BrainMap VBM: An environment for structural meta-analysis.

The BrainMap database is a community resource that curates peer-reviewed, coordinate-based human neuroimaging literature. By pairing the results of neuroimaging studies with their relevant meta-data, BrainMap facilitates coordinate-based meta-analysis (CBMA) of the neuroimaging literature en masse or at the level of experimental paradigm, clinical disease, or anatomic location. Initially dedicated to the functional, task-activation literature, BrainMap is now expanding to include voxel-based morphometry (VBM) studies in a separate sector, titled: BrainMap VBM. VBM is a whole-brain, voxel-wise method that measures significant structural differences between or within groups which are reported as standardized, peak x-y-z coordinates. Here we describe BrainMap VBM, including the meta-data structure, current data volume, and automated reverse inference functions (region-to-disease profile) of this new community resource. CBMA offers a robust methodology for retaining true-positive and excluding false-positive findings across studies in the VBM literature. As with BrainMap's functional database, BrainMap VBM may be synthesized en masse or at the level of clinical disease or anatomic location. As a use-case scenario for BrainMap VBM, we illustrate a trans-diagnostic data-mining procedure wherein we explore the underlying network structure of 2,002 experiments representing over 53,000 subjects through independent components analysis (ICA). To reduce data-redundancy effects inherent to any database, we demonstrate two data-filtering approaches that proved helpful to ICA. Finally, we apply hierarchical clustering analysis (HCA) to measure network- and disease-specificity. This procedure distinguished psychiatric from neurological diseases. We invite the neuroscientific community to further exploit BrainMap VBM with other modeling approaches.

Implementation errors in the GingerALE Software: Description and recommendations.

Neuroscience imaging is a burgeoning, highly sophisticated field the growth of which has been fostered by grant-funded, freely distributed software libraries that perform voxel-wise analyses in anatomically standardized three-dimensional space on multi-subject, whole-brain, primary datasets. Despite the ongoing advances made using these non-commercial computational tools, the replicability of individual studies is an acknowledged limitation. Coordinate-based meta-analysis offers a practical solution to this limitation and, consequently, plays an important role in filtering and consolidating the enormous corpus of functional and structural neuroimaging results reported in the peer-reviewed literature. In both primary data and meta-analytic neuroimaging analyses, correction for multiple comparisons is a complex but critical step for ensuring statistical rigor. Reports of errors in multiple-comparison corrections in primary-data analyses have recently appeared. Here, we report two such errors in GingerALE, a widely used, US National Institutes of Health (NIH)-funded, freely distributed software package for coordinate-based meta-analysis. These errors have given rise to published reports with more liberal statistical inferences than were specified by the authors. The intent of this technical report is threefold. First, we inform authors who used GingerALE of these errors so that they can take appropriate actions including re-analyses and corrective publications. Second, we seek to exemplify and promote an open approach to error management. Third, we discuss the implications of these and similar errors in a scientific environment dependent on third-party software. Hum Brain Mapp 38:7-11, 2017. © 2016 Wiley Periodicals, Inc.

Meta-analytic connectivity and behavioral parcellation of the human cerebellum.

The cerebellum historically has been thought to mediate motor and sensory signals between the body and cerebral cortex, yet cerebellar lesions are also associated with altered cognitive behavioral performance. Neuroimaging evidence indicates that the cerebellum contributes to a wide range of cognitive, perceptual, and motor functions. Here, we used the BrainMap database to investigate whole-brainco-activation patterns between cerebellar structures and regions of the cerebral cortex, as well as associations with behavioral tasks. Hierarchical clustering was performed to meta-analytically identify cerebellar structures with similar cortical co-activation, and independently, with similar correlations to specific behavioral tasks. Strong correspondences were observed in these separate but parallel analyses of meta-analytic connectivity and behavioral metadata. We recovered differential zones of cerebellar co-activation that are reflected across the literature. Furthermore, the behaviors and tasks associated with the different cerebellar zones provide insight into the specialized function of the cerebellum, relating to high-order cognition, emotion, perception, interoception, and action. Taken together, these task-basedmeta-analytic results implicate distinct zones of the cerebellum as critically involved in the monitoring and mediation of psychological responses to internal and external stimuli.

Neural architecture underlying classification of face perception paradigms.

We present a novel strategy for deriving a classification system of functional neuroimaging paradigms that relies on hierarchical clustering of experiments archived in the BrainMap database. The goal of our proof-of-concept application was to examine the underlying neural architecture of the face perception literature from a meta-analytic perspective, as these studies include a wide range of tasks. Task-based results exhibiting similar activation patterns were grouped as similar, while tasks activating different brain networks were classified as functionally distinct. We identified four sub-classes of face tasks: (1) Visuospatial Attention and Visuomotor Coordination to Faces, (2) Perception and Recognition of Faces, (3) Social Processing and Episodic Recall of Faces, and (4) Face Naming and Lexical Retrieval. Interpretation of these sub-classes supports an extension of a well-known model of face perception to include a core system for visual analysis and extended systems for personal information, emotion, and salience processing. Overall, these results demonstrate that a large-scale data mining approach can inform the evolution of theoretical cognitive models by probing the range of behavioral manipulations across experimental tasks.

Meta-analytic connectivity modelling of functional magnetic resonance imaging studies in autism spectrum disorders.

Social and non-social deficits in autism spectrum disorders (ASD) persist into adulthood and may share common regions of aberrant neural activations. The current meta-analysis investigated activation differences between ASD and neurotypical controls irrespective of task type. Activation likelihood estimation meta-analyses were performed to examine consistent hypo-activated and/or hyper-activated regions for all tasks combined, and for social and non-social tasks separately; meta-analytic connectivity modelling and behavioral/paradigm analyses were performed to examine co-activated regions and associated behaviors. One hundred studies (mean age range = 18-41 years) were included. For all tasks combined, the ASD group showed significant (p < .05) hypo-activation in one cluster around the left amygdala (peak - 26, -2, -20, volume = 1336 mm3, maximum ALE = 0.0327), and this cluster co-activated with two other clusters around the right cerebellum (peak 42, -56, -22, volume = 2560mm3, maximum ALE = 0.049) Lobule VI/Crus I and the left fusiform gyrus (BA47) (peak - 42, -46, -18, volume = 1616 mm3, maximum ALE = 0.046) and left cerebellum (peak - 42, -58, -20, volume = 1616mm3, maximum ALE = 0.033) Lobule VI/Crus I. While the left amygdala was associated with negative emotion (fear) (z = 3.047), the left fusiform gyrus/cerebellum Lobule VI/Crus I cluster was associated with language semantics (z = 3.724) and action observation (z = 3.077). These findings highlight the left amygdala as a region consistently hypo-activated in ASD and suggest the potential involvement of fusiform gyrus and cerebellum in social cognition in ASD. Future research should further elucidate if and how amygdala-fusiform/cerebellar connectivity relates to social and non-social cognition in adults with ASD.

On the spatial organization of sound processing in the human temporal lobe: a meta-analysis.

In analogy to visual object recognition, proposals have been made that auditory object recognition is organized by sound class (e.g., vocal/non-vocal, linguistic/non-linguistic) and linked to several pathways or processing streams with specific functions. To test these proposals, we analyzed temporal lobe activations from 297 neuroimaging studies on vocal, musical and environmental sound processing. We found that all sound classes elicited activations anteriorly, posteriorly and ventrally of primary auditory cortex. However, rather than being sound class (e.g., voice) or attribute (e.g., complexity) specific, these processing streams correlated with sound knowledge or experience. Specifically, an anterior stream seemed to support general, sound class independent sound recognition and discourse-level semantic processing. A posterior stream could be best explained as supporting the embodiment of sound associated actions and a ventral stream as supporting multimodal conceptual representations. Vocalizations and music engaged these streams evenly in the left and right hemispheres, whereas environmental sounds produced a left-lateralized pattern. Together, these results both challenge and confirm existing proposal of temporal lobe specialization. Moreover, they suggest that the temporal lobe maintains the neuroanatomical building blocks for an all-purpose sound comprehension system that, instead of being preset for a particular sound class, is shaped in interaction with an individual's sonic environment.

The functional connectivity of the human caudate: an application of meta-analytic connectivity modeling with behavioral filtering.

Meta-analysis based techniques are emerging as powerful, robust tools for developing models of connectivity in functional neuroimaging. Here, we apply meta-analytic connectivity modeling to the human caudate to 1) develop a model of functional connectivity, 2) determine if meta-analytic methods are sufficiently sensitive to detect behavioral domain specificity within region-specific functional connectivity networks, and 3) compare meta-analytic driven segmentation to structural connectivity parcellation using diffusion tensor imaging. Results demonstrate strong coherence between meta-analytic and data-driven methods. Specifically, we found that behavioral filtering resulted in cognition and emotion related structures and networks primarily localized to the head of the caudate nucleus, while perceptual and action specific regions localized to the body of the caudate, consistent with early models of nonhuman primate histological studies and postmortem studies in humans. Diffusion tensor imaging (DTI) revealed support for meta-analytic connectivity modeling's (MACM) utility in identifying both direct and indirect connectivity. Our results provide further validation of meta-analytic connectivity modeling, while also highlighting an additional potential, namely the extraction of behavioral domain specific functional connectivity.

Correspondence of the brain's functional architecture during activation and rest.

Neural connections, providing the substrate for functional networks, exist whether or not they are functionally active at any given moment. However, it is not known to what extent brain regions are continuously interacting when the brain is "at rest." In this work, we identify the major explicit activation networks by carrying out an image-based activation network analysis of thousands of separate activation maps derived from the BrainMap database of functional imaging studies, involving nearly 30,000 human subjects. Independently, we extract the major covarying networks in the resting brain, as imaged with functional magnetic resonance imaging in 36 subjects at rest. The sets of major brain networks, and their decompositions into subnetworks, show close correspondence between the independent analyses of resting and activation brain dynamics. We conclude that the full repertoire of functional networks utilized by the brain in action is continuously and dynamically "active" even when at "rest."

Behavioral interpretations of intrinsic connectivity networks.

An increasingly large number of neuroimaging studies have investigated functionally connected networks during rest, providing insight into human brain architecture. Assessment of the functional qualities of resting state networks has been limited by the task-independent state, which results in an inability to relate these networks to specific mental functions. However, it was recently demonstrated that similar brain networks can be extracted from resting state data and data extracted from thousands of task-based neuroimaging experiments archived in the BrainMap database. Here, we present a full functional explication of these intrinsic connectivity networks at a standard low order decomposition using a neuroinformatics approach based on the BrainMap behavioral taxonomy as well as a stratified, data-driven ordering of cognitive processes. Our results serve as a resource for functional interpretations of brain networks in resting state studies and future investigations into mental operations and the tasks that drive them.

Brain pathology recapitulates physiology: A network meta-analysis.

Network architecture is a brain-organizational motif present across spatial scales from cell assemblies to distributed systems. Structural pathology in some neurodegenerative disorders selectively afflicts a subset of functional networks, motivating the network degeneration hypothesis (NDH). Recent evidence suggests that structural pathology recapitulating physiology may be a general property of neuropsychiatric disorders. To test this possibility, we compared functional and structural network meta-analyses drawing upon the BrainMap database. The functional meta-analysis included results from >7,000 experiments of subjects performing >100 task paradigms; the structural meta-analysis included >2,000 experiments of patients with >40 brain disorders. Structure-function network concordance was high: 68% of networks matched (pFWE < 0.01), confirming the broader scope of NDH. This correspondence persisted across higher model orders. A positive linear association between disease and behavioral entropy (p = 0.0006;R2 = 0.53) suggests nodal stress as a common mechanism. Corroborating this interpretation with independent data, we show that metabolic 'cost' significantly differs along this transdiagnostic/multimodal gradient.

Convergence Analysis of Micro-Lesions (CAML): An approach to mapping of diffuse lesions from carotid revascularization.

Carotid revascularization (endarterectomy, stenting) prevents stroke; however, procedure-related embolization is common and results in small brain lesions easily identified by diffusion weighted magnetic resonance imaging (DWI). A crucial barrier to understanding the clinical significance of these lesions has been the lack of a statistical approach to identify vulnerable brain areas. The problem is that the lesions are small, numerous, and non-overlapping. Here we address this problem with a new method, the Convergence Analysis of Micro-Lesions (CAML) technique, an extension of the Anatomic Likelihood Analysis (ALE). The method combines manual lesion tracing, constraints based on known lesion patterns, and convergence analysis to represent regions vulnerable to lesions as probabilistic brain atlases. Two studies were conducted over the course of 12 years in an active, vascular surgery clinic. An analysis in an initial group of 126 patients at 1.5 T MRI was cross-validated in a second group of 80 patients at 3T MRI. In CAML, lesions were manually defined and center points identified. Brains were aligned according to side of surgery since this factor powerfully determines lesion distribution. A convergence based analysis, was performed on each of these groups. Results indicated the most consistent region of vulnerability was in motor and premotor cortex regions. Smaller regions common to both groups included the dorsolateral prefrontal cortex and medial parietal regions. Vulnerability of motor cortex is consistent with previous work showing changes in hand dexterity associated with these procedures. The consistency of CAML also demonstrates the feasibility of this new approach to characterize small, diffuse, non-overlapping lesions in patients with multifocal pathologies.

Automated regional behavioral analysis for human brain images.

Behavioral categories of functional imaging experiments along with standardized brain coordinates of associated activations were used to develop a method to automate regional behavioral analysis of human brain images. Behavioral and coordinate data were taken from the BrainMap database (http://www.brainmap.org/), which documents over 20 years of published functional brain imaging studies. A brain region of interest (ROI) for behavioral analysis can be defined in functional images, anatomical images or brain atlases, if images are spatially normalized to MNI or Talairach standards. Results of behavioral analysis are presented for each of BrainMap's 51 behavioral sub-domains spanning five behavioral domains (Action, Cognition, Emotion, Interoception, and Perception). For each behavioral sub-domain the fraction of coordinates falling within the ROI was computed and compared with the fraction expected if coordinates for the behavior were not clustered, i.e., uniformly distributed. When the difference between these fractions is large behavioral association is indicated. A z-score ≥ 3.0 was used to designate statistically significant behavioral association. The left-right symmetry of ~100K activation foci was evaluated by hemisphere, lobe, and by behavioral sub-domain. Results highlighted the classic left-side dominance for language while asymmetry for most sub-domains (~75%) was not statistically significant. Use scenarios were presented for anatomical ROIs from the Harvard-Oxford cortical (HOC) brain atlas, functional ROIs from statistical parametric maps in a TMS-PET study, a task-based fMRI study, and ROIs from the ten "major representative" functional networks in a previously published resting state fMRI study. Statistically significant behavioral findings for these use scenarios were consistent with published behaviors for associated anatomical and functional regions.

Thalamic medial dorsal nucleus atrophy in medial temporal lobe epilepsy: A VBM meta-analysis.

PURPOSE: Medial temporal lobe epilepsy (MTLE) is associated with MTLE network pathology within and beyond the hippocampus. The purpose of this meta-analysis was to identify consistent MTLE structural change to guide subsequent targeted analyses of these areas. METHODS: We performed an anatomic likelihood estimation (ALE) meta-analysis of 22 whole-brain voxel-based morphometry experiments from 11 published studies. We grouped these experiments in three ways. We then constructed a meta-analytic connectivity model (MACM) for regions of consistent MTLE structural change as reported by the ALE analysis. KEY FINDINGS: ALE reported spatially consistent structural change across VBM studies only in the epileptogenic hippocampus and the bilateral thalamus; within the thalamus, the medial dorsal nucleus of the thalamus (MDN thalamus) represented the greatest convergence (P < 0.05 corrected for multiple comparisons). The subsequent MACM for the hippocampus and ipsilateral MDN thalamus demonstrated that the hippocampus and ipsilateral MDN thalamus functionally co-activate and are nodes within the same network, suggesting that MDN thalamic damage could result from MTLE network excitotoxicity. SIGNIFICANCE: Notwithstanding our large sample of studies, these findings are more restrictive than previous reports and demonstrate the utility of our inclusion filters and of recently modified meta-analytic methods in approximating clinical relevance. Thalamic pathology is commonly observed in animal and human studies, suggesting it could be a clinically useful indicator. Thalamus-specific research as a clinical marker awaits further investigation.

The BrainMap strategy for standardization, sharing, and meta-analysis of neuroimaging data.

BACKGROUND: Neuroimaging researchers have developed rigorous community data and metadata standards that encourage meta-analysis as a method for establishing robust and meaningful convergence of knowledge of human brain structure and function. Capitalizing on these standards, the BrainMap project offers databases, software applications, and other associated tools for supporting and promoting quantitative coordinate-based meta-analysis of the structural and functional neuroimaging literature. FINDINGS: In this report, we describe recent technical updates to the project and provide an educational description for performing meta-analyses in the BrainMap environment. CONCLUSIONS: The BrainMap project will continue to evolve in response to the meta-analytic needs of biomedical researchers in the structural and functional neuroimaging communities. Future work on the BrainMap project regarding software and hardware advances are also discussed.

Automated analysis of meta-analysis networks.

The high information content in large data sets from voxel-based meta-analyses is complex, making it hard to readily resolve details. Using the meta-analysis network as a standardized data structure, network analysis algorithms can examine complex interrelationships and resolve hidden details. Two new network analysis algorithms have been adapted for use with meta-analysis networks. The first, called replicator dynamics network analysis (RDNA), analyzes co-occurrence of activations, whereas the second, called fractional similarity network analysis (FSNA), uses binary pattern matching to form similarity subnets. These two network analysis methods were evaluated using data from activation likelihood estimation (ALE)-based meta-analysis of the Stroop paradigm. Two versions of these data were evaluated, one using a more strict ALE threshold (P < 0.01) with a 13-node meta-analysis network, and the other a more lax threshold (P < 0.05) with a 22-node network. Java-based applications were developed for both RDNA and FSNA. The RDNA algorithm was modified to provide multiple subnets or maximal cliques for meta-analysis networks. Three different similarity measures were evaluated with FSNA to form subsets of nodes and experiments. RDNA provides a means to gauge importance of metanalysis subnets and complements FSNA, which provides a more comprehensive assessment of node similarity subsets, experiment similarity subsets, and overall node-to-factors similarity. The need to use both presence and absence of activations was an important finding in similarity analyses. FSNA revealed details from the pooled Stroop meta-analysis that would otherwise require separate highly filtered meta-analyses. These new analysis tools demonstrate how network analysis strategies can simplify greatly and enhance voxel-based meta-analyses.

ALE meta-analysis: controlling the false discovery rate and performing statistical contrasts.

Activation likelihood estimation (ALE) has greatly advanced voxel-based meta-analysis research in the field of functional neuroimaging. We present two improvements to the ALE method. First, we evaluate the feasibility of two techniques for correcting for multiple comparisons: the single threshold test and a procedure that controls the false discovery rate (FDR). To test these techniques, foci from four different topics within the literature were analyzed: overt speech in stuttering subjects, the color-word Stroop task, picture-naming tasks, and painful stimulation. In addition, the performance of each thresholding method was tested on randomly generated foci. We found that the FDR method more effectively controls the rate of false positives in meta-analyses of small or large numbers of foci. Second, we propose a technique for making statistical comparisons of ALE meta-analyses and investigate its efficacy on different groups of foci divided by task or response type and random groups of similarly obtained foci. We then give an example of how comparisons of this sort may lead to advanced designs in future meta-analytic research.

BrainMap taxonomy of experimental design: description and evaluation.

Coordinate-based, voxel-wise meta-analysis is an exciting recent addition to the human functional brain mapping literature. In view of the critical importance of selection criteria for any valid meta-analysis, a taxonomy of experimental design should be an important tool for aiding in the design of rigorous meta-analyses. The coding scheme of experimental designs developed for and implemented within the BrainMap database provides a candidate taxonomy. In this study, the BrainMap experimental-design taxonomy is described and evaluated by comparing taxonomy fields to data-filtering choices made by subject-matter experts carrying out meta-analyses of the functional imaging literature. Fifteen publications reporting a total of 46 voxel-wise meta-analyses were included in this assessment. Collectively these 46 meta-analyses pooled data from 351 publications, selected for experimental similarity within each meta-analysis. Filter implementations within BrainMap were graded by ease-of-use (A-C) and by stage-of-use (1-3). Quality filters and content filters were tabulated separately. Quality filters required for data entry into BrainMap were classed as mandatory (five filters), being above the use grading system. All authors spontaneously adopted the five mandatory filters in constructing their meta-analysis, indicating excellent agreement on data quality among authors and between authors and the BrainMap development team. Two non-mandatory quality filters (group size and imaging modality) were applied by all authors; both were Stage 1, Grade A filters. Field-of-view filters were the least-accessible quality filters (Stage 3, Grade C); two field-of-view filters were applied by six and four authors, respectively. Authors made a total of 115 content-filter choices. Of these, 78 (68%) were Stage 1, Grade A filters; 16 (14%) were Stage 2, Grade A; and 21 (18%) were Stage 2, Grade C. No author-applied filter was absent from the taxonomy.