ABSTRACT
Purpose: To help radiologists examine the growing number of computed tomography (CT) scans, automatic anomaly detection is an ongoing focus of medical imaging research. Radiologists must analyze a CT scan by searching for any deviation from normal healthy anatomy. We propose an approach to detecting abnormalities in axial 2D CT slice images of the brain. Although much research has been done on detecting abnormalities in magnetic resonance images of the brain, there is little work on CT scans, where abnormalities are more difficult to detect due to the low image contrast that must be represented by the model used. Approach: We use a generative adversarial network (GAN) to learn normal brain anatomy in the first step and compare two approaches to image reconstruction: training an encoder in the second step and using iterative optimization during inference. Then, we analyze the differences from the original scan to detect and localize anomalies in the brain. Results: Our approach can reconstruct healthy anatomy with good image contrast for brain CT scans. We obtain median Dice scores of 0.71 on our hemorrhage test data and 0.43 on our test set with additional tumor images from publicly available data sources. We also compare our models to a state-of-the-art autoencoder and a diffusion model and obtain qualitatively more accurate reconstructions. Conclusions: Without defining anomalies during training, a GAN-based network was used to learn healthy anatomy for brain CT scans. Notably, our approach is not limited to the localization of hemorrhages and tumors and could thus be used to detect structural anatomical changes and other lesions.
ABSTRACT
Thalamic aphasia results from focal thalamic lesions that cause dysfunction of remote but functionally connected cortical areas due to language network perturbation. However, specific local and network-level neural substrates of thalamic aphasia remain incompletely understood. Using lesion symptom mapping, we demonstrate that lesions in the left ventrolateral and ventral anterior thalamic nucleus are most strongly associated with aphasia in general and with impaired semantic and phonemic fluency and complex comprehension in particular. Lesion network mapping (using a normative connectome based on fMRI data from 1000 healthy individuals) reveals a Thalamic aphasia network encompassing widespread left-hemispheric cerebral connections, with Broca's area showing the strongest associations, followed by the superior and middle frontal gyri, precentral and paracingulate gyri, and globus pallidus. Our results imply the critical involvement of the left ventrolateral and left ventral anterior thalamic nuclei in engaging left frontal cortical areas, especially Broca's area, during language processing.
Subject(s)
Aphasia , Magnetic Resonance Imaging , Stroke , Thalamus , Ventral Thalamic Nuclei , Humans , Male , Middle Aged , Female , Ventral Thalamic Nuclei/physiopathology , Ventral Thalamic Nuclei/diagnostic imaging , Aphasia/physiopathology , Aphasia/etiology , Aphasia/diagnostic imaging , Stroke/complications , Stroke/physiopathology , Thalamus/physiopathology , Thalamus/diagnostic imaging , Aged , Adult , Connectome , Frontal Lobe/physiopathology , Frontal Lobe/diagnostic imaging , Nerve Net/physiopathology , Nerve Net/diagnostic imaging , Neural Pathways/physiopathologyABSTRACT
INTRODUCTION: When time since stroke onset is unknown, DWI-FLAIR mismatch rating is an established technique for patient stratification. A visible DWI lesion without corresponding parenchymal hyperintensity on FLAIR suggests time since onset of under 4.5 h and thus a potential benefit from intravenous thrombolysis. To improve accuracy and availability of the mismatch concept, deep learning might be able to augment human rating and support decision-making in these cases. METHODS: We used unprocessed DWI and coregistered FLAIR imaging data to train a deep learning model to predict dichotomized time since ischemic stroke onset. We analyzed the performance of Group Convolutional Neural Networks compared to other deep learning methods. Unlabeled imaging data was used for pre-training. Prediction performance of the best deep learning model was compared to the performance of four independent junior and senior raters. Additionally, in cases deemed indeterminable by human raters, model ratings were used to augment human performance. Post-hoc gradient-based explanations were analyzed to gain insights into model predictions. RESULTS: Our best predictive model performed comparably to human raters. Using model ratings in cases deemed indeterminable by human raters improved rating accuracy and interrater agreement for junior and senior ratings. Post-hoc explainability analyses showed that the model localized stroke lesions to derive predictions. DISCUSSION: Our analysis shows that deep learning based clinical decision support has the potential to improve the accessibility of the DWI-FLAIR mismatch concept by supporting patient stratification.
Subject(s)
Brain Ischemia , Deep Learning , Ischemic Stroke , Stroke , Humans , Diffusion Magnetic Resonance Imaging/methods , Time Factors , Stroke/diagnostic imaging , Stroke/pathologyABSTRACT
Introduction: The automatic segmentation of brain parenchyma and cerebrospinal fluid-filled spaces such as the ventricular system is the first step for quantitative and qualitative analysis of brain CT data. For clinical practice and especially for diagnostics, it is crucial that such a method is robust to anatomical variability and pathological changes such as (hemorrhagic or neoplastic) lesions and chronic defects. This study investigates the increase in overall robustness of a deep learning algorithm that is gained by adding hemorrhage training data to an otherwise normal training cohort. Methods: A 2D U-Net is trained on subjects with normal appearing brain anatomy. In a second experiment the training data includes additional subjects with brain hemorrhage on image data of the RSNA Brain CT Hemorrhage Challenge with custom reference segmentations. The resulting networks are evaluated on normal and hemorrhage test casesseparately, and on an independent test set of patients with brain tumors of the publicly available GLIS-RT dataset. Results: Adding data with hemorrhage to the training set significantly improves the segmentation performance over an algorithm trained exclusively on normally appearing data, not only in the hemorrhage test set but also in the tumor test set. The performance on normally appearing data is stable. Overall, the improved algorithm achieves median Dice scores of 0.98 (parenchyma), 0.91 (left ventricle), 0.90 (right ventricle), 0.81 (third ventricle), and 0.80 (fourth ventricle) on the hemorrhage test set. On the tumor test set, the median Dice scores are 0.96 (parenchyma), 0.90 (left ventricle), 0.90 (right ventricle), 0.75 (third ventricle), and 0.73 (fourth ventricle). Conclusion: Training on an extended data set that includes pathologies is crucial and significantly increases the overall robustness of a segmentation algorithm for brain parenchyma and ventricular system in CT data, also for anomalies completely unseen during training. Extension of the training set to include other diseases may further improve the generalizability of the algorithm.
ABSTRACT
Purpose: Automated lesion segmentation is increasingly used in acute ischemic stroke magnetic resonance imaging (MRI). We explored in detail the performance of apparent diffusion coefficient (ADC) thresholding for delineating baseline diffusion-weighted imaging (DWI) lesions. Methods: Retrospective, exploratory analysis of the prospective observational single-center 1000Plus study from September 2008 to June 2013 (clinicaltrials.org; NCT00715533). We built a fully automated lesion segmentation algorithm using a fixed ADC threshold (≤620 × 10-6 mm2/s) to delineate the baseline DWI lesion and analyzed its performance compared to manual assessments. Diagnostic capabilities of best possible ADC thresholds were investigated using receiver operating characteristic curves. Influential patient factors on ADC thresholding techniques' performance were studied by conducting multiple linear regression. Results: 108 acute ischemic stroke patients were selected for analysis. The median Dice coefficient for the algorithm was 0.43 (IQR 0.20-0.64). Mean ADC values in the DWI lesion (ß = -0.68, p < 0.001) and DWI lesion volumes (ß = 0.29, p < 0.001) predicted performance. Optimal individual ADC thresholds differed between subjects with a median of ≤691 × 10-6 mm2/s (IQR ≤660-750 × 10-6 mm2/s). Mean ADC values in the DWI lesion (ß = -0.96, p < 0.001) and mean ADC values in the brain parenchyma (ß = 0.24, p < 0.001) were associated with the performance of individual thresholds. Conclusion: The performance of ADC thresholds for delineating acute stroke lesions varies substantially between patients. It is influenced by factors such as lesion size as well as lesion and parenchymal ADC values. Considering the inherent noisiness of ADC maps, ADC threshold-based automated delineation of very small lesions is not reliable.
ABSTRACT
BACKGROUND: White matter hyperintensities of presumed vascular origin (WMH) are the most prominent imaging feature of cerebral small vessel disease (cSVD). Previous studies suggest a link between cSVD burden and intracerebral hemorrhage and worse functional outcome after thrombolysis in acute ischemic stroke. We aimed to determine the impact of WMH burden on efficacy and safety of thrombolysis in the MRI-based randomized controlled WAKE-UP trial of intravenous alteplase in unknown onset stroke. METHODS: The design of this post hoc study was an observational cohort design of a secondary analysis of a randomized trial. WMH volume was quantified on baseline fluid-attenuated inversion recovery images of patients randomized to either alteplase or placebo in the WAKE-UP trial. Excellent outcome was defined as score of 0-1 on the modified Rankin Scale after 90 days. Hemorrhagic transformation was assessed on follow-up imaging 24-36 hours after randomization. Treatment effect and safety were analyzed by fitting multivariable logistic regression models. RESULTS: Quality of scans was sufficient in 441 of 503 randomized patients to delineate WMH. Median age was 68 years, 151 patients were female, and 222 patients were assigned to receive alteplase. Median WMH volume was 11.4 mL. Independent from treatment, WMH burden was statistically significantly associated with worse functional outcome (odds ratio, 0.72 [95% CI, 0.57-0.92]), but not with higher chances of any hemorrhagic transformation (odds ratio, 0.78 [95% CI, 0.60-1.01]). There was no interaction of WMH burden and treatment group for the likelihood of excellent outcome (P=0.443) or any hemorrhagic transformation (P=0.151). In a subgroup of 166 patients with severe WMH, intravenous thrombolysis was associated with higher odds of excellent outcome (odds ratio, 2.40 [95% CI, 1.19-4.84]) with no significant increase in the rate of hemorrhagic transformation (odds ratio, 1.96 [95% CI, 0.80-4.81]). CONCLUSIONS: Although WMH burden is associated with worse functional outcome, there is no association with treatment effect or safety of intravenous thrombolysis in patients with ischemic stroke of unknown onset. REGISTRATION: URL: https://www. CLINICALTRIALS: gov; Unique identifier: NCT01525290.
Subject(s)
Brain Ischemia , Ischemic Stroke , Stroke , White Matter , Humans , Female , Aged , Male , Tissue Plasminogen Activator , Fibrinolytic Agents , Brain Ischemia/diagnostic imaging , Brain Ischemia/drug therapy , Brain Ischemia/etiology , Thrombolytic Therapy/methods , Ischemic Stroke/drug therapy , White Matter/diagnostic imaging , Stroke/diagnostic imaging , Stroke/drug therapy , Stroke/etiology , Treatment OutcomeABSTRACT
Background Cerebral microbleeds (CMBs) are increasingly recognized as "covert" brain lesions indicating increased risk of future neurological events. However, data on CMBs in patients undergoing catheter-based structural heart interventions are scarce. Therefore, we assessed occurrence and predictors of new CMBs in patients undergoing catheter-based left atrial appendage closure and percutaneous mitral valve repair using the MitraClip System. Methods and Results We conducted an exploratory analysis using data derived from 2 prospective, observational studies. Eligible patients underwent cerebral magnetic resonance imaging (3 Tesla) examinations and cognitive tests (using the Montreal Cognitive Assessment) before and after catheter-based left atrial appendage closure and percutaneous mitral valve repair. Forty-seven patients (53% men; median age, 77 years) were included. New CMBs occurred in 17 of 47 patients (36%) following catheter-based structural heart interventions. Occurrences of new CMBs did not differ significantly between patients undergoing catheter-based left atrial appendage closure and percutaneous mitral valve repair (7/25 versus 10/22; P=0.348). In univariable analysis, longer procedure time was significantly associated with new CMBs. Adjustment for heparin attenuated this association (adjusted odds ratio [per 30 minutes]: 1.77 [95% CI, 0.92-3.83]; P=0.090). Conclusions New CMBs occur in approximately one-third of patients after catheter-based left atrial appendage closure and percutaneous mitral valve repair using the MitraClip System. Our data suggest that longer duration of the procedure may be a risk factor for new CMBs. Future studies in larger populations are needed to further investigate their clinical relevance. Clinical Trial Registration German Clinical Trials Register: DRKS00010300 (https://drks.de/search/en/trial/DRKS00010300); ClinicalTrials.gov : NCT03104556 (https://clinicaltrials.gov/ct2/show/NCT03104556?term=NCT03104556&draw=2&rank=1).
Subject(s)
Cardiac Surgical Procedures , Mitral Valve Insufficiency , Aged , Female , Humans , Male , Cardiac Surgical Procedures/adverse effects , Cardiac Surgical Procedures/methods , Catheters/adverse effects , Cerebral Hemorrhage/diagnostic imaging , Cerebral Hemorrhage/etiology , Magnetic Resonance Imaging , Mitral Valve Insufficiency/diagnostic imaging , Mitral Valve Insufficiency/surgery , Mitral Valve Insufficiency/complications , Prospective Studies , Treatment OutcomeABSTRACT
Brain arteries are routinely imaged in the clinical setting by various modalities, e.g., time-of-flight magnetic resonance angiography (TOF-MRA). These imaging techniques have great potential for the diagnosis of cerebrovascular disease, disease progression, and response to treatment. Currently, however, only qualitative assessment is implemented in clinical applications, relying on visual inspection. While manual or semi-automated approaches for quantification exist, such solutions are impractical in the clinical setting as they are time-consuming, involve too many processing steps, and/or neglect image intensity information. In this study, we present a deep learning-based solution for the anatomical labeling of intracranial arteries that utilizes complete information from 3D TOF-MRA images. We adapted and trained a state-of-the-art multi-scale Unet architecture using imaging data of 242 patients with cerebrovascular disease to distinguish 24 arterial segments. The proposed model utilizes vessel-specific information as well as raw image intensity information, and can thus take tissue characteristics into account. Our method yielded a performance of 0.89 macro F1 and 0.90 balanced class accuracy (bAcc) in labeling aggregated segments and 0.80 macro F1 and 0.83 bAcc in labeling detailed arterial segments on average. In particular, a higher F1 score than 0.75 for most arteries of clinical interest for cerebrovascular disease was achieved, with higher than 0.90 F1 scores in the larger, main arteries. Due to minimal pre-processing, simple usability, and fast predictions, our method could be highly applicable in the clinical setting.
ABSTRACT
The symptoms of acute ischemic stroke can be attributed to disruption of the brain network architecture. Systemic thrombolysis is an effective treatment that preserves structural connectivity in the first days after the event. Its effect on the evolution of global network organisation is, however, not well understood. We present a secondary analysis of 269 patients from the randomized WAKE-UP trial, comparing 127 imaging-selected patients treated with alteplase with 142 controls who received placebo. We used indirect network mapping to quantify the impact of ischemic lesions on structural brain network organisation in terms of both global parameters of segregation and integration, and local disruption of individual connections. Network damage was estimated before randomization and again 22 to 36 h after administration of either alteplase or placebo. Evolution of structural network organisation was characterised by a loss in integration and gain in segregation, and this trajectory was attenuated by the administration of alteplase. Preserved brain network organization was associated with excellent functional outcome. Furthermore, the protective effect of alteplase was spatio-topologically nonuniform, concentrating on a subnetwork of high centrality supported in the salvageable white matter surrounding the ischemic cores. This interplay between the location of the lesion, the pathophysiology of the ischemic penumbra, and the spatial embedding of the brain network explains the observed potential of thrombolysis to attenuate topological network damage early after stroke. Our findings might, in the future, lead to new brain network-informed imaging biomarkers and improved prognostication in ischemic stroke.
Subject(s)
Brain Ischemia , Ischemic Stroke , Stroke , Humans , Tissue Plasminogen Activator/therapeutic use , Tissue Plasminogen Activator/adverse effects , Thrombolytic Therapy/adverse effects , Thrombolytic Therapy/methods , Fibrinolytic Agents/therapeutic use , Fibrinolytic Agents/adverse effects , Brain Ischemia/diagnostic imaging , Brain Ischemia/drug therapy , Stroke/diagnostic imaging , Stroke/drug therapy , Brain/diagnostic imaging , Treatment OutcomeABSTRACT
Sharing labeled data is crucial to acquire large datasets for various Deep Learning applications. In medical imaging, this is often not feasible due to privacy regulations. Whereas anonymization would be a solution, standard techniques have been shown to be partially reversible. Here, synthetic data using a Generative Adversarial Network (GAN) with differential privacy guarantees could be a solution to ensure the patient's privacy while maintaining the predictive properties of the data. In this study, we implemented a Wasserstein GAN (WGAN) with and without differential privacy guarantees to generate privacy-preserving labeled Time-of-Flight Magnetic Resonance Angiography (TOF-MRA) image patches for brain vessel segmentation. The synthesized image-label pairs were used to train a U-net which was evaluated in terms of the segmentation performance on real patient images from two different datasets. Additionally, the Fréchet Inception Distance (FID) was calculated between the generated images and the real images to assess their similarity. During the evaluation using the U-Net and the FID, we explored the effect of different levels of privacy which was represented by the parameter ϵ. With stricter privacy guarantees, the segmentation performance and the similarity to the real patient images in terms of FID decreased. Our best segmentation model, trained on synthetic and private data, achieved a Dice Similarity Coefficient (DSC) of 0.75 for ϵ = 7.4 compared to 0.84 for ϵ = ∞ in a brain vessel segmentation paradigm (DSC of 0.69 and 0.88 on the second test set, respectively). We identified a threshold of ϵ <5 for which the performance (DSC <0.61) became unstable and not usable. Our synthesized labeled TOF-MRA images with strict privacy guarantees retained predictive properties necessary for segmenting the brain vessels. Although further research is warranted regarding generalizability to other imaging modalities and performance improvement, our results mark an encouraging first step for privacy-preserving data sharing in medical imaging.
ABSTRACT
Deep learning requires large labeled datasets that are difficult to gather in medical imaging due to data privacy issues and time-consuming manual labeling. Generative Adversarial Networks (GANs) can alleviate these challenges enabling synthesis of shareable data. While 2D GANs have been used to generate 2D images with their corresponding labels, they cannot capture the volumetric information of 3D medical imaging. 3D GANs are more suitable for this and have been used to generate 3D volumes but not their corresponding labels. One reason might be that synthesizing 3D volumes is challenging owing to computational limitations. In this work, we present 3D GANs for the generation of 3D medical image volumes with corresponding labels applying mixed precision to alleviate computational constraints. We generated 3D Time-of-Flight Magnetic Resonance Angiography (TOF-MRA) patches with their corresponding brain blood vessel segmentation labels. We used four variants of 3D Wasserstein GAN (WGAN) with: 1) gradient penalty (GP), 2) GP with spectral normalization (SN), 3) SN with mixed precision (SN-MP), and 4) SN-MP with double filters per layer (c-SN-MP). The generated patches were quantitatively evaluated using the Fréchet Inception Distance (FID) and Precision and Recall of Distributions (PRD). Further, 3D U-Nets were trained with patch-label pairs from different WGAN models and their performance was compared to the performance of a benchmark U-Net trained on real data. The segmentation performance of all U-Net models was assessed using Dice Similarity Coefficient (DSC) and balanced Average Hausdorff Distance (bAVD) for a) all vessels, and b) intracranial vessels only. Our results show that patches generated with WGAN models using mixed precision (SN-MP and c-SN-MP) yielded the lowest FID scores and the best PRD curves. Among the 3D U-Nets trained with synthetic patch-label pairs, c-SN-MP pairs achieved the highest DSC (0.841) and lowest bAVD (0.508) compared to the benchmark U-Net trained on real data (DSC 0.901; bAVD 0.294) for intracranial vessels. In conclusion, our solution generates realistic 3D TOF-MRA patches and labels for brain vessel segmentation. We demonstrate the benefit of using mixed precision for computational efficiency resulting in the best-performing GAN-architecture. Our work paves the way towards sharing of labeled 3D medical data which would increase generalizability of deep learning models for clinical use.
Subject(s)
Image Processing, Computer-Assisted , Magnetic Resonance Angiography , Humans , Image Processing, Computer-Assisted/methods , Imaging, Three-DimensionalABSTRACT
BACKGROUND: New ischaemic brain lesions on magnetic resonance imaging (MRI) are reported in up to 86% of patients after transcatheter edge-to-edge repair of the mitral valve (TEER-MV). Knowledge of the exact procedural step(s) that carry the highest risk for cerebral embolisation may help to further improve the procedure. AIMS: The aim of this study was to identify the procedural step(s) that are associated with an increased risk of cerebral embolisation during TEER-MV with the MitraClip system. Furthermore, the risk of overt stroke and silent brain ischaemia after TEER-MV was assessed. METHODS: In this prospective, pre-specified observational study, all patients underwent continuous transcranial Doppler examination during TEER-MV to detect microembolic signals (MES). MES were assigned to specific procedural steps: (1) transseptal puncture and placement of the guide, (2) advancing and adjustment of the clip in the left atrium, (3) device interaction with the MV, and (4) removal of the clip delivery system and the guide. Neurological examination using the National Institutes of Health Stroke Scale (NIHSS) and cerebral MRI were performed before and after TEER-MV. RESULTS: Fifty-four patients were included. The number of MES differed significantly between the procedural steps with the highest numbers observed during device interaction with the MV. Mild neurological deterioration (NIHSS ≤3) occurred in 9/54 patients. New ischaemic lesions were detected in 21/24 patients who underwent MRI. Larger infarct volume was significantly associated with neurological deterioration. CONCLUSIONS: Cerebral embolisation is immanent to TEER-MV and predominantly occurs during device interaction with the MV. Improvements to the procedure may focus on this procedural step.
Subject(s)
Heart Valve Prosthesis Implantation , Mitral Valve Insufficiency , Stroke , Cardiac Catheterization/methods , Heart Valve Prosthesis Implantation/methods , Humans , Mitral Valve/diagnostic imaging , Mitral Valve/surgery , Mitral Valve Insufficiency/diagnostic imaging , Mitral Valve Insufficiency/surgery , Prospective Studies , Stroke/etiology , Treatment OutcomeABSTRACT
BACKGROUND: Fluid-attenuated inversion recovery (FLAIR) sequences have gained a role to guide treatment of patients with unknown time of stroke symptom onset. Evolution of signal intensities in FLAIR is associated with time since stroke onset with continuous linear increases. AIMS: Estimating symptom onset during night-sleep in patients from the WAKE-UP trial based on relative signal intensities FLAIR (FLAIR-rSI) from acute stroke lesions an independent dataset (PRE-FLAIR study). METHODS: FLAIR-rSI was quantified in stroke lesions in PRE-FLAIR and WAKE-UP. The PRE-FLAIR study was a multicenter observational trial establishing FLAIR as a surrogate parameter for time since stroke onset. WAKE-UP was a randomized controlled trial that revealed a benefit for alteplase in patients selected based on a DWI-FLAIR mismatch. Stroke onset times were recorded in PRE-FLAIR and used to fit a linear regression model with FLAIR-rSI, adjusted for patient age and lesion volume. The model was applied to FLAIR-rSI of stroke lesions to estimate onset times in those patients enrolled in WAKE-UP who had symptom onset during night-sleep. RESULTS: FLAIR-rSI was quantified in 399 patients from PRE-FLAIR. Linear regression indicated a significant association of age (p = 0.001), lesion volume (p = 0.005) and FLAIR-rSI (p < 0.001) with time since symptom onset (adjusted R2 = 0.179). In 813 patients from WAKE-UP, distribution of times of last seen well, symptom recognition and MRI examination were recorded. Median times of last seen well were 1 h before midnight (IQR 2.4 h) and symptom recognition 7 h after midnight (IRQ 2.2 h). Based on the FLAIR-rSI profiles, we estimated median stroke onset 6.1 h after midnight (IQR 2.7 h). CONCLUSION: Nocturnal strokes during night-sleep may predominantly occur during the early morning hours. Our results are in line with evidence of characteristic diurnal patterns of cardiovascular events.
Subject(s)
Brain Ischemia , Stroke , Brain Ischemia/drug therapy , Diffusion Magnetic Resonance Imaging/methods , Humans , Magnetic Resonance Imaging/methods , Stroke/diagnostic imaging , Stroke/drug therapy , Thrombolytic Therapy/methods , Time FactorsABSTRACT
PURPOSE: Cerebral neoplasms of various histological origins may show comparable appearances on conventional Magnetic Resonance Imaging (MRI). Vessel size imaging (VSI) is an MRI technique that enables noninvasive assessment of microvasculature by providing quantitative estimates of microvessel size and density. In this study, we evaluated the potential of VSI to differentiate between brain tumor types based on their microvascular morphology. METHODS: Using a clinical 3T MRI scanner, VSI was performed on 25 patients with cerebral neoplasms, 10 with glioblastoma multiforme (GBM), 8 with primary CNS lymphoma (PCNSL) and 7 with cerebral lung cancer metastasis (MLC). Following the postprocessing of VSI maps, mean vessel diameter (vessel size index, vsi) and microvessel density (Q) were compared across tumors, peritumoral areas, and healthy tissues. RESULTS: The MLC tumors have larger and less dense microvasculature compared to PCNSLs in terms of vsi and Q (pâ¯= 0.0004 and pâ¯< 0.0001, respectively). GBM tumors have higher yet non-significantly different vsi values than PCNSLs (pâ¯= 0.065) and non-significant differences in Q. No statistically significant differences in vsi or Q were present between GBMs and MLCs. GBM tumor volume was positively correlated with vsi (râ¯= 0.502, pâ¯= 0.0017) and negatively correlated with Q (râ¯= -0.531, pâ¯= 0.0007). CONCLUSION: Conventional MRI parameters are helpful in differentiating between PCNSLs, GBMs, and MLCs. Additionally incorporating VSI parameters into the diagnostic protocol could help in further differentiating between PCNSLs and metastases and potentially between PCNSLs and GBMs. Future studies in larger patient cohorts are required to establish diagnostic cut-off values for VSI.
Subject(s)
Brain Neoplasms , Glioblastoma , Supratentorial Neoplasms , Brain Neoplasms/diagnostic imaging , Brain Neoplasms/pathology , Diagnosis, Differential , Glioblastoma/diagnostic imaging , Glioblastoma/pathology , Humans , Magnetic Resonance Imaging/methodsABSTRACT
BACKGROUND AND OBJECTIVES: Cerebral microbleeds (CMBs) are common in patients with acute ischemic stroke and are associated with increased risk of intracerebral hemorrhage (ICH) after intravenous thrombolysis. Whether CMBs modify the treatment effect of thrombolysis is unknown. METHODS: We performed a prespecified analysis of the prospective randomized controlled multicenter Efficacy and Safety of MRI-Based Thrombolysis in Wake-Up Stroke (WAKE-UP) trial including patients with acute ischemic stroke with unknown time of symptom onset and diffusion-weighted imaging-fluid-attenuated inversion recovery mismatch on MRI receiving alteplase or placebo. Patients were screened and enrolled between September 2012 and June 2017 (with final follow-up in September 2017). Patients were randomized to treatment with IV thrombolysis with alteplase at 0.9 mg/kg body weight or placebo. CMB status (presence, number, and distribution) was assessed after study completion by 3 raters blinded to clinical information following a standardized protocol. Outcome measures were excellent functional outcome at 90 days, defined by modified Rankin Scale (mRS) score ≤1, and symptomatic ICH according to National Institutes of Neurological Disease and Stroke trial criteria 22 to 36 hours after treatment. RESULTS: Of 503 patients enrolled in the WAKE-UP trial, 459 (91.3%; 288 [63%] men) were available for analysis. Ninety-eight (21.4%) had at least 1 CMB on baseline imaging; 45 (9.8%) had exactly 1 CMB; 37 (8.1%) had 2 to 4 CMBs; and 16 (3.5%) had ≥5 CMBs. Presence of CMBs was associated with a nonsignificant increased risk of symptomatic ICH (11.2% vs 4.2%; adjusted odds ratio [OR] 2.32, 95% confidence interval [CI] 0.99-5.43, p = 0.052) but had no effect on functional outcome at 90 days (mRS score ≤1: 45.8% vs 50.7%; adjusted OR 0.99, 95% CI 0.59-1.64, p = 0.955). Patients receiving alteplase had better functional outcome (mRS score ≤1: 54.6% vs 44.6%, adjusted OR 1.61, 95% CI 1.07-2.43, p = 0.022) without evidence of heterogeneity in relation to CMB presence (p of the interactive term = 0.546). Results were similar for subpopulations with strictly lobar (presumed cerebral amyloid angiopathy related) or not strictly lobar CMB distribution. DISCUSSION: In the randomized-controlled WAKE-UP trial, we saw no evidence of reduced treatment effect of alteplase in patients with acute ischemic stroke with ≥1 CMBs. Additional studies are needed to determine the treatment effect of alteplase and its benefit-harm ratio in patients with a larger number of CMBs. TRIAL REGISTRATION INFORMATION: ClinicalTrials.gov identifier NCT01525290; ClinicalTrialsRegister.EU identifier 2011-005906-32. CLASSIFICATION OF EVIDENCE: This study provides Class II evidence that for patients with acute ischemic stroke with unknown time of onset and diffusion-weighted imaging-fluid-attenuated inversion recovery mismatch who received IV alteplase, CMBs are not significantly associated with functional outcome at 90 days.
Subject(s)
Ischemic Stroke , Stroke , Cerebral Hemorrhage/diagnostic imaging , Cerebral Hemorrhage/drug therapy , Cerebral Hemorrhage/etiology , Fibrinolytic Agents , Humans , Male , Prospective Studies , Stroke/diagnostic imaging , Stroke/drug therapy , Thrombolytic Therapy/adverse effects , Thrombolytic Therapy/methods , Tissue Plasminogen Activator/adverse effects , Treatment OutcomeABSTRACT
BACKGROUND: Variants of the Circle of Willis (vCoW) may impede correct identification of ischemic lesion patterns and stroke etiology. We assessed reclassifications of ischemic lesion patterns due to vCoW. METHODS: We analyzed vCoW in patients with acute ischemic stroke from the 1000+ study using time-of-flight magnetic resonance angiography (TOF MRA) of intracranial arteries. We assessed A1 segment agenesis or hypoplasia in the anterior circulation and fetal posterior cerebral artery in the posterior circulation. Stroke patterns were classified as one or more-than-one territory stroke pattern. We examined associations between vCoW and stroke patterns and the frequency of reclassifications of stroke patterns due to vCoW. RESULTS: Of 1000 patients, 991 had evaluable magnetic resonance angiography. At least one vCoW was present in 37.1%. VCoW were more common in the posterior than in the anterior circulation (33.3% vs. 6.7%). Of 238 patients initially thought to have a more-than-one territory stroke pattern, 20 (8.4%) had to be reclassified to a one territory stroke pattern after considering vCoW. All these patients had fetal posterior cerebral artery and six (30%) additionally had carotid artery disease. Of 753 patients initially presumed to have a one-territory stroke pattern, four (0.5%) were reclassified as having more-than-one territory pattern. CONCLUSIONS: VCoW are present in about one in three stroke patients and more common in the posterior circulation. Reclassifications of stroke lesion patterns due to vCoW occurred predominantly in the posterior circulation with fetal posterior cerebral artery mimicking multiple territory stroke pattern. Considering vCoW in these cases may uncover symptomatic carotid disease.
Subject(s)
Ischemic Stroke , Stroke , Cerebral Infarction/complications , Cerebrovascular Circulation , Circle of Willis/abnormalities , Circle of Willis/diagnostic imaging , Circle of Willis/pathology , Humans , Magnetic Resonance Angiography , Stroke/etiologyABSTRACT
Lesion analysis is a fundamental and classical approach for inferring the causal contributions of brain regions to brain function. However, many studies have been limited by the shortcomings of methodology or clinical data. Aiming to overcome these limitations, we here use an objective multivariate approach based on game theory, Multi-perturbation Shapley value Analysis, in conjunction with data from a large cohort of 394 acute stroke patients, to derive causal contributions of brain regions to four principal functional components of the widely used National Institutes of Health Stroke Score measure. The analysis was based on a high-resolution parcellation of the brain into 294 grey and white matter regions. Through initial lesion symptom mapping for identifying all potential candidate regions and repeated iterations of the game-theoretical approach to remove non-significant contributions, the analysis derived the smallest sets of regions contributing to each of the four principal functional components as well as functional interactions among the regions. Specifically, the factor 'language and consciousness' was related to contributions of cortical regions in the left hemisphere, including the prefrontal gyrus, the middle frontal gyrus, the ventromedial putamen and the inferior frontal gyrus. Right and left motor functions were associated with contributions of the left and right dorsolateral putamen and the posterior limb of the internal capsule, correspondingly. Moreover, the superior corona radiata and the paracentral lobe of the right hemisphere as well as the right caudal area 23 of the cingulate gyrus were mainly related to left motor function, while the prefrontal gyrus, the external capsule and the sagittal stratum fasciculi of the left hemisphere contributed to right motor function. Our approach demonstrates a practically feasible strategy for applying an objective lesion inference method to a high-resolution map of the human brain and distilling a small, characteristic set of grey and white matter structures contributing to fundamental brain functions. In addition, we present novel findings of synergistic interactions between brain regions that provide insight into the functional organization of brain networks.
ABSTRACT
BACKGROUND: Arterial brain vessel segmentation allows utilising clinically relevant information contained within the cerebral vascular tree. Currently, however, no standardised performance measure is available to evaluate the quality of cerebral vessel segmentations. Thus, we developed a performance measure selection framework based on manual visual scoring of simulated segmentation variations to find the most suitable measure for cerebral vessel segmentation. METHODS: To simulate segmentation variations, we manually created non-overlapping segmentation errors common in magnetic resonance angiography cerebral vessel segmentation. In 10 patients, we generated a set of approximately 300 simulated segmentation variations for each ground truth image. Each segmentation was visually scored based on a predefined scoring system and segmentations were ranked based on 22 performance measures common in the literature. The correlation of visual scores with performance measure rankings was calculated using the Spearman correlation coefficient. RESULTS: The distance-based performance measures balanced average Hausdorff distance (rank = 1) and average Hausdorff distance (rank = 2) provided the segmentation rankings with the highest average correlation with manual rankings. They were followed by overlap-based measures such as Dice coefficient (rank = 7), a standard performance measure in medical image segmentation. CONCLUSIONS: Average Hausdorff distance-based measures should be used as a standard performance measure in evaluating cerebral vessel segmentation quality. They can identify more relevant segmentation errors, especially in high-quality segmentations. Our findings have the potential to accelerate the validation and development of novel vessel segmentation approaches.