AJNR. American journal of neuroradiology最新文献

Adult-onset genetic leukoencephalopathies (AGL) are frequently misdiagnosed due to overlap with more common acquired white matter (WM) diseases. The WM Rounds Network includes clinicians/scientists from more than 15 centers around the world who meet monthly to discuss undiagnosed WM diseases. MRI remains a central tool in narrowing the differential diagnosis of WM diseases; however, a key barrier to optimizing diagnosis was the lack of protocol standardization between institutions. We aimed to 1. assess the state of practice of current protocols for investigating suspected AGLs and 2. propose a core standard protocol.We used a modified Delphi method to facilitate group judgements. We developed a survey and submitted it for feedback to a panel of neuroimaging experts. The final survey was circulated to the entire WM Rounds Network. The results were analyzed, and specific recommendations were put forward during Delphi rounds, for which the stop criterion was >75% agreement.The average summed sequence time of our MR protocols was 40 minutes (range 25-60) and included the following sequences: 3D T1 MPRAGE and 3D FLAIR obtained in the sagittal plane, axial T2 and T2-FLAIR, axial DWI/ADC, and axial SWI. Cervical and thoracic spine imaging (T1 sagittal, T2 sagittal and axial) were also frequently performed. A standardized core imaging protocol inclusive of the above-listed sequences would help harmonize sequence acquisition across institutions and promote cost-effective optimization of the imaging workup of AGLs. Four additional recommendations were proposed: 1) Contrast-enhanced T1 sequences should be performed for patients with demonstrable clinical or radiological evidence of AGL. 2) Lumbar MR spine imaging has limited utility. 3) Head CT is of little added value and should not be included routinely. 4) MRS, DTI, and other advanced imaging techniques are not yet ready to be implemented in the protocol but may be helpful in supporting a working diagnosis.ABBREVIATIONS: AGL＝ adult-onset genetic leukoencephalopathies; WM＝ white matter.

Background and purpose: The International League Against Epilepsy (ILAE) classifies hippocampal sclerosis (HS) based on the degree of neuronal loss affecting hippocampal subfields. Due to the differences in clinical presentation and postoperative outcomes, preoperative prediction of HS subtypes by MRI can aid in surgical decision-making. The current study aims to evaluate the utility of preoperative high-resolution MRI in this prediction.

Materials and methods: With institutional ethical committee approval, a retrospective observational study was conducted in a single tertiary referral centre. Consecutive histopathology-proven cases of unilateral mesial temporal sclerosis from January 2018 to June 2023 were included, if they also had a pre-operative MR study on a 3 Tesla, performed with a high-resolution epilepsy protocol. Two neuroradiologists, blinded to the histopathology results, independently performed MR grading by assessing volume loss in hippocampal subfields on a high-resolution 2D coronal T2-weighted sequence. The discrepancies were resolved by a senior neuroradiologist. An agreement was sought between the MR grading and the histopathology grading of HS.

Results: A total of 106 patients (57 males) were included with a mean age of 23.8 years. Type 1 HS was the most common subtype both on MR (n: 71) and histopathology (n: 89). MRI detected 'No hippocampal sclerosis/gliosis only' (GO) subtype of HS with 72.7% (95% CI, 43.4%-90.3%) sensitivity and 97.9% (95% CI, 92.7%-99.4%) specificity, and Type 1 HS with 75.3% (95% CI, 65.4%-83.1%) sensitivity and 76.5% (95% CI, 52.7%-90.4%) specificity. Despite a good interobserver agreement (Cohen's kappa: 0.770, P <0.01), MR showed only a fair agreement with histopathology (Cohen's kappa: 0.328, P < 0.01) in subclassifying HS. For individual HS subtypes, MR showed good agreement in predicting GO subtype (Cohen's kappa: 0.736, P <0.01) and fair agreement (Cohen's kappa: 0.362, P <0.01) in predicting HS type 1.

Conclusions: Hippocampal subfield analysis is feasible with high-resolution 3T MRI, with good interobserver agreement. MR grading based on a qualitative assessment of volume loss in hippocampal subfields can predict the GO subtype with good agreement compared to histopathology. However, differentiation of HS types 1, 2, and 3 subtypes from each other may not be feasible pre-operatively with the current clinical 3T MR protocols.

Abbreviations: CA= Cornu Ammonis, HS= Hippocampal Sclerosis, GO= No hippocampal sclerosis/gliosis only, ILAE= International League Against Epilepsy, SRLM= Strata radiatum, lacunosum, and moleculare, NeuN= Neuronal nuclear antigen, GFAP= Glial fibrillary acidic protein.

Background and purpose: The relationship between regional brain edema caused by anti-amyloid monoclonal antibodies (ARIA-E) and the degree of regional β-amyloid (Aβ) positron emission tomography (PET) signal reduction is unknown.

Materials and methods: In patients with moderate or severe ARIA-E, we quantified changes in Aβ PET signal before and after ARIA-E resolution, comparing regions affected by ARIA-E with unaffected regions.

Results: In four of five patients treated with lecanemab or donanemab and who had moderate or severe ARIA-E, Aβ PET signal decreased significantly more in regions that had been involved with ARIA-E.

Conclusions: Greater regional Aβ PET signal reduction in areas affected by ARIA-E may reflect enhanced local Aβ clearance, reduced tracer binding site availability, impaired glymphatic flow from immune complex deposition, or other mechanisms. The finding of greater regional Aβ PET signal reduction in ARIA-E regions refines the characterization of ARIA-E and raises the possibility that its occurrence may have beneficial as well as adverse implications.

Abbreviations: ARIA＝ amyloid related imaging abnormalities; Aβ＝ β-amyloid; SUVR = standard uptake value ratio.

Focused ultrasound (FUS) is an emerging therapeutic and diagnostic technology in neuro-oncology, offering new strategies for molecular diagnosis, drug delivery, and tumor ablation across a range of brain tumors, including glioblastoma (GBM), brain metastases, and diffuse intrinsic pontine glioma (DIPG). The prognosis for aggressive brain tumors remains poor, despite advances in surgery, radiation, and chemotherapy. A considerable challenge is the limited ability to deliver therapeutics across the blood-brain barrier (BBB), particularly to infiltrative or non-enhancing tumor regions. FUS introduces an incisionless approach to the molecular subtyping of brain tumors, enhancing therapeutic delivery, and offers novel therapeutic approaches such as sonodynamic therapy (SDT). This review summarizes the FUS mechanisms and highlights the critical role of imaging modalities confirming target engagement, assessing bioeffects and outcomes, and ensuring safety. We also explore future directions, including the integration of liquid biopsy, artificial intelligence, and outpatient-ready FUS platforms, which will position FUS as a promising adjunct to standard neuro-oncologic care.ABBREVIATIONS: GBM = glioblastoma; FUS = focused ultrasound; HIFU = high-intensity focused ultrasound; LIFU = low-intensity focused ultrasound; MRgFUS = magnetic resonance guided focused ultrasound; BBBO = blood brain barrier opening; SDT = sonodynamic therapy; 5-ALA = 5-aminolevulinic acid.

Background and purpose: Identifying amyloid-beta (Aβ)-positive patients is essential for Alzheimer disease clinical trials and disease-modifying treatments but currently requires PET or CSF sampling. Previous MRI-based deep learning models using only T1-weighted (T1w) images have shown moderate performance.

Materials and methods: Multicontrast MRI- and PET-based quantitative Aβ deposition were retrospectively obtained from 3 public data sets: ADNI, OASIS3, and A4. Aβ positivity was defined using the recommended Centiloid threshold of each data set. Two EfficientNet models were trained to predict amyloid-positivity: one by using only T1w images and another incorporating both T1w and T2 FLAIR. Model performance was assessed using an internal held-out test set, evaluating area under the curve (AUC), accuracy, sensitivity, and specificity. External validation was conducted using an independent cohort from Stanford Alzheimer Disease Research Center. DeLong and McNemar tests were used to compare AUC and accuracy, respectively.

Results: A total of 4056 examinations (mean age: 71.6 [SD, 6.3] years; 55% female; 55% amyloid-positive) were used for network development, and 149 examinations were used for external testing (mean age: 72.1 [SD] 9.6] years; 57% female; 56% amyloid-positive). The multicontrast model outperformed the single-technique model in the internal held-out test set (AUC: 0.67; 95% CI, 0.65-0.70; P < .001; accuracy: 0.63; 95% CI, 0.62-0.65; P < .001) compared with the T1w-only model (AUC: 0.61; accuracy: 0.59). Among cognitive subgroups, the highest performance (AUC: 0.71) was observed in mild cognitive impairment. The multicontrast model also demonstrated consistent performance in the external test set (AUC: 0.65; 95% CI, 0.60-0.71; P = .014; accuracy: 0.62; 95% CI, 0.58-0.65; P < .001).

Conclusions: The use of multicontrast MRI, specifically incorporating T2 FLAIR in addition to T1w images, significantly improved the predictive accuracy of PET-determined amyloid status from MRIs by using a deep learning approach.

Background and purpose: Visual disturbance is a major complication in nonfunctioning pituitary adenoma (NFPA) due to chiasmal compression. While neuroimaging studies have established brain dysfunction in visually impaired patients from chiasmal compression, the brain dynamic features of spontaneous activity and functional connectivity remain underexplored. This cross-sectional study aims to explore changes in temporal variability of spontaneous activity and connectivity in visually impaired patients with NFPA by resting-state functional MRI.

Materials and methods: Thirty-six patients with NFPA with visual impairment and 36 healthy controls were recruited and underwent resting-state fMRI scans. Dynamic amplitude of low-frequency fluctuation (dALFF) and dynamic functional connectivity (dFC) analyses were performed to assess temporal variability in brain activity and interregional communication. Associations between altered dALFF/dFC and the severity and duration of chiasmal compression, as well as visual field defect severity (quantified by mean deviation), were further evaluated.

Results: Compared with healthy controls, patients with a nonfunctioning pituitary exhibited significantly reduced dALFF in the right lingual gyrus (LING) and bilateral calcarine fissure and surrounding cortex (CAL). Additionally, patients showed a significant reduction in dFC between the right LING and the bilateral precuneus. Our exploratory correlation analyses revealed that the altered dALFF values in the bilateral CAL and right LING were positively correlated with chiasmal volume, and the altered dALFF in the left CAL was negatively correlated with suprasellar extension distance. Additionally, the altered dALFF values in the bilateral CAL were positively correlated with mean deviation.

Conclusions: Patients with NFPA with visual impairment exhibited decreased temporal variability in brain activity and functional connectivity within visual-related regions, offering new insights into the neuropathologic mechanisms underlying visual disturbance in nonfunctioning pituitary adenoma.

Background and purpose: Multiphase CT Angiography (mCTA) has shown potential as a diagnostic tool for acute ischemic stroke because it captures dynamic changes in the cerebral vasculature. However, mCTA has limitations in assessing brain tissue perfusion, which reduces its clinical interpretability. To address this limitation, we aimed to develop a generative adversarial network (GAN) that generates CTP-like maps from mCTA. This approach aims to improve the interpretability of mCTA.

Materials and methods: A total of 714 cases with NCCT, CTP, mCTA, and follow-up NCCT/MRI were analyzed across internal and external data sets. A GAN was trained to generate multiparametric CTP maps (Tmax, CBF, CBV). The performance of the model was evaluated using the Structural Similarity Index (SSIM), peak signal-to-noise ratio (PSNR), and Fréchet Inception Distance (FID) compared with actual CTP maps. Clinical utility was assessed by predicting infarct core and penumbra using threshold-based segmentation and evaluating metrics such as the Dice coefficient, area under the receiver operating characteristic curve (AUC) of dichotomized infarct volumes of < 70 mL, and mismatch ratio following DEFUSE 3 criteria, compared with the ground truth of actual CTP prediction.

Results: The GAN achieved SSIM, 0.65-0.66; PSNR, 20.4-20.8; and FID, 15.8-17.0 on internal data, surpassing both CycleGAN (SSIM: 0.608-0.642, PSNR: 18.2-19.7, FID: 27.6-32.5) and Pix2Pix (SSIM: 0.630-0.645, PSNR: 19.5-19.7, FID: 19.4-20.8) across all metrics. Predicted penumbra and infarct core showed Dice coefficients of 0.672 and 0.468, with strong correlations (penumbra: 0.921, core: 0.902) and AUCs of 0.854 (95% CI, 0.819-0.888) (mismatch ratio) and 0.850 (95% CI, 0.817-0.884) (dichotomized infarct core). External data validation yielded Dice coefficients of 0.481 (penumbra) and 0.301 (core) with AUCs of 0.720 (95% CI, 0.589-0.808) (mismatch ratio) and 0.703 (95% CI, 0.528-0.794) (dichotomized infarct core).

Conclusions: The GAN effectively generated CTP-like maps from mCTA, improving interpretability and demonstrating promising diagnostic performance, particularly for resource-limited settings.

Background and purpose: Differentiating between a brain tumor and a nontumorous lesion remains a diagnostic challenge, particularly when conventional imaging modalities such as CT and MRI provide inconclusive results. While ¹¹C-methionine PET (MET-PET) has shown potential in neuro-oncology, its diagnostic performance across a broad spectrum of brain pathologies has not been comprehensively evaluated. This study, therefore, assessed the sensitivity, specificity, and uptake patterns of MET-PET in a large cohort of brain lesions.

Materials and methods: This single-center retrospective study analyzed 426 consecutive patients with undiagnosed brain lesions who underwent MET-PET imaging between January 2019 and May 2024. Tumor-to-normal region ratios (TNRs) were calculated by using a threshold of 1.5 for positive findings. Histologic diagnoses were established on the basis of the World Health Organization 2021 criteria, including isocitrate dehydrogenase (IDH) mutation status and 1p/19q-codeletion.

Results: Among the cohort, 342 cases (67.8%) were confirmed as having tumorous lesions; 76 (17.8%), as having nontumorous lesions; and 61 (14.3%) remained undiagnosed. MET-PET exhibited high sensitivity (86.2%) but limited specificity (47.4%) for tumor detection. In multiple sclerosis cases, MET-PET showed a remarkably high positivity rate (n = 10/12) that was significantly higher than for other nontumorous lesions. In terms of tumors, IDH wild-type glioblastomas had significantly higher TNRs compared with IDH-mutant gliomas, while oligodendrogliomas had higher TNRs compared with astrocytomas, in which TNR values correlated with tumor grade.

Conclusions: MET-PET demonstrated robust sensitivity for brain tumor detection but was limited by low specificity due to false-positives in inflammatory conditions and false-negatives for low-grade tumors. These findings imply the importance of integrating MET-PET with other imaging modalities to enhance diagnostic accuracy.

Background and purpose: Delayed neurologic sequelae are among the most serious complications of carbon monoxide poisoning. However, no reliable tools are available for evaluating their potential risk. We aimed to assess whether machine learning models using imaging features that were automatically extracted from brain MRI can predict the potential delayed neurologic sequelae risk in patients with acute carbon monoxide poisoning.

Materials and methods: This single-center, retrospective, observational study analyzed a prospectively collected registry of patients with acute carbon monoxide poisoning who visited our emergency department from April 2011 to December 2015. Overall, 1618 radiomics and 4 lesion-segmentation features from DWI b1000 and ADC images, as well as 62 clinical variables, were extracted from each patient. The entire data set was divided into 5 subsets, with 1 serving as the hold-out test set and the remaining 4 used for training and tuning. Four machine learning models, linear regression, support vector machine, random forest, and extreme gradient boosting, as well as an ensemble model, were trained and evaluated by using 20 different data configurations. The primary evaluation metric was the mean and 95% CI of the area under the receiver operating characteristic curve. Shapley additive explanations were calculated and visualized to enhance model interpretability.

Results: Of the 373 patients, delayed neurologic sequelae occurred in 99 (26.5%) patients (mean age 43.0 ± 15.2; 62.0% men). The means [95% CIs] of the area under the receiver operating characteristic curve (AUROC), accuracy, sensitivity, and specificity of the best performing machine learning model for predicting the development of delayed neurologic sequelae were 0.88 [0.86-0.9], 0.82 [0.8-0.83], 0.81 [0.79-0.83], and 0.82 [0.8-0.84], respectively. Among imaging features, the presence, size, and number of acute brain lesions on DWI b1000 and ADC images more accurately predicted delayed neurologic sequelae risk than advanced radiomics features based on shape, texture, and wavelet transformation.

Conclusions: Machine learning models developed using automatically extracted brain MRI features with clinical features can distinguish patients at delayed neurologic sequelae risk. The models enable effective prediction of delayed neurologic sequelae in patients with acute carbon monoxide poisoning, facilitating timely treatment planning for prevention.

Background and purpose: Distinguishing GM from WM is essential for CT of the brain. The recently established photon-counting detector (PCD)-CT technology uses a novel detection technique that might allow more precise measurement of tissue attenuation for an improved delineation of attenuation values (Hounsfield units [HU]) and improved image quality in comparison with energy-integrating detector (EID)-CT. To investigate this, we compared HU, GM versus WM contrast, and image noise by using automated deep learning-based brain segmentations.

Materials and methods: We retrospectively included patients who received either PCD-CT or EID-CT and did not display a cerebral pathology. A deep learning-based segmentation of the GM and WM was used to extract HU. From this, the gray-white matter ratio (GWR) and contrast-to-noise ratio (CNR) were calculated.

Results: We included 329 patients with EID-CT (mean age 59.8 ± 20.2 years) and 180 with PCD-CT (mean age 64.7 ± 16.5 years). GM and WM showed significantly lower HU in PCD-CT (GM: 40.4 ± 2.2 HU; WM: 33.4 ± 1.5 HU) compared with EID-CT (GM: 45.1 ± 1.6 HU; WM: 37.4 ± 1.6 HU; P < .001). Standard deviations of HU were also lower in PCD-CT (GM and WM both P < .001) and CNR was significantly higher in PCD-CT compared with EID-CT (P < .001). GWRs were not significantly different across both modalities (P > .99). In an age-matched subset (n=157 patients from both cohorts), all findings were replicated.

Conclusions: This comprehensive comparison of HU in cerebral GM and WM revealed substantially reduced image noise and an average offset with lower HU in PCD-CT while the ratio between GM and WM remained constant. The potential need to adapt windowing presets based on this finding should be investigated in future studies.