3 criteria yielded similar patterns, but with overall lower MM prevalence in the late window. When employing rCBF <38% as a surrogate for ischemic core, a higher percentage of patients were classified as MM across both time windows compared to rCBF <30%.

Conclusion: The prevalence of MM in large ischemic core patients varies significantly depending on the imaging criteria and time from LKW. Notably, MM was more prevalent in the early time window across all criteria used. Additional RCTs are needed to determine if this definition of MM identifies patients who will benefit most from EVT.

Purpose: Subarachnoid haemorrhage is a potentially fatal consequence of intracranial aneurysm rupture, however, it is difficult to predict if aneurysms will rupture. Prophylactic treatment of an intracranial aneurysm also involves risk, hence identifying rupture-prone aneurysms is of substantial clinical importance. This systematic review aims to evaluate the performance of machine learning algorithms for predicting intracranial aneurysm rupture risk.

Methods: MEDLINE, Embase, Cochrane Library and Web of Science were searched until December 2023. Studies incorporating any machine learning algorithm to predict the risk of rupture of an intracranial aneurysm were included. Risk of bias was assessed using the Prediction Model Risk of Bias Assessment Tool (PROBAST). PROSPERO registration: CRD42023452509.

Results: Out of 10,307 records screened, 20 studies met the eligibility criteria for this review incorporating a total of 20,286 aneurysm cases. The machine learning models gave a 0.66-0.90 range for performance accuracy. The models were compared to current clinical standards in six studies and gave mixed results. Most studies posed high or unclear risks of bias and concerns for applicability, limiting the inferences that can be drawn from them. There was insufficient homogenous data for a meta-analysis.

Conclusions: Machine learning can be applied to predict the risk of rupture for intracranial aneurysms. However, the evidence does not comprehensively demonstrate superiority to existing practice, limiting its role as a clinical adjunct. Further prospective multicentre studies of recent machine learning tools are needed to prove clinical validation before they are implemented in the clinic.

Background: Recent advances have highlighted the efficacy of endovascular thrombectomy (EVT) in patients with large ischemic core stroke, yet a significant portion still experience very poor outcomes, defined as a 90-day modified Rankin Score (mRS) of 5-6. This study aims to investigate the hypoperfusion intensity ratio (HIR) as a prognostic imaging parameter for these outcomes.

Methods: In a multicenter retrospective cohort study, data from consecutive patients undergoing EVT for acute ischemic stroke with large vessel occlusion (AIS-LVO) at two comprehensive stroke centers were analyzed. The study included patients with an Alberta Stroke Program Early CT Score (ASPECTS) of 5 or less and utilized pretreatment perfusion imaging to calculate HIR. The primary outcome was very poor outcomes (90 days mRS 5-6).

Results: Among 102 patients included, 59 (57.8%) had very poor outcome (90 days mRS 5-6). Multivariable logistic regression analysis adjusting for multiple covariates including admission National Institutes of Health Stroke Scale (NIHSS) and EVT revealed that higher admission NIHSS (adjusted odds ratio [aOR] 1.224, 95% CI 1.089-1.374, p = 0.001) and HIR (aOR per 0.1 incremental change, 1.34, 95% CI 1.02-1.82, P = 0.042) were independently associated with very poor outcomes.

Conclusion: This study demonstrates that admission NIHSS and HIR are independently associated with very poor outcome (90 days mRS 5-6) in patients with large ischemic core strokes. These findings highlight the importance of collateral status and perfusion imaging in predicting outcomes in this patient population, suggesting a potential role for HIR in the triage and management of large core stroke patients.

Purpose: In acute ischemic stroke with large-vessel occlusion (LVO), collateral assessment with single-phase computed tomography angiography (CTA) might underestimate pial collateral supply in a considerable proportion of patients. We aimed to compare time-resolved magnetic resonance imaging (MRI)-based quantitative collateral mapping to conventional collateral imaging with CTA.

Methods: This retrospective single-center study covering a period of 6 years (2012-2018) included drip-and-ship LVO patients who underwent MR imaging after initial imaging evaluation with CT. For MRI-based collateral assessment, T2*-weighted time series from perfusion-weighted imaging (PWI) were processed to compute a quantitative collateral vessel index (CVI_PWI) based on the magnitude of signal variance across the entire acquisition time. CTA-based collateral scores (Tan and Maas) and CVI_PWI were investigated in terms of inter-modality associations between collateral measures, as well as their relationships with stroke severity, infarct volume and early functional outcome.

Results: The final analysis included n = 56 patients (n = 31 female, mean age 69.9 ± 14.21 years). No significant relationship was found between MR-based quantitative collateral supply (CVI_PWI) and CT-based collateral scores (r = -0.00057, p = 0.502 and r = -0.124, p = 0.797). In contrast to CVI_PWI, CTA-based collateral scores showed no significant relationship with clinical stroke severity and infarct volume. While MR-based CVI_PWI was independently associated with favorable early functional outcome in multivariate analysis (OR 1.075, 95% CI 1.001-1.153, p = 0.046), CTA-based collateral scores were not significantly associated with outcome.

Conclusions: Since collateral scores based on single-phase CTA do not accurately reflect infarct progression and might underestimate pial collateralization in a relevant proportion of patients, they are not associated with early functional outcome in LVO patients. In contrast, CVI_PWI represents a robust imaging parameter of collateral supply and is independently associated with functional outcome.

Purpose: To evaluate a novel flow-independent sequence (Relaxation-Enhanced Angiography without Contrast and Triggering (REACT)) for imaging of the extracranial arteries in acute ischemic stroke (AIS) at 1.5 T.

Methods: This retrospective single-center study included 47 AIS patients who received REACT (scan time: 3:01 min) and contrast-enhanced MRA (CE-MRA) of the extracranial arteries at 1.5 T in clinical routine. Two radiologists assessed scans for proximal internal carotid artery (ICA) stenosis, stated their diagnostic confidence and rated the image quality of cervical arteries, impact of artifacts and image noise. Apparent signal- and contrast-to-noise ratios (aSNR/aCNR) were measured for the common carotid artery and ICA.

Results: REACT achieved a sensitivity of 95.0% and a specificity of 97.3% for ICA stenoses in high agreement with CE-MRA (κ = 0.83) with equal diagnostic confidence (p = 0.22). Image quality was rated higher for CE-MRA at the aortic arch (p = 0.002) and vertebral arteries (p < 0.001), whereas REACT provided superior results for the extracranial ICA (p = 0.008). Both sequences were only slightly affected by artifacts (p = 0.60), while image noise was more pronounced in CE-MRA (p < 0.001) in line with higher aSNR (p < 0.001) and aCNR (p < 0.001) values in REACT for all vessels.

Conclusion: Given its good diagnostic performance while yielding comparable image quality and scan time to CE-MRA, REACT may be suitable for the imaging of the extracranial arteries in acute ischemic stroke at 1.5 T.

Background: Dual-layer stents have fallen into disrepute after several studies reported high rates of in-stent occlusions in acute stroke treatments. The CGuard stent is a new-generation hybrid dual-layer stent that has been designed to provide less thrombogenicity and to prevent peri- and postinterventional emboli. The aim of the study is to evaluate the safety and efficacy of the CGuard stent for the acute treatment of occlusion or high-grade stenosis of the extracranial internal carotid artery (ICA) in patients with acute ischemic stroke (AIS) with and without concomitant intracranial large vessel occlusion (LVO).

Methods: All patients who underwent emergent carotid artery stenting (CAS) with the CGuard stent were identified and analyzed from the stroke registries from four tertiary German stroke centers. Clinical, procedural, and imaging data were evaluated. Stent patency within 72 h, intracranial hemorrhage, and modified Rankin score (mRS) at discharge were the safety and efficacy end points.

Results: Overall, ninety-six patients were included (mean age 70.2 ± 11.8, 66 males (68.8%), median NIHSS score at admission 11 (7-17), IV lysis: n = 44 (45.8%)). Stent placement was successful in all patients. Eighty-three (86.4%) patients had tandem occlusions. In-stent occlusion occurred in 5 patients (5.2%) and 3 patients developed early in-stent stenosis (3.1%). Median mRS at discharge was 2 (1-4).

Conclusion: In this multicenter study, the use of the dual-layer CGuard stent for emergent CAS, particularly in tandem occlusions, was safe and resulted in low rates of in-stent occlusions.

Objective: In 2022, arterioectatic spinal angiopathy (AESA) of childhood was reported as a fatal, progressive, multi-segment myelopathy associated with a unique form of non-inflammatory spinal angiopathy involving diffuse dilatation of the anterior spinal artery and cord congestion in children. In this study, we present four more cases of AESA, using early and long-term conventional imaging and flat detector computed tomography angiography (FDCTA) imaging to assess the probability of disease regression and prevent unnecessary interventions.

Methods: We retrospectively reviewed the clinical and radiological findings of four patients with AESA seen in two neuroradiology departments between 2014 and 2023.

Results: The study included three boys and one girl. Two of the boys were siblings. Although the clinical and radiological presentation in the early stages of the clinical course overlapped the definition of AESA, the clinical course was more benign in three of the cases. The clinical courses of the two siblings with monosegmental cord involvement and largely reversible radiological findings suggest that some of the features in the initial definition of the disease cannot be standardized for all patients. The siblings had a mutation of the NDUFS gene, which is involved in mitochondrial function and clinical-radiological reversibility in these patients.

Conclusion: Many mitochondrial diseases, such as this NDUFS mutation, present with myelopathy, and mitochondrial diseases can sometimes show spontaneous recovery. It is crucial to identify other genetic mutations or environmental factors that trigger the accompanying vascular ectatic findings in AESA in larger multicenter studies to prevent its potential lethal course and possible unnecessary surgical-endovascular interventions.

Purpose: Flat-panel detector computed tomography (FDCT) is increasingly used in (neuro)interventional angiography suites. This study aimed to compare FDCT perfusion (FDCTP) with conventional multidetector computed tomography perfusion (MDCTP) in patients with acute ischemic stroke.

Methods: In this study, 19 patients with large vessel occlusion in the anterior circulation who had undergone mechanical thrombectomy, baseline MDCTP and pre-interventional FDCTP were included. Hypoperfused tissue volumes were manually segmented on time to maximum (Tmax) and time to peak (TTP) maps based on the maximum visible extent. Absolute and relative thresholds were applied to the maximum visible extent on Tmax and relative cerebral blood flow (rCBF) maps to delineate penumbra volumes and volumes with a high likelihood of irreversible infarcted tissue ("core"). Standard comparative metrics were used to evaluate the performance of FDCTP.

Results: Strong correlations and robust agreement were found between manually segmented volumes on MDCTP and FDCTP Tmax maps (r = 0.85, 95% CI 0.65-0.94, p < 0.001; ICC = 0.85, 95% CI 0.69-0.94) and TTP maps (r = 0.91, 95% CI 0.78-0.97, p < 0.001; ICC = 0.90, 95% CI 0.78-0.96); however, direct quantitative comparisons using thresholding showed lower correlations and weaker agreement (MDCTP versus FDCTP Tmax 6 s: r = 0.35, 95% CI -0.13-0.69, p = 0.15; ICC = 0.32, 95% CI 0.07-0.75). Normalization techniques improved results for Tmax maps (r = 0.78, 95% CI 0.50-0.91, p < 0.001; ICC = 0.77, 95% CI 0.55-0.91). Bland-Altman analyses indicated a slight systematic underestimation of FDCTP Tmax maximum visible extent volumes and slight overestimation of FDCTP TTP maximum visible extent volumes compared to MDCTP.

Conclusion: FDCTP and MDCTP provide qualitatively comparable volumetric results on Tmax and TTP maps; however, direct quantitative measurements of infarct core and hypoperfused tissue volumes showed lower correlations and agreement.