Meta-analysis in human brain mapping

点击翻译按钮获取中文摘要

This is the competing renewal of R01MH074457-13, which sustains the BrainMap Project (www.brainmap.org). The overall goal of the BrainMap Project is to provide the human neuroimaging community with curated data sets, metadata, computational tools, and related resources that enable coordinate-based meta-analyses (CBMA), meta-analytic connectivity modeling (MACM), meta-data informed interpretation (“decoding”) of imaging results, and meta-analytic priors for mining (including machine learning) primary (per-subject) neuroimaging data. To date, the BrainMap Project has designed and populated two coordinate-based databases: 1) a task-activation repository (TA DB); and, 2) a voxel-based morphometry repository (VBM DB). The TA DB contains >17,200 experiments, collectively representing > 78,000 subjects and > 110 task- activation paradigms. The VBM DB contains > 3,100 experiments, collectively representing > 81,000 subjects with > 80 psychiatric, neurologic and developmental disorders with ICD-10 coding. The BrainMap Project has created, optimized and validated an integrated pipeline of multi-platform (Javascript), open-access tools to curate (Scribe), filter and retrieve (Sleuth), analyze (GingerALE), visualize (Mango) and interpret analysis output (BrainMap meta-data plugins for Mango). Several network-modeling approaches have been applied to BrainMap data -- MACM, independent components analysis (ICA), graph theory modeling (GTM), author-topic modeling (ATM), structural equation modeling (SEM), and connectivity-based parcellation (CBP) – but none are yet pipeline components. Utilization of these CBMA resources is substantial: BrainMap software, data and meta-data have been used in > 825 peer-reviewed publications. Of these, > 350 were published within the current funding period (April 2015-March 2019; brainmap.org/pubs). In this competing renewal, four tool- development aims are proposed, each of which extends this high-impact research resource. Aim 1. Database Expansion. BrainMap data repositories will be expanded. Aim 2. Meta-analytic Network Modeling. Network modeling will be added to the BrainMap pipeline. Aim 3. Large-Scale Simulations, Comparisons and Validations. Data simulations, characterizations and validations will be performed. Aim 4. Meta-data Inferential tools. Tools for mining BrainMap’s location-linked meta-data will be expanded. Data Sharing Plan. BrainMap data, meta-data, pipeline tools, and templates created by whole-database modeling (e.g., ICA and ATM network masks) are shared at BrainMap.org. Of all new data entries, more than half are contributed by BrainMap users, i.e., community data sharing via BrainMap.org. For community-coded entries, the BrainMap team provides curation and quality control. Comprehensive database images (database dumps) are available to tool developers through Collaborative Use Agreements.