Improving the analysis, storage and sharing of neuroimaging data using relational databases and distributed computing - PubMed (original) (raw)

Improving the analysis, storage and sharing of neuroimaging data using relational databases and distributed computing

Uri Hasson et al. Neuroimage. 2008.

Abstract

The increasingly complex research questions addressed by neuroimaging research impose substantial demands on computational infrastructures. These infrastructures need to support management of massive amounts of data in a way that affords rapid and precise data analysis, to allow collaborative research, and to achieve these aims securely and with minimum management overhead. Here we present an approach that overcomes many current limitations in data analysis and data sharing. This approach is based on open source database management systems that support complex data queries as an integral part of data analysis, flexible data sharing, and parallel and distributed data processing using cluster computing and Grid computing resources. We assess the strengths of these approaches as compared to current frameworks based on storage of binary or text files. We then describe in detail the implementation of such a system and provide a concrete description of how it was used to enable a complex analysis of fMRI time series data.

PubMed Disclaimer

Figures

Figure 1

Figure 1. Sharing and analyzing data using databases

fMRI data collected at one center (the Data Source) is stored on a Master database, and is replicated to a collaborator, as well as to a 100-node computing cluster. Collaborators can either analyze the data locally, or query data from the master database. The computing cluster holds two copies of the data using two separate DBMS servers, to serve 100 clients simultaneously.

Figure 2

Figure 2. Example of database scheme for storing data from an fMRI experiment

Each titled table reflects a table in the database and the information it maintains. Separate tables store the time series data and signal estimates (green). The database returns the data of voxels satisfying a certain criteria. If no criteria are specified, the data for all voxels is returned. Criteria are specified as constraints based on the filter tables (orange). Some filters are linked to individual participants (single-participant filters) whereas others are linked to the entire group of participants in the experiment (group-level filters).

Figure 3

Figure 3. Results of a reverse correlation analysis performed using a database and Grid computing

Orange lines are the hemodynamic response in the ventral premotor cortex during (A) the gesture condition and (B) the self-adaptor gesture condition, in which gestures were uninformative with respect to story content. Grey lines are the gamma functions fit to each maxima in the response. These were used to objectively determine which stimulus aspects produce maxima and minima (see text). Blue arrowed lines point to maxima while black arrowed lines point to minima. Meaningful gestures were far more likely to occur at maxima in the response than in minima, whereas non-meaningful self-adapting hand movements are as likely to occur at maxima as minima.

Figure 4

Figure 4. Example of database schema for storing time series data from an fMRI experiment

The database schema affords selecting the time series of any given set of voxels on the basis of the voxel’s estimated signal intensity or anatomical location. In addition, for each voxel it is possible to select either the entire time series, or just those time points in the series where specific experimental condition or conditions occurred.

References

    1. Argall BD, Saad ZS, Beauchamp MS. Simplified intersubject averaging on the cortical surface using SUMA. Human Brain Mapping. 2005;27:14–27. -PMC -PubMed
    1. Ascoli GA. Mobilizing the base of neuroscience data: the case of neuronal morphologies. Nature Reviews Neuroscience. 2006;7:318–324. -PubMed
    1. Ashburner J, Friston KJ. Voxel-based morphometry -- the methods. NeuroImage. 2000;11:805–821. -PubMed
    1. Bowden DM, Martin RF. NeuroNames brain hierarchy. NeuroImage. 1995;2(1):63–83. -PubMed
    1. Buchsbaum BR, Olsen RK, Koch PF, Kohn P, Kippenhan JS, Berman KF. Reading, hearing, and the planum temporale. NeuroImage. 2005;24:444–454. -PubMed

Publication types

MeSH terms

LinkOut - more resources