Distinct brain morphometry patterns revealed by deep learning improve prediction of post-stroke aphasia severity.

BACKGROUND: Emerging evidence suggests that post-stroke aphasia severity depends on the integrity of the brain beyond the lesion. While measures of lesion anatomy and brain integrity combine synergistically to explain aphasic symptoms, substantial interindividual variability remains unaccounted. One explanatory factor may be the spatial distribution of morphometry beyond the lesion (e.g., atrophy), including not just specific brain areas, but distinct three-dimensional patterns. METHODS: Here, we test whether deep learning with Convolutional Neural Networks (CNNs) on whole brain morphometry (i.e., segmented tissue volumes) and lesion anatomy better predicts chronic stroke individuals with severe aphasia (N = 231) than classical machine learning (Support Vector Machines; SVMs), evaluating whether encoding spatial dependencies identifies uniquely predictive patterns. RESULTS: CNNs achieve higher balanced accuracy and F1 scores, even when SVMs are nonlinear or integrate linear or nonlinear dimensionality reduction. Parity only occurs when SVMs access features learned by CNNs. Saliency maps demonstrate that CNNs leverage distributed morphometry patterns, whereas SVMs focus on the area around the lesion. Ensemble clustering of CNN saliencies reveals distinct morphometry patterns unrelated to lesion size, consistent across individuals, and which implicate unique networks associated with different cognitive processes as measured by the wider neuroimaging literature. Individualized predictions depend on both ipsilateral and contralateral features outside the lesion. CONCLUSIONS: Three-dimensional network distributions of morphometry are directly associated with aphasia severity, underscoring the potential for CNNs to improve outcome prognostication from neuroimaging data, and highlighting the prospective benefits of interrogating spatial dependence at different scales in multivariate feature space.

Some stroke survivors experience difficulties understanding and producing language. We performed brain imaging to capture information about brain structure in stroke survivors and used it to predict which survivors have more severe language problems. We found that a type of artificial intelligence (AI) specifically designed to find patterns in spatial data was more accurate at this task than more traditional methods. AI found more complex patterns of brain structure that distinguish stroke survivors with severe language problems by analyzing the brain's spatial properties. Our findings demonstrate that AI tools can provide new information about brain structure and function following stroke. With further developments, these models may be able to help clinicians understand the extent to which language problems can be improved after a stroke.

brainlife.io: a decentralized and open-source cloud platform to support neuroscience research.

Neuroscience is advancing standardization and tool development to support rigor and transparency. Consequently, data pipeline complexity has increased, hindering FAIR (findable, accessible, interoperable and reusable) access. brainlife.io was developed to democratize neuroimaging research. The platform provides data standardization, management, visualization and processing and automatically tracks the provenance history of thousands of data objects. Here, brainlife.io is described and evaluated for validity, reliability, reproducibility, replicability and scientific utility using four data modalities and 3,200 participants.

Distinct brain morphometry patterns revealed by deep learning improve prediction of aphasia severity.

Emerging evidence suggests that post-stroke aphasia severity depends on the integrity of the brain beyond the stroke lesion. While measures of lesion anatomy and brain integrity combine synergistically to explain aphasic symptoms, significant interindividual variability remains unaccounted for. A possible explanatory factor may be the spatial distribution of brain atrophy beyond the lesion. This includes not just the specific brain areas showing atrophy, but also distinct three-dimensional patterns of atrophy. Here, we tested whether deep learning with Convolutional Neural Networks (CNN) on whole brain morphometry (i.e., segmented tissue volumes) and lesion anatomy can better predict which individuals with chronic stroke (N=231) have severe aphasia, and whether encoding spatial dependencies in the data might be capable of improving predictions by identifying unique individualized spatial patterns. We observed that CNN achieves significantly higher accuracy and F1 scores than Support Vector Machine (SVM), even when the SVM is nonlinear or integrates linear and nonlinear dimensionality reduction techniques. Performance parity was only achieved when the SVM was directly trained on the latent features learned by the CNN. Saliency maps demonstrated that the CNN leveraged widely distributed patterns of brain atrophy predictive of aphasia severity, whereas the SVM focused almost exclusively on the area around the lesion. Ensemble clustering of CNN saliency maps revealed distinct morphometry patterns that were unrelated to lesion size, highly consistent across individuals, and implicated unique brain networks associated with different cognitive processes as measured by the wider neuroimaging literature. Individualized predictions of severity depended on both ipsilateral and contralateral features outside of the location of stroke. Our findings illustrate that three-dimensional network distributions of atrophy in individuals with aphasia are directly associated with aphasia severity, underscoring the potential for deep learning to improve prognostication of behavioral outcomes from neuroimaging data, and highlighting the prospective benefits of interrogating spatial dependence at different scales in multivariate feature space.

brainlife.io: A decentralized and open source cloud platform to support neuroscience research.

Neuroscience research has expanded dramatically over the past 30 years by advancing standardization and tool development to support rigor and transparency. Consequently, the complexity of the data pipeline has also increased, hindering access to FAIR data analysis to portions of the worldwide research community. brainlife.io was developed to reduce these burdens and democratize modern neuroscience research across institutions and career levels. Using community software and hardware infrastructure, the platform provides open-source data standardization, management, visualization, and processing and simplifies the data pipeline. brainlife.io automatically tracks the provenance history of thousands of data objects, supporting simplicity, efficiency, and transparency in neuroscience research. Here brainlife.io's technology and data services are described and evaluated for validity, reliability, reproducibility, replicability, and scientific utility. Using data from 4 modalities and 3,200 participants, we demonstrate that brainlife.io's services produce outputs that adhere to best practices in modern neuroscience research.

Cerebral blood flow in patients recovered from mild COVID-19.

BACKGROUND AND PURPOSE: Cerebral hypoperfusion has been described in both severe and mild forms of symptomatic Coronavirus Disease 2019 (COVID-19) infection. The purpose of this study was to investigate global and regional cerebral blood flow (CBF) in asymptomatic COVID-19 patients. METHODS: Cases with mild COVID-19 infection and age-, sex-, and race-matched healthy controls were drawn from the Aging Brain Consortium at The University of South Carolina data repository. Demographics, risk factors, and data from the Montreal Cognitive Assessment were collected. Mean CBF values for gray matter (GM), white matter (WM), and the whole brain were calculated by averaging CBF values of standard space-normalized CBF image values falling within GM and WM masks. Whole brain region of interest-based analyses were used to create standardized CBF maps and explore differences between groups. RESULTS: Twenty-eight cases with prior mild COVID-19 infection were compared with 28 controls. Whole-brain CBF (46.7 ± 5.6 vs. 49.3 ± 3.7, p = .05) and WM CBF (29.3 ± 2.6 vs. 31.0 ± 1.6, p = .03) were noted to be significantly lower in COVID-19 cases as compared to controls. Predictive models based on these data predicted COVID-19 group membership with a high degree of accuracy (85.2%, p < .001), suggesting CBF patterns are an imaging marker of mild COVID-19 infection. CONCLUSION: In this study, lower WM CBF, as well as widespread regional CBF changes identified using quantitative MRI, was found in mild COVID-19 patients. Further studies are needed to determine the reliability of this newly identified COVID-19 brain imaging marker and determine what drives these CBF changes.

Right hemispheric white matter hyperintensities improve the prediction of spatial neglect severity in acute stroke.

White matter hyperintensities (WMH) are frequently observed in brain scans of elderly people. They are associated with an increased risk of stroke, cognitive decline, and dementia. However, it is unknown yet if measures of WMH provide information that improve the understanding of poststroke outcome compared to only state-of-the-art stereotaxic structural lesion data. We implemented high-dimensional machine learning models, based on support vector regression, to predict the severity of spatial neglect in 103 acute right hemispheric stroke patients. We found that (1) the additional information of right hemispheric or bilateral voxel-based topographic WMH extent indeed yielded a significant improvement in predicting acute neglect severity (compared to the voxel-based stroke lesion map alone). (2) Periventricular WMH appeared more relevant for prediction than deep subcortical WMH. (3) Among different measures of WMH, voxel-based maps as measures of topographic extent allowed more accurate predictions compared to the use of traditional ordinally assessed visual rating scales (Fazekas-scale, Cardiovascular Health Study-scale). In summary, topographic WMH appear to be a valuable clinical imaging biomarker for predicting the severity of cognitive deficits and bears great potential for rehabilitation guidance of acute stroke patients.

Disconnectomics to unravel the network underlying deficits of spatial exploration and attention.

Spatial attention and exploration are related to a predominantly right hemispheric network structure. However, the areas of the brain involved and their exact role is still debated. Spatial neglect following right hemispheric stroke lesions has been frequently viewed as a model to study these processes in humans. Previous investigations on the anatomical basis on spatial neglect predominantly focused on focal brain damage and lesion-behaviour mapping analyses. This approach might not be suited to detect remote areas structurally spared but which might contribute to the behavioural deficit. In the present study of a sample of 203 right hemispheric stroke patients, we combined connectome lesion-symptom mapping with multivariate support vector regression to unravel the complex and disconnected network structure in spatial neglect. We delineated three central nodes that were extensively disconnected from other intrahemispheric areas, namely the right superior parietal lobule, the insula, and the temporal pole. Additionally, the analysis allocated central roles within this network to the inferior frontal gyrus (pars triangularis and opercularis), right middle temporal gyrus, right temporal pole and left and right orbitofrontal cortices, including interhemispheric disconnection. Our results suggest that these structures-although not necessarily directly damaged-might play a role within the network underlying spatial neglect in humans.

Brain age predicts long-term recovery in post-stroke aphasia.

The association between age and language recovery in stroke remains unclear. Here, we used neuroimaging data to estimate brain age, a measure of structural integrity, and examined the extent to which brain age at stroke onset is associated with (i) cross-sectional language performance, and (ii) longitudinal recovery of language function, beyond chronological age alone. A total of 49 participants (age: 65.2 ± 12.2 years, 25 female) underwent routine clinical neuroimaging (T1) and a bedside evaluation of language performance (Bedside Evaluation Screening Test-2) at onset of left hemisphere stroke. Brain age was estimated from enantiomorphically reconstructed brain scans using a machine learning algorithm trained on a large sample of healthy adults. A subsample of 30 participants returned for follow-up language assessments at least 2 years after stroke onset. To account for variability in age at stroke, we calculated proportional brain age difference, i.e. the proportional difference between brain age and chronological age. Multiple regression models were constructed to test the effects of proportional brain age difference on language outcomes. Lesion volume and chronological age were included as covariates in all models. Accelerated brain age compared with age was associated with worse overall aphasia severity (F(1, 48) = 5.65, P = 0.022), naming (F(1, 48) = 5.13, P = 0.028), and speech repetition (F(1, 48) = 8.49, P = 0.006) at stroke onset. Follow-up assessments were carried out ≥2 years after onset; decelerated brain age relative to age was significantly associated with reduced overall aphasia severity (F(1, 26) = 5.45, P = 0.028) and marginally failed to reach statistical significance for auditory comprehension (F(1, 26) = 2.87, P = 0.103). Proportional brain age difference was not found to be associated with changes in naming (F(1, 26) = 0.23, P = 0.880) and speech repetition (F(1, 26) = 0.00, P = 0.978). Chronological age was only associated with naming performance at stroke onset (F(1, 48) = 4.18, P = 0.047). These results indicate that brain age as estimated based on routine clinical brain scans may be a strong biomarker for language function and recovery after stroke.

Mapping the human praxis network: an investigation of white matter disconnection in limb apraxia of gesture production.

Left hemispheric cerebral stroke can cause apraxia, a motor cognitive disorder characterized by deficits of higher-order motor skills such as the failure to accurately produce meaningful gestures. This disorder provides unique insights into the anatomical and cognitive architecture of the human praxis system. The present study aimed to map the structural brain network that is damaged in apraxia. We assessed the ability to perform meaningful gestures with the hand in 101 patients with chronic left hemisphere stroke. Structural white matter fibre damage was directly assessed by diffusion tensor imaging and fractional anisotropy mapping. We used multivariate topographical inference on tract-based fractional anisotropy topographies to identify white matter disconnection associated with apraxia. We found relevant pathological white matter alterations in a densely connected fronto-temporo-parietal network of short and long association fibres. Hence, the findings suggest that heterogeneous topographical results in previous lesion mapping studies might not only result from differences in study design, but also from the general methodological limitations of univariate topographical mapping in uncovering the structural praxis network. A striking role of middle and superior temporal lobe disconnection, including temporo-temporal short association fibres, was found, suggesting strong involvement of the temporal lobe in the praxis network. Further, the results stressed the importance of subcortical disconnections for the emergence of apractic symptoms. Our study provides a fine-grain view into the structural connectivity of the human praxis network and suggests a potential value of disconnection measures in the clinical prediction of behavioural post-stroke outcome.

Effect of Stroke on Contralateral Functional Connectivity.

Introduction: Stroke can induce large-scale functional reorganization of the brain; however, the spatial patterns of this reorganization remain largely unknown. Methods: Using a large (N = 116) sample of participants who were in the chronic stages of stroke, we present a systematic study of the association between brain damage and functional connectivity (FC) within the intact hemisphere. We computed correlations between regional cortical damage and contralateral FC. Results: We identified left-hemisphere regions that had the most pronounced effect on the right-hemisphere FC, and, conversely, right-hemisphere connections where the effect of damage was particularly strong. Notably, the vast majority of significant correlations were positive: damage was associated with an increase in regional contralateral connectivity. Discussion: These findings lend evidence of the reorganization of contralateral cortical networks as a response to brain damage, which is more pronounced in a set of well-connected regions where connectivity increases with the amount of damage. Impact statement The relatively large sample size combined with our best-of-breed analysis methods provides us with sufficient statistical power and spatial sensitivity to identify a set of brain regions where damage has the strongest impact on contralateral networks, and a set of contralateral functional connections that increase in strength in response to brain damage. Our results demonstrate that the brain's ability to reorganize itself after extensive damage is not distributed equally in space, but is more likely to occur in specific core regions. We believe that the associations between brain damage and increased connectivity in the "intact" hemisphere provide novel, and important, insight into the plasticity of the adult brain.

Independent contributions of structural and functional connectivity: Evidence from a stroke model.

Altered functional connectivity is related to severity of language impairment in poststroke aphasia. However, it is not clear whether this finding specifically reflects loss of functional coherence, or more generally, is related to decreased structural connectivity due to cortical necrosis. The aim of the current study was to investigate this issue by factoring out structural connectivity from functional connectivity measures and then relating the residual data to language performance poststroke. Ninety-seven participants with a history of stroke were assessed using language impairment measures (Auditory Verbal Comprehension and Spontaneous Speech scores from the Western Aphasia Battery-Revised) and MRI (structural, diffusion tensor imaging, and resting-state functional connectivity). We analyzed the association between functional connectivity and language and controlled for multiple potential neuroanatomical confounders, namely structural connectivity. We identified functional connections within the left hemisphere ventral stream where decreased functional connectivity, independent of structural connectivity, was associated with speech comprehension impairment. These connections exist in frontotemporal and temporoparietal regions. Our results suggest poor speech comprehension in aphasia is at least partially caused by loss of cortical synchrony in a left hemisphere ventral stream network and is not only reflective of localized necrosis or structural connectivity.

Clinical Implementation of Transcranial Direct Current Stimulation in Aphasia: A Survey of Speech-Language Pathologists.

Purpose The objectives of this study are to (a) identify speech-language pathologists' (SLPs') familiarity with transcranial direct current stimulation (tDCS), (b) quantify what SLPs consider necessary tDCS-related improvement in aphasia severity (i.e., tDCS enhancement; desired improvement above and beyond traditional behavioral therapy) to implement this adjuvant therapy for the clinical management of aphasia, and (c) identify concerns that could potentially hinder the clinical adoption of tDCS. Method A brief (14-question) survey was disseminated via e-mail and social media outlets targeting SLPs working with individuals with aphasia. Results Two hundred twenty-one individuals responded, and 155 valid surveys were analyzed. Seventy-one percent of participants reported familiarity with tDCS prior to taking the survey. Clinicians reported a desired mean enhancement of 22.9% additional points on the Western Aphasia Battery-Revised Aphasia Quotient. Importantly, 94.2% of SLPs reported concerns regarding the implementation of tDCS in clinical settings (i.e., safety, cost, administrative approval, reimbursement and training). Conclusions This is the first study to identify SLPs' perspectives regarding the clinical adoption of tDCS. Results suggest the majority of queried SLPs were familiar with tDCS prior to taking the survey. Although SLPs report a desired improvement of approximately 23% additional points on the Western Aphasia Battery-Revised Aphasia Quotient to consider adopting tDCS into practice, many SLPs reported concerns regarding clinical adoption. Responses from the current survey offer important preliminary evidence to begin bridging the research-to-practice gap as it relates to the clinical implementation of tDCS. Relatedly, these results will inform future clinical trials.

Using machine learning-based lesion behavior mapping to identify anatomical networks of cognitive dysfunction: Spatial neglect and attention.

Previous lesion behavior studies primarily used univariate lesion behavior mapping techniques to map the anatomical basis of spatial neglect after right brain damage. These studies led to inconsistent results and lively controversies. Given these inconsistencies, the idea of a wide-spread network that might underlie spatial orientation and neglect has been pushed forward. In such case, univariate lesion behavior mapping methods might have been inherently limited in detecting the presumed network due to limited statistical power. By comparing various univariate analyses with multivariate lesion-mapping based on support vector regression, we aimed to validate the network hypothesis directly in a large sample of 203 newly recruited right brain damaged patients. If the exact same correction factors and parameter combinations (FDR correction and dTLVC for lesion size control) were used, both univariate as well as multivariate approaches uncovered the same complex network pattern underlying spatial neglect. At the cortical level, lesion location dominantly affected the temporal cortex and its borders into inferior parietal and occipital cortices. Beyond, frontal and subcortical gray matter regions as well as white matter tracts connecting these regions were affected. Our findings underline the importance of a right network in spatial exploration and attention and specifically in the emergence of the core symptoms of spatial neglect.

Reprint of: Mapping human brain lesions and their functional consequences.

Neuroscience has a long history of inferring brain function by examining the relationship between brain injury and subsequent behavioral impairments. The primary advantage of this method over correlative methods is that it can tell us if a certain brain region is necessary for a given cognitive function. In addition, lesion-based analyses provide unique insights into clinical deficits. In the last decade, statistical voxel-based lesion behavior mapping (VLBM) emerged as a powerful method for understanding the architecture of the human brain. This review illustrates how VLBM improves our knowledge of functional brain architecture, as well as how it is inherently limited by its mass-univariate approach. A wide array of recently developed methods appear to supplement traditional VLBM. This paper provides an overview of these new methods, including the use of specialized imaging modalities, the combination of structural imaging with normative connectome data, as well as multivariate analyses of structural imaging data. We see these new methods as complementing rather than replacing traditional VLBM, providing synergistic tools to answer related questions. Finally, we discuss the potential for these methods to become established in cognitive neuroscience and in clinical applications.

Lead-DBS v2: Towards a comprehensive pipeline for deep brain stimulation imaging.

Deep brain stimulation (DBS) is a highly efficacious treatment option for movement disorders and a growing number of other indications are investigated in clinical trials. To ensure optimal treatment outcome, exact electrode placement is required. Moreover, to analyze the relationship between electrode location and clinical results, a precise reconstruction of electrode placement is required, posing specific challenges to the field of neuroimaging. Since 2014 the open source toolbox Lead-DBS is available, which aims at facilitating this process. The tool has since become a popular platform for DBS imaging. With support of a broad community of researchers worldwide, methods have been continuously updated and complemented by new tools for tasks such as multispectral nonlinear registration, structural/functional connectivity analyses, brain shift correction, reconstruction of microelectrode recordings and orientation detection of segmented DBS leads. The rapid development and emergence of these methods in DBS data analysis require us to revisit and revise the pipelines introduced in the original methods publication. Here we demonstrate the updated DBS and connectome pipelines of Lead-DBS using a single patient example with state-of-the-art high-field imaging as well as a retrospective cohort of patients scanned in a typical clinical setting at 1.5T. Imaging data of the 3T example patient is co-registered using five algorithms and nonlinearly warped into template space using ten approaches for comparative purposes. After reconstruction of DBS electrodes (which is possible using three methods and a specific refinement tool), the volume of tissue activated is calculated for two DBS settings using four distinct models and various parameters. Finally, four whole-brain tractography algorithms are applied to the patient's preoperative diffusion MRI data and structural as well as functional connectivity between the stimulation volume and other brain areas are estimated using a total of eight approaches and datasets. In addition, we demonstrate impact of selected preprocessing strategies on the retrospective sample of 51 PD patients. We compare the amount of variance in clinical improvement that can be explained by the computer model depending on the preprocessing method of choice. This work represents a multi-institutional collaborative effort to develop a comprehensive, open source pipeline for DBS imaging and connectomics, which has already empowered several studies, and may facilitate a variety of future studies in the field.

Removal of artifacts from resting-state fMRI data in stroke.

We examined the effect of lesion on the resting-state functional connectivity in chronic post-stroke patients. We found many instances of strong correlations in BOLD signal measured at different locations within the lesion, making it hard to distinguish from the connectivity between intact and strongly connected regions. Regression of the mean cerebro-spinal fluid signal did not alleviate this problem. The connectomes computed by exclusion of lesioned voxels were not good predictors of the behavioral measures. We came up with a novel method that utilizes Independent Component Analysis (as implemented in FSL MELODIC) to identify the sources of variance in the resting-state fMRI data that are driven by the lesion, and to remove this variance. The resulting functional connectomes show better correlations with the behavioral measures of speech and language, and improve the out-of-sample prediction accuracy of multivariate analysis. We therefore advocate this preprocessing method for studies of post-stroke functional connectivity, particularly in samples with large lesions.

Biased temporal order judgments in chronic neglect influenced by trunk position.

Numerous studies have reported that temporal order perception is biased in neurological patients with extinction and neglect. These individuals tend to perceive two objectively simultaneous stimuli as occurring asynchronously, with the ipsilesional item being perceived as appearing prior to the contralesional item. Likewise, they report that two stimuli occurred simultaneously in situations where the contralesional item is presented substantially prior to the ipsilesional item. Therefore, they exhibit a biased point of subjective simultaneity (PSS). Here we demonstrate that the magnitude of this effect is modulated by the relative position of the stimuli with respect to the patient's trunk. This effect was only observed in patients who still exhibited neglect symptoms, and neither the pathological bias nor substantial modulation were observed in individuals who had recovered from neglect, those who never had neglect or neurologically healthy controls. Crucially, our design kept the retinal and head-centered coordinates of these stimuli constant, providing a pure measure for the influence of egocentric trunk position. This finding emphasizes the interaction of egocentric spatial position on the temporal symptoms observed in these individuals.

Mapping human brain lesions and their functional consequences.

Neuroscience has a long history of inferring brain function by examining the relationship between brain injury and subsequent behavioral impairments. The primary advantage of this method over correlative methods is that it can tell us if a certain brain region is necessary for a given cognitive function. In addition, lesion-based analyses provide unique insights into clinical deficits. In the last decade, statistical voxel-based lesion behavior mapping (VLBM) emerged as a powerful method for understanding the architecture of the human brain. This review illustrates how VLBM improves our knowledge of functional brain architecture, as well as how it is inherently limited by its mass-univariate approach. A wide array of recently developed methods appear to supplement traditional VLBM. This paper provides an overview of these new methods, including the use of specialized imaging modalities, the combination of structural imaging with normative connectome data, as well as multivariate analyses of structural imaging data. We see these new methods as complementing rather than replacing traditional VLBM, providing synergistic tools to answer related questions. Finally, we discuss the potential for these methods to become established in cognitive neuroscience and in clinical applications.