Predicting final infarct size using acute and subacute multiparametric MRI measurements in patients with ischemic stroke

Artificial Intelligence Algorithm-Based MRI in Evaluating the Treatment Effect of Acute Cerebral Infarction

The study is aimed at exploring the application of artificial intelligence algorithm-based magnetic resonance imaging (MRI) in the diagnosis of acute cerebral infarction, expected to provide a reference for diagnosis and effect evaluation of acute cerebral infarction. In this study, 80 patients diagnosed with suspected acute cerebral infarction per Diagnostic Criteria for Cerebral Infarction were selected as the research subjects. MRI images were reconstructed by deep dictionary learning to improve their recognition ability. At the same time, the same diagnostic operation was performed by Computed Tomography (CT) images to compare with MRI. The results of the interalgorithm comparison showed the image reconstruction effect of the deep dictionary learning model is significantly better than SAE reconstruction, single-layer dictionary reconstruction model, and KAVD reconstruction. After comparison, the results of MRI based on artificial intelligence algorithm and CT evaluation were statistically significant (P < 0.05). In the lesion image, the diameter of MRI lesions (3.81 ± 0.77 cm) based on artificial intelligence algorithm and the diameter of lesions in CT (3.66 ± 1.65 cm) also had significant statistical significance (P < 0.05). The results showed that MRI based on deep learning was more sensitive than CT imaging for diagnosis and evaluation of patients with acute cerebral infarction, with only 1 case misdiagnosed. The rate of disease detection and lesion image quality had a higher improvement. The results can provide effective support for the clinical application of MRI based on artificial intelligence algorithm in the diagnosis of acute cerebral infarction.

PS1 FAD mutants decrease ephrinB2-regulated angiogenic functions, ischemia-induced brain neovascularization and neuronal survival

Microvascular pathology and ischemic lesions contribute substantially to neuronal dysfunction and loss that lead to Alzheimer disease (AD). To facilitate recovery, the brain stimulates neovascularization of damaged tissue via sprouting angiogenesis, a process regulated by endothelial cell (EC) sprouting and the EphB4/ephrinB2 system. Here, we show that in cultures of brain ECs, EphB4 stimulates the VE-cadherin/Rok-α angiogenic complexes known to mediate sprouting angiogenesis. Importantly, brain EC cultures expressing PS1 FAD mutants decrease the EphB4-stimulated γ-secretase cleavage of ephrinB2 and reduce production of the angiogenic peptide ephrinB2/CTF2, the VE-cadherin angiogenic complexes and EC sprouting and tube formation. These data suggest that FAD mutants may attenuate ischemia-induced brain angiogenesis. Supporting this hypothesis, ischemia-induced VE-cadherin angiogenic complexes, levels of neoangiogenesis marker Endoglin, vascular density, and cerebral blood flow recovery, are all decreased in brains of mouse models expressing PS1 FAD mutants. Ischemia-induced brain neuronal death and cognitive deficits also increase in these mice. Furthermore, a small peptide comprising the C-terminal sequence of peptide ephrinB2/CTF2 rescues angiogenic functions of brain ECs expressing PS1 FAD mutants. Together, our data show that PS1 FAD mutations impede the EphB4/ephrinB2-mediated angiogenic functions of ECs and impair brain neovascularization, neuronal survival and cognitive recovery following ischemia. Furthermore, our data reveal a novel brain angiogenic mechanism targeted by PS1 FAD mutants and a potential therapeutic target for ischemia-induced neurodegeneration. Importantly, FAD mutant effects occur in absence of neuropathological hallmarks of AD, supporting that such hallmarks may form downstream of mutant effects on neoangiogenesis and neuronal survival.

Machine learning identifies stroke features between species

Identification and localization of ischemic stroke (IS) lesions is routinely performed to confirm diagnosis, assess stroke severity, predict disability and plan rehabilitation strategies using magnetic resonance imaging (MRI). In basic research, stroke lesion segmentation is necessary to study complex peri-infarction tissue changes. Moreover, final stroke volume is a critical outcome evaluated in clinical and preclinical experiments to determine therapy or intervention success. Manual segmentations are performed but they require a specialized skill set, are prone to inter-observer variation, are not entirely objective and are often not supported by histology. The task is even more challenging when dealing with large multi-center datasets, multiple experimenters or large animal cohorts. On the other hand, current automatized segmentation approaches often lack histological validation, are not entirely user independent, are often based on single parameters, or in the case of complex machine learning methods, require vast training datasets and are prone to a lack of model interpretation.

Methods: We induced IS using the middle cerebral artery occlusion model on two rat cohorts. We acquired apparent diffusion coefficient (ADC) and T2-weighted (T2W) images at 24 h and 1-week after IS induction. Subsets of the animals at 24 h and 1-week post IS were evaluated using histology and immunohistochemistry. Using a Gaussian mixture model, we segmented voxel-wise interactions between ADC and T2W parameters at 24 h using one of the rat cohorts. We then used these segmentation results to train a random forest classifier, which we applied to the second rat cohort. The algorithms' stroke segmentations were compared to manual stroke delineations, T2W and ADC thresholding methods and the final stroke segmentation at 1-week. Volume correlations to histology were also performed for every segmentation method. Metrics of success were calculated with respect to the final stroke volume. Finally, the trained random forest classifier was tested on a human dataset with a similar temporal stroke on-set. Manual segmentations, ADC and T2W thresholds were again used to evaluate and perform comparisons with the proposed algorithms' output.

Results: In preclinical rat data our framework significantly outperformed commonly applied automatized thresholding approaches and segmented stroke regions similarly to manual delineation. The framework predicted the localization of final stroke regions in 1-week post-stroke MRI with a median Dice similarity coefficient of 0.86, Matthew's correlation coefficient of 0.80 and false positive rate of 0.04. The predicted stroke volumes also strongly correlated with final histological stroke regions (Pearson correlation = 0.88, P < 0.0001). Lastly, the stroke region characteristics identified by our framework in rats also identified stroke lesions in human brains, largely outperforming thresholding approaches in stroke volume prediction (P<0.01).

Conclusion: Our findings reveal that the segmentation produced by our proposed framework using 24 h MRI rat data strongly correlated with the final stroke volume, denoting a predictive effect. In addition, we show for the first time that the stroke imaging features can be directly translated between species, allowing identification of acute stroke in humans using the model trained on animal data. This discovery reduces the gap between the clinical and preclinical fields, unveiling a novel approach to directly co-analyze clinical and preclinical data. Such methods can provide further biological insights into human stroke and highlight the differences between species in order to help improve the experimental setups and animal models of the disease.

Matrix Metalloproteinase-9 Relationship With Infarct Growth and Hemorrhagic Transformation in the Era of Thrombectomy

Objective: To assess the relationship between matrix metalloproteinase 9 (MMP-9), a proteolytic enzyme involved in the breakdown of the blood-brain barrier, and infarct growth and hemorrhagic transformation in acute ischemic stroke (AIS) with large vessel occlusion (LVO) in the era of mechanical thrombectomy (MT) using the kinetics of MMP-9 and sequential magnetic resonance imaging (MRI).

Methods: HIBISCUS-STROKE is a cohort study including AIS patients with LVO treated with MT following admission MRI. Patients underwent sequential assessment of MMP-9, follow-up CT at day 1, and MRI at day 6. The CT scan at day 1 classified any hemorrhagic transformation according to the European Co-operative Acute Stroke Study-II (ECASS II) classification. Infarct growth was defined as the difference between final Fluid-Attenuated Inversion Recovery volume and baseline diffusion-weighted imaging volume. Conditional logistic regression analyses were adjusted for main confounding variables including reperfusion status.

Results: One hundred and forty-eight patients represent the study population. A high MMP-9 level at 6 h from admission (H6) (p = 0.02), a high glucose level (p = 0.01), a high temperature (p = 0.04), and lack of reperfusion (p = 0.02) were associated with infarct growth. A high MMP-9 level at H6 (p = 0.03), a high glucose level (p = 0.03) and a long delay from symptom onset to groin puncture (p = 0.01) were associated with hemorrhagic transformation.

Conclusions: In this MT cohort study, MMP-9 level at H6 predicts infarct growth and hemorrhagic transformation.

Arterial spin labelling shows functional depression of non-lesion tissue in chronic Wernicke's aphasia

Behavioural impairment post-stroke is a consequence of structural damage and altered functional network dynamics. Hypoperfusion of intact neural tissue is frequently observed in acute stroke, indicating reduced functional capacity of regions outside the lesion. However, cerebral blood flow (CBF) is rarely investigated in chronic stroke. This study investigated CBF in individuals with chronic Wernicke's aphasia (WA) and examined the relationship between lesion, CBF and neuropsychological impairment.

Arterial spin labelling CBF imaging and structural MRIs were collected in 12 individuals with chronic WA and 13 age-matched control participants. Joint independent component analysis (jICA) investigated the relationship between structural lesion and hypoperfusion. Partial correlations explored the relationship between lesion, hypoperfusion and language measures.

Joint ICA revealed significant differences between the control and WA groups reflecting a large area of structural lesion in the left posterior hemisphere and an associated area of hypoperfusion extending into grey matter surrounding the lesion. Small regions of remote cortical hypoperfusion were observed, ipsilateral and contralateral to the lesion. Significant correlations were observed between the neuropsychological measures (naming, repetition, reading and semantic association) and the jICA component of interest in the WA group. Additional ROI analyses found a relationship between perfusion surrounding the core lesion and the same neuropsychological measures.

This study found that core language impairments in chronic WA are associated with a combination of structural lesion and abnormal perfusion in non-lesioned tissue. This indicates that post-stroke impairments are due to a wider disruption of neural function than observable on structural T1w MRI.

ANTONIA perfusion and stroke. A software tool for the multi-purpose analysis of MR perfusion-weighted datasets and quantitative ischemic stroke assessment

OBJECTIVES

The objective of this work is to present the software tool ANTONIA, which has been developed to facilitate a quantitative analysis of perfusion-weighted MRI (PWI) datasets in general as well as the subsequent multi-parametric analysis of additional datasets for the specific purpose of acute ischemic stroke patient dataset evaluation.

METHODS

Three different methods for the analysis of DSC or DCE PWI datasets are currently implemented in ANTONIA, which can be case-specifically selected based on the study protocol. These methods comprise a curve fitting method as well as a deconvolution-based and deconvolution-free method integrating a previously defined arterial input function. The perfusion analysis is extended for the purpose of acute ischemic stroke analysis by additional methods that enable an automatic atlas-based selection of the arterial input function, an analysis of the perfusion-diffusion and DWI-FLAIR mismatch as well as segmentation-based volumetric analyses.

RESULTS

For reliability evaluation, the described software tool was used by two observers for quantitative analysis of 15 datasets from acute ischemic stroke patients to extract the acute lesion core volume, FLAIR ratio, perfusion-diffusion mismatch volume with manually as well as automatically selected arterial input functions, and follow-up lesion volume. The results of this evaluation revealed that the described software tool leads to highly reproducible results for all parameters if the automatic arterial input function selection method is used.

CONCLUSION

Due to the broad selection of processing methods that are available in the software tool, ANTONIA is especially helpful to support image-based perfusion and acute ischemic stroke research projects.

Infarct volume prediction using apparent diffusion coefficient maps during middle cerebral artery occlusion and soon after reperfusion in the rat

Middle cerebral artery occlusion (MCAO) in rodents causes brain infarctions of variable sizes that depend on multiple factors, particularly in models of ischemia/reperfusion. This is a major problem for infarct volume comparisons between different experimental groups since unavoidable variability can induce biases in the results and imposes the use of large number of subjects. MRI can help to minimize these difficulties by ensuring that the severity of ischemia is comparable between groups. Furthermore, several studies showed that infarct volumes can be predicted with MRI data obtained soon after ischemia onset. However, such predictive studies require multiparametric MRI acquisitions that cannot be routinely performed, and data processing using complex algorithms that are often not available. The aim here was to provide a simplified method for infarct volume prediction using apparent diffusion coefficient (ADC) data in a model of transient MCAO in rats. ADC images were obtained before, during MCAO and after 60 min of reperfusion. Probability histograms were generated using ADC data obtained either during MCAO, after reperfusion, or both combined. The results were compared to real infarct volumes, i.e.T2 maps obtained at day 7. Assessment of the performance of the estimations showed better results combining ADC data obtained during occlusion and at reperfusion. Therefore, ADC data alone can provide sufficient information for a reasonable prediction of infarct volume if the MRI information is obtained both during the occlusion and soon after reperfusion. This approach can be used to check whether drug administration after MRI acquisition can change infarct volume prediction.

A comparison of VLSM and VBM in a cohort of patients with post-stroke aphasia

Studies attempting to map post-stroke cognitive or motor symptoms to lesion location have been available in the literature for over 150 years. In the last two decades, two computational techniques have been developed to identify the lesion sites associated with behavioural impairments. Voxel Based Morphometry (VBM) has now been used extensively for this purpose in many different patient populations. More recently, Voxel-based Lesion Symptom Mapping (VLSM) was developed specifically for the purpose of identifying lesion–symptom relationships in stroke patients, and has been used extensively to study, among others functions, language, motor abilities and attention. However, no studies have compared the results of these two techniques so far. In this study we compared VLSM and VBM in a cohort of 20 patients with chronic post-stroke aphasia. Comparison of the two techniques showed overlap in regions previously found to be relevant for the tasks used, suggesting that using both techniques and looking for overlaps between them can increase the reliability of the results obtained. However, overall VBM and VLSM provided only partially concordant results and the differences between the two techniques are discussed.

Predicting final extent of ischemic infarction using artificial neural network analysis of multi-parametric MRI in patients with stroke

In hemispheric ischemic stroke, the final size of the ischemic lesion is the most important correlate of clinical functional outcome. Using a set of acute-phase MR images (Diffusion-weighted - DWI, T₁-weighted – T1WI, T₂-weighted-T2WI, and proton density weighted - PDWI) for inputs, and the chronic T2WI at 3 months as an outcome measure, an Artificial Neural Network (ANN) was trained to predict the 3-month outcome in the form of a voxel-by-voxel forecast of the chronic T2WI. The ANN was trained and tested using 12 subjects (with 83 slices and 140218 voxels) using a leave-one-out cross-validation method with calculation of the Area Under the Receiver Operator Characteristic Curve (AUROC) for training, testing and optimization of the ANN. After training and optimization, the ANN produced maps of predicted outcome that were well correlated (r = 0.80, p<0.0001) with the T2WI at 3 months for all 12 patients. This result implies that the trained ANN can provide an estimate of 3-month ischemic lesion on T2WI in a stable and accurate manner (AUROC = 0.89).

Hyperacute detection of neurofilament heavy chain in serum following stroke: a transient sign

Serological biomarkers which enable quick and reliable diagnosis or measurement of the extent of irreversible brain injury early in the course of stroke are eagerly awaited. Neurofilaments (Nf) are a group of proteins integrated into the scaffolding of the neuronal and axonal cytoskeleton and an established biomarker of neuro-axonal damage. The Nf heavy chain (NfH(SMI35)) was assessed together with brain-specific astroglial proteins GFAP and S100B in hyperacute stroke (6 and 24 h from symptom onset) and daily for up to 6 days. Twenty-two patients with suspected stroke (median NIHSS 8) were recruited in a prospective observational study. Evidence for an ischaemic or haemorrhagic lesion on neuroimaging was found in 18 (ischaemia n = 16, intracerebral haemorrhage n = 2). Serum NfH(SMI35) levels became detectable within 24 h post-stroke (P < 0.0001) and elevated levels persisted over the study course. While GFAP was not detectable during the entire course, S100B levels peaked at the end of the observation period. The data indicate that significant in vivo information on the pathophysiology of stroke may be obtained by the determination of NfH(SMI35). Further studies are required to evaluate whether NfH(SMI35) in hyperacute stroke reflects the extent of focal ischaemic injury seen on neuroimaging or is a consequence of more diffuse neuro-axonal damage.

Prediction of infarct volume and neurologic outcome by using automated multiparametric perfusion-weighted magnetic resonance imaging in a primate model of permanent middle cerebral artery occlusion

By optimizing thresholds, we identified the perfusion-weighted magnetic resonance imaging (PWI) parameters that accurately predict final infarct volume and neurologic outcome in a primate model of permanent middle cerebral artery (MCA) occlusion. Ten cynomolgus monkeys underwent PWI and diffusion-weighted imaging (DWI) at 3 and 47 hours, respectively, after right MCA occlusion using platinum coils, and were killed at 48 hours. Volumes of the hypoperfused areas on PWI were automatically measured using different thresholds and 11 parametric maps to determine the optimum threshold (at which least difference was found between the average volumes on PWI and those determined using specimens or DWI). In the case of arrival time (AT), cerebral blood volume (CBV), time to peak (TTP), time to maximum (T(max)), and cerebral blood flow (CBF) determined using deconvolution techniques, the volume of the hypoperfused area significantly correlated with the infarct volumes and the neurologic deficit scores with small variations, whereas in the case of mean transit time and nondeconvolution CBF, relatively poor correlations with large variations were seen. At optimum threshold, AT, CBV, TTP, T(max), and deconvolution CBF can accurately predict the final infarct volume and neurologic outcome in monkeys with permanent MCA occlusion.

Multiparametric magnetic resonance imaging of brain disorders

Magnetic resonance imaging (MRI) has been shown to improve the diagnosis and management of patients with brain disorders. Multiparametric MRI offers the possibility of noninvasively assessing multiple facets of pathophysiological processes that exist simultaneously, thereby further assisting in patient treatment management. Voxel-based analysis approaches, such as tissue theme mapping, have the benefit over volumetric approaches in being able to identify spatially heterogeneous colocalized changes on multiple parametric MR images that are not readily discernible. Tissue theme maps seem to be a promising tool for integrating the plethora of novel imaging contrasts that are being developed for the noninvasive investigation of the different stages of disease progression into easily interpretable maps of brain injury. We describe here various implementations for combining multiparametric imaging and their merits in the evaluation of brain diseases.

Correlating tissue outcome with quantitative multiparametric MRI of acute cerebral ischemia in rats

Predicting tissue outcome remains a challenge for stroke magnetic resonance imaging (MRI). In this study, we have acquired multiparametric MRI data sets (including absolute T(1), T(2), diffusion, T(1rho) using continuous wave and adiabatic pulse approaches, cerebral blood flow (CBF), and amide proton transfer ratio (APTR) images) during and after 65 mins of middle cerebral artery occlusion (MCAo) in rats. The MRI scans were repeated 24 h after MCAo, when the animals were killed for quantitative histology. Magnetic resonance imaging parameters acquired at three acute time points were correlated with regionally matching cell count at 24 h. The results emphasize differences in the temporal profile of individual MRI contrasts during MCAo and especially during early reperfusion, and suggest that complementary information from CBF and tissue damage can be obtained with appropriate MRI contrasts. The data show that by using three to four MRI parameters, sensitive to both hemodynamic changes and different aspects of parenchymal changes, the fate of the tissue can be predicted with increased correlation compared with single-parameter techniques. Combined multiparametric MRI data and multiparametric analysis may provide an excellent tool for preclinical testing of new treatments and also has the potential to facilitate decision-making in the management of acute stroke patients.

Inter-individual variability in the capacity for motor recovery after ischemic stroke

BACKGROUND

Motor recovery after stroke is predicted only moderately by clinical variables, implying that there is still a substantial amount of unexplained, biologically meaningful variability in recovery. Regression diagnostics can indicate whether this is associated simply with Gaussian error or instead with multiple subpopulations that vary in their relationships to the clinical variables.

OBJECTIVE

To perform regression diagnostics on a linear model for recovery versus clinical predictors.

METHODS

Forty-one patients with ischemic stroke were studied. Impairment was assessed using the upper extremity Fugl-Meyer Motor Score. Motor recovery was defined as the change in the upper extremity Fugl-Meyer Motor Score from 24 to 72 hours after stroke to 3 or 6 months later. The clinical predictors in the model were age, gender, infarct location (subcortical vs cortical), diffusion weighted imaging infarct volume, time to reassessment, and acute upper extremity Fugl-Meyer Motor Score. Regression diagnostics included a Kolmogorov-Smirnov test for Gaussian errors and a test for outliers using Studentized deleted residuals.

RESULTS

In the random sample, clinical variables explained only 47% of the variance in recovery. Among the patients with the most severe initial impairment, there was a set of regression outliers who recovered very poorly. With the outliers removed, explained variance in recovery increased to 89%, and recovery was well approximated by a proportional relationship with initial impairment (recovery congruent with 0.70 x initial impairment).

CONCLUSIONS

Clinical variables only moderately predict motor recovery. Regression diagnostics demonstrated the existence of a subpopulation of outliers with severe initial impairment who show little recovery. When these outliers were removed, clinical variables were good predictors of recovery among the remaining patients, showing a tight proportional relationship to initial impairment.

Fiber connections between the cerebral cortex and the corpus callosum in Alzheimer's disease: a diffusion tensor imaging and voxel-based morphometry study

Regional cortical atrophy in Alzheimer's disease (AD) most likely reflects the loss of cortical neurons. Several diffusion tensor imaging studies reported reduced fractional anisotropy (FA) in the corpus callosum in AD. The aim of this study was to investigate the association between reduced FA in the corpus callosum and gray matter atrophy in AD. Thirteen patients with AD with a mean (+/-standard deviation) age of 68.3 years (+/-11.5) and mean Mini Mental State Examination (MMSE) score of 21.8 (+/-4.8) were recruited. There were 13 control subjects with a mean age of 66.7 years (+/-6.4) and MMSE of 29.1 (+/-0.7). We used voxel-based morphometry of gray matter maps and region of interest-based analysis of FA in the corpus callosum. FA values of the anterior corpus callosum in AD patients were significantly correlated with gray matter volume in the prefrontal cortex and left parietal lobes. FA values of the posterior corpus callosum were significantly correlated with gray matter volume in the bilateral frontal, temporal, right parietal, and occipital lobes. In control subjects, no correlations were detected. Our findings suggest that decline of FA in the corpus callosum may be related to neuronal degeneration in corresponding cortical areas.