BrainMap Community Portal: Meta-analytic Connectivity Modeling in an HPC Environment
Poster No:
1769
Submission Type:
Abstract Submission
Authors:
Peter Fox1, William Allen2, Mickle Fox3, Angela Uecker3, Michaela Robertson3, Jonathan Towne4, Mohamad Habes5, Simon Eickhoff6
Institutions:
1The University of Texas Health Science Center at San Antonio, San Antonio, TX, 2University of Texas, Austin, TX, 3University of Texas Health Science Center at San Antonio, San Antonio, TX, 4UT Health San Antonio, San Antonio, TX, 5University of Texas Health San Antonio, San Antonio, TX, 6Institute for Systems Neuroscience, Medical Faculty, Heinrich-Heine University Düsseldorf, Düsseldorf, North Rhine–Westphalia Land
First Author:
Co-Author(s):
Simon Eickhoff
Institute for Systems Neuroscience, Medical Faculty, Heinrich-Heine University Düsseldorf
Düsseldorf, North Rhine–Westphalia Land
Institute for Systems Neuroscience, Medical Faculty, Heinrich-Heine University Düsseldorf
Düsseldorf, North Rhine–Westphalia Land
Introduction:
Meta-analysis offers powerful approaches for synthesis of neuroimaging findings across laboratories, subject cohorts, and imaging modalities. Coordinate-Based Meta-Analysis (CBMA) computes between-study, effect-location replicability from atlas-referenced, 3-D coordinates. The preponderance of CBMA studies to date are mass univariate, computing 3-D meta-analytic maps. Multivariate CBMA methods, by contrast, identify the neural network architectures underlying task performance and brain disorders. Disorder-network modeling, in particular, is underutilized, as current evidence suggests that the vast majority of neurolologic, psychiatric and developmental disorders are network based. Cardinal limitations to wider adoption of multivariate CBMA ("meta-connectomics") for task and disorder network modeling are computational demand, pipeline availability, and data availability. The BrainMap Community Portal addresses these limitations by providing three online databases of coordinate-reporting task and disorder data sets in a high performance computing environment with multivariate and univariate CBMA applications.
Methods:
The BrainMap Community Portal applies the community-portal (science gateway) architecture (Lawrence et al. 2015). Data, pipelined applications and HPC access are provided in an integrated environment as a standalone deployment of the Texas Advanced Computing Center (TACC) "Core Experience Portal" codebase. Computation is provided by TACC resources: Stampede2, Lonestar6, Maverick2, Frontera and Longhorn. Data and meta-data are accessed via mirrored instances of three BrainMap Databases (DBs), implemented in Oracle®. These are: Task-Activation (TA DB; Fox et al. 2005), Voxel-Based Morphometry (VBM; DB Vanasse et al. 2018), and Voxel-Based Physiology (VBP DB).
Containerized implementations of BrainMap tools are provided in both graphical and command line formats. Sleuth 3.0.4 performs data retrieval filtered by the meta-data taxonomy to create an editable workspace. GingerALE 3.0.2 performs mass-univariate CBMA on the workspace using the latest implementation of the alteration-likelihood estimation (ALE) algorithm (Turkeltaub et al., 2002, 2012) with best-practices default settings (Eickhoff et al. 2016; Frahm L et al., 2023). Mango provides data visualization and regional interpretation of output by anatomy, function and disease.
For multivariate analysis, five algorithms are implemented as pipelined, script-controlled, command-line applications. Meta-analytic connectivity modeling (MACM; Robinson et al., 2010) is implemented de novo as mass-multivariate application of ALE, using a novel cooccurrence architecture. Connectivity-based parcellation (CBP; Bzdok et al. 2013) is implemented in an ALE-specific manner using shared code (Reuter et al., 2020) and the MACM co-occurrence architecture. ICA is implemented as a containerized instance of MELODIC (Beckmann et al. 2004) as adapted for CBMA (Smith et al., 2009). GTM (Cauda et al., 2018) and Author-topic modeling (Yeo et al. 2015; Ngo et al. 2019) are implemented in an ALE-specific format using code shared by the originators.
Containerized implementations of BrainMap tools are provided in both graphical and command line formats. Sleuth 3.0.4 performs data retrieval filtered by the meta-data taxonomy to create an editable workspace. GingerALE 3.0.2 performs mass-univariate CBMA on the workspace using the latest implementation of the alteration-likelihood estimation (ALE) algorithm (Turkeltaub et al., 2002, 2012) with best-practices default settings (Eickhoff et al. 2016; Frahm L et al., 2023). Mango provides data visualization and regional interpretation of output by anatomy, function and disease.
For multivariate analysis, five algorithms are implemented as pipelined, script-controlled, command-line applications. Meta-analytic connectivity modeling (MACM; Robinson et al., 2010) is implemented de novo as mass-multivariate application of ALE, using a novel cooccurrence architecture. Connectivity-based parcellation (CBP; Bzdok et al. 2013) is implemented in an ALE-specific manner using shared code (Reuter et al., 2020) and the MACM co-occurrence architecture. ICA is implemented as a containerized instance of MELODIC (Beckmann et al. 2004) as adapted for CBMA (Smith et al., 2009). GTM (Cauda et al., 2018) and Author-topic modeling (Yeo et al. 2015; Ngo et al. 2019) are implemented in an ALE-specific format using code shared by the originators.
Results:
In-house and beta testing of the BrainMap Community Portal confirms that the portal interface, the BrainMap database, CBMA applications are operational and ready for community access.
Conclusions:
The BrainMap Community Portal is ready for access at: portal.brainmap.org. We invite the community to explore this new resource and provide feedback Brainmap.org/forum). We encourage users to expand this resource by: 1) coding CBMA-compliant articles for entry; 2) sharing CBMA
workspaces and other work products; 3) implementing new CBMA pipelines within the community portal (Yeung et al., 2023).
workspaces and other work products; 3) implementing new CBMA pipelines within the community portal (Yeung et al., 2023).
Modeling and Analysis Methods:
fMRI Connectivity and Network Modeling 1
Image Registration and Computational Anatomy
Multivariate Approaches 2
Keywords:
Computational Neuroscience
Meta- Analysis
Statistical Methods
Workflows
1|2Indicates the priority used for review
Provide references using author date format
Imaging. IEEE Transaction on Medical Imaging Vol 23 (2) 137-152.
Bzdok D (2013) An investigation of the structural, connectional and functional subspecialization in the human
amygdala. Hum Brain Mapp 34(12_:3247-3266.
Eickhoff SB (2016) Behavior, sensitivity, and power of activation likelihood estimation characterized by massive
empirical simulation. Neuroimage 137:70-85.
Fox PT (2005) BrainMap Taxonomy of Experimental Design: Description and Evaluation. Hum Brain Mapp
25:185-198.
Fox PT (2014) Meta-Analysis in Human Neuroimaging: Computation Modeling of Large-Scale Databases.
Annual Review of Neurosciences 37:409-34.
Frahm L (in review) ALE meta-analysis of voxel-based morphometry Studies: Parameter validation via largescale
simulation.
Lawrence KA Science gateways today and tomorrow: positive perspective of nearly 5000 members of the
research community. Concurrency Computat: Pract. Exper. 2015: 27:4252-4268.
Ngo G (2019) Beyond consensus: Embracing heterogeneity in curated neuroimaging meta-analysis.
NeuroImage 300:142-158.
Reuter N (2020) CBPtools: a Python package for regional connectivity-based parcellation. Brain Structure &
Function 225:1261-1275.
Robinson JR (2010) Meta-analytic connectivity modeling: Delineating the Functional Connectivity of the Human
Amygdala. Hum Brain Mapp 31:173-184.
Smith SM (2009) The functional architecture of the human brain: Correspondence between resting FMRI and
task-activation studies. Proc Natl Acad Sci USA 106:13040-13045.
Towne J (in review) BrainMap VBP: a resource for meta-analysis of voxel-based physiological literature.
Turkeltaub P (2002) Meta-analysis of the functional neuroanatomy of single-word reading: Method and
Validation. NeuroImage 16: 765-780.
Vanasse T (2018) BrainMap VBM: An environment for structural meta-analysis. Human Brain Mapp 39(8) 3308-
3325.
Yeo BTT (2015) Functional specialization and flexibility in human association cortex. Cerebr Cortex 25:3654-
3672.
Yeung AWK et al. (2023) Trends in the sample size, statistics, and contributions to the BrainMap database of
activation likelihood estimation meta-analyses: An empirical study of 10-year data. Human Brain Mapp (in
press).