COORDINATE-BASED META-ANALYTIC SEARCH

FOR THE SPM NEUROIMAGING PIPELINE

The BredeQuery Plugin for SPM5

Bartłomiej Wilkowski, Marcin Szewczyk, Peter Mondrup Rasmussen

Lars Kai Hansen and Finn

Arup Nielsen

Informatics and Mathematical Modelling, Technical University of Denmark, Kongens Lyngby, Denmark

Keywords:

SPM, fMRI, PET, Neuroinformatics, Talairach, MNI, Brain region, Brain function, Brain activity, Article

retrieval, Coordinate-based search, Bibtex, Brede Database, Ontology.

Abstract:

Large amounts of neuroimaging studies are collected and have changed our view on human brain function.

By integrating multiple studies in meta-analysis a more complete picture is emerging. Brain locations are

usually reported as coordinates with reference to a speciﬁc brain atlas, thus some of the databases offer so-

called coordinate-based searching to the users (e.g. Brede, BrainMap). For such search, the publications,

which relate to the brain locations represented by the user coordinates, are retrieved. In this paper we present

BredeQuery – a plugin for the widely used SPM5 data analytic pipeline. BredeQuery offers a direct link from

SPM5 to the Brede Database coordinate-based search engine. BredeQuery is able to ‘grab’ brain location

coordinates from the SPM windows and enter them as a query for the Brede Database. Moreover, results of

the query can be displayed in an SPM window and/or exported directly to some popular bibliographic ﬁle

formats (BibTeX, Reference Manager, etc).

1 INTRODUCTION

The growing number of functional neuroimaging

studies of increasingly sophisticated human brain ac-

tivity brings the demand for new tools/services for

integration of research ﬁndings, wider exchange of

information between laboratories from the same re-

search area and efﬁcient searching of related articles,

reviews and other literature (Wager et al., 2007).

The dominant paradigm in current neuroimaging

is that of functional localization. Functional locali-

zation hypothesizes that a given human behavior is

established by a change in brain activity in a rela-

tively limited number of spatially segregated proce-

ssing units. Thus the result of an experiment under

this paradigm consists of a Statistical Parametric Map

(SPM) indicating the local involvement. Often the

SPM is summarized as a list of regions, see e.g., (Fris-

ton et al., 2007; Pekar, 2006), in which the SPM has

been judged to be signiﬁcantly different from zero

(regions were the null hypothesis is rejected). As

the typical neuroimaging experiment investigates a

highly controlled behavior and often involves a rela-

tively limited number of subjects, there is strong need

for tools to integrate multiple experiments in order to

increase the robustness to the experiment speciﬁc im-

plementation of the given behavior and to statistical

ﬂuctuation due to limited sample sizes.

Several methods have been proposed for neu-

roimaging meta-analysis and for estimation of associ-

ations between the brain locations and textual repre-

sentations of behavior, for a recent review, see e.g.,

(Wager et al., 2007). A set of methods are based on

the so-called Brede Database (Nielsen, 2003). Me-

thods for integration include estimation of conditional

probability density functions representing the loca-

lized probability of activation in response to a given

behavior ‘word’ (Nielsen and Hansen, 2002; Nielsen

and Hansen, 2004) and multivariate methods based

on non-negative matrix factorization that aim to rep-

resent global dependencies between brain activation

and semantic text labels from neuroscience publica-

tions (Nielsen et al., 2004).

Brain locations are reported as region coordinates

relative to a speciﬁc brain atlas (usually MNI or Ta-

lairach spaces), hence, there is an interest for effec-

Wilkowski B., Szewczyk M., Rasmussen P., Hansen L. and Nielsen F. (2009).

COORDINATE-BASED META-ANALYTIC SEARCH FOR THE SPM NEUROIMAGING PIPELINE - The BredeQuery Plugin for SPM5.

In Proceedings of the International Conference on Health Informatics, pages 11-17

DOI: 10.5220/0001510000110017

 SciTePress

Figure 1: Graphical user interface of the BredeQuery plugin for SPM5. Firstly, the user can choose if the coordinates used

for querying will be grabbed from an SPM’s results window or will be typed manually. The grabbed (retrieved) coordinates

are shown on the list. The user can switch on an interactive mode – the coordinate selected in the SPM window will be

automatically selected in the plugin on the coordinates list. Moreover, the coordinates are grabbed using the chosen MNI to

Talairach transformation (Brett or Lancaster MTT afﬁne transformations). Afterwards, the user is able to display the query

results in the Matlab web browser or to import them into the speciﬁed bibliographic format.

tive search for experiments, hence, scientiﬁc papers,

which report similar coordinate sets in brain. Brain-

Map (Laird et al., 2005) and Brede (Nielsen, 2003)

are the databases which offer the coordinate-based

searching. For Brede it is available on both the web-

page and in a standalone application.

In order to enable a neuroimaging scientists to per-

form meta-analysis in the context of a speciﬁc on-

going study we here propose a tool that integrates

retrieval of related research within the data analysis

pipeline. The dominant tool for human brain map-

ping is undisputable the SPM set of tools developed

and distributed by the Functional Imaging Laboratory

(London, (Friston et al., 2007)). For an analysis of the

usage of imaging pipelines see e.g., (Nielsen et al.,

2006). Thus we have initiated the development of a

plugin for SPM5, which offers high integration with

the Brede Database.

The BredeQuery plugin (see Figure 1) provides

the opportunity to perform coordinate-based query

and retrieval of the related articles references directly

from the SPM (Matlab) environment.

2 BREDE DATABASE

The Brede Database available through the web-

page: http://hendrix.imm.dtu.dk/services/jerne/brede/

records published neuroimaging experiments that list

stereotaxic coordinates in so-called MNI or Talairach

space (Talairach and Tournoux, 1988). Presently,

close to 4000 coordinates from 186 papers with a total

of 586 experiments are available.

HEALTHINF 2009 - International Conference on Health Informatics

Figure 2: Screenshot from one of the pages in the Brede

Database showing coordinates in Talairach space. This is

one of presently 586 experiments recorded in the database

– an fMRI experiment resulting in 29 reported coordinates.

The data is stored in XML ﬁles, and Matlab func-

tions generate static webpages with visualization of

the entries in the database, see Figure 2. Web-based

searching is possible from the homepage, but up till

now it has required that the researcher manually typed

in the query or extracted results from the image ana-

lysis program. Webpages are also available with the

Brede Database ontologies, which exist for brain re-

gions and brain functions, see Figure 3.

3 RELATED TOOLS

There a a few available tools that have aims related to

those of the BredeQuery plugin.

The AMAT SPM toolbox is also meta-analysis

toolbox (Hamilton, 2005) for Matlab based meta-

analysis of fMRI data. It provides coordinate-based

search for over 5000 coordinates from 213 published

papers of which some were derived from the Brede

Database. The coordinates are in MNI or Talairach

space. The toolbox can locate neighboring coordi-

nates to a given coordinate, as well as publications

for a given author or year. The tool was last updated

in 2005.

Another related toolbox, is xjView (Cui and

Li, 2007), which offers the SPM user, apart

from viewing the images in glass view, sec-

tion view or 3D render view, search of selected

brain regions in databases in order to eluci-

date their function. It searches among others

in Google Scholar (http://scholar.google.com/),

PubMed (http://www.ncbi.nlm.nih.gov/pubmed/)

and the xBrain database, available on the webpage:

http://sig.biostr.washington.edu/projects/xbrain/.

The XCEDE SPM Toolbox (Keator et al., 2006)

is also a toolbox distributed within the SPM com-

munity. This is a toolbox for SPM99 and SPM2,

which enables the users to capture activation data

for PET/fMRI analysis and save them to the XML

ﬁle in a XCEDE XML schema. Moreover, it is

extending the exported XML ﬁle by adding auto-

matically the anatomical labeling of the region in

the brain for the given activity coordinates. It is

achieved through two SPM toolboxes: Talairach Dae-

mon available on http://www.talairach.org/ and Au-

tomated Anatomical Labeling (AAL) available on

http://www.cyceron.fr/freeware/.

4 SOFTWARE DESCRIPTION

The recent version of the BredeQuery plugin, together

with the User’s Guide, can be downloaded from the

webpage: http://neuroinf.imm.dtu.dk/BredeQuery/. A

graphical user interface of the BredeQuery plugin is

divided into ﬁve areas where different user-actions

can be performed. Firstly, the activity coordinates

can be ‘grabbed’ from the SPM results ﬁgure into

the plugin. Since the coordinates can be presented

in MNI or Talairach spaces, some transformations are

introduced for interoperability. The coordinate-based

search in the Brede Database is based on the Talairach

space coordinates, thus the BredeQuery plugin of-

fers two MNI to Talairach transformations, which can

be chosen by the user. The piece-wise afﬁne trans-

formation proposed by Matthew Brett is one of the

available transformation (Brett, 1999). Also included

is the afﬁne transformation MNI-to-Talairach (MTT),

suggested by Jack Lancaster et al. (Lancaster et al.,

2007). Three separate transformation were suggested

by the group: one for SPM, one for FSL and a com-

bined ’pooled’ transformation. The MTT

SPM

trans-

formation is set as default in the BredeQuery plugin.

When the coordinates have been ‘grabbed’ and

shown in the BredeQuery plugin, the coordinate-

based querying with Brede Database can be done.

One or more coordinates can be selected for querying

and the results from the Brede Database (publications

related to the given activity coordinate) are displayed

by the plugin in a web browser (see Figure 5), ex-

ported to an XML ﬁle or saved in the bibliographic

ﬁle format (BibTeX, Reference Manager, RefWorks

or EndNote). We mention that the coordinates need

not necessarily be grabbed from SPM in order to

make a query. The coordinates can also be entered

manually in a manner similar to the functionality on

the webpage of the Brede Database.

COORDINATE-BASED META-ANALYTIC SEARCH FOR THE SPM NEUROIMAGING PIPELINE - The BredeQuery

Plugin for SPM5

Figure 3: Relationships and taxonomy of the regions in brain associated with superior temporal gyrus. The entire ontologies

for brain regions and brain functions are available together with the Brede Database.

The user is also able to perform an ‘experi-

ment search’ (available in the Brede Database

service) via the BredeQuery. It has previously

been suggested how a similarity can be computed

between one set of coordinates and a volume or

another set of coordinates (Nielsen and Hansen,

2004). This procedure required the conversion

of the set of coordinates to a volume by kernel

density estimation. It is, however, not necessary

to convert the coordinates to a volume if only the

similarity between two coordinates sets are to be

compared. It will then generally be faster to compute

the similarity based on all coordinate-coordinate

pair-wise similarities and perform a weighted sum-

mation. There are multiple ways to compute the

similarity. Presently, the web-service for the Brede

Database uses the following Gaussian/Euclidean form

q,e

√

∑

m=1

∑

n=1

exp



−x

)

+ (y

−y

)

+ (z

−z

)

−2σ



where σ is set to 10 millimeters, (x

) is the

mth of M three-dimensional query coordinates,

while (x

) is the nth of N three-dimensional

coordinates in the Brede Database. The factor 1/

√

aims to regularize for the number of coordinates in

each set so that sets with many coordinates do not

dominate the search result. A corresponding weight

for the query coordinates is not necessary, since this

factor will be equal for all queried sets of coordinates

of the database.

Following the terminology of BrainMap, a set of

coordinates is in the Brede Database called an ‘expe-

riment’ (Fox et al., 1994), thus the name ‘experiment

search’.

The Perl function that presently pro-

vides the search functionality from the Brede

Database webpage is part of the Brede Tool-

box, and this toolbox is available on the Internet

http://hendrix.imm.dtu.dk/software/brede/.

5 EXAMPLE SESSION

In this section we present the example steps for the

user session in SPM5 using BredeQuery plugin. The

sample data – Single subject epoch (block) audi-

tory fMRI activation data, taken from SPM webpage:

http://www.ﬁl.ion.ucl.ac.uk/spm/data/auditory/, was

used for this demonstration purposes. An example

SPM5-BredeQuery session may proceed with the fol-

lowing steps:

1. In an analysis of PET/fMRI data with SPM5, the

data set is loaded, a statistical analysis is per-

HEALTHINF 2009 - International Conference on Health Informatics

Figure 4: The demonstration of example usage of the BredeQuery plugin in SPM. The user can see regions with signiﬁcant

brain activation in the SPM results window. The region of activation represented by the coordinate (50, -34, 10) in MNI space

is selected (a). The same coordinate, transformed to the Talairach space using Lancaster’s MTT

SPM

transformation is marked

on the BredeQuery’s coordinates list as (45, -35, 11) (b). Afterwards, the user is able to submit coordinate-based queries to

the Brede Database and get the articles related to the same (or nearby) brain regions.

formed and, ﬁnally, results are displayed in an

SPM window. The regions of signiﬁcant brain

activation are marked and their coordinates can

be retrieved. The currently selected activation is

marked with a small red arrow. In our example

(Figure 4), the user has loaded and analyzed the

data. After the analysis, he has selected the acti-

vation, represented by the coordinate in the MNI

space as (50, -34, 10) – see Figure 4a.

2. The BredeQuery plugin is run by choosing the

BredeQuery entry in the SPM’s toolbox pop-up

menu. All coordinates are grabbed by the plugin

and shown in the coordinates list. They are trans-

formed according to the chosen MNI-to-Talairach

transformation. In our example, the coordinates

were transformed using the Lancaster’s MTT

SPM

afﬁne transformation. The user has pressed the

Mark red SPM chosen value button and the pre-

viously selected coordinate (50, -34, 10) in MNI

space, transformed in the plugin to (45, -35, 11)

in Talairach space, is marked in the plugin’s coor-

dinates list – see Figure 4b.

3. The user has pressed the Query button in the

Query Brede database [web browser] panel

(shown on Figure 1) and the webpage with the

query results (related articles) has appeared. The

user is now able to compare the obtained results

and conclusions with those from the retrieved ar-

ticles. The webpage results from our example are

displayed on Figure 5. The ﬁrst match from the

Brede Database is to a coordinate in superior tem-

poral gyrus from an experiment titled Tics during

Tourette’s syndrome. There is a link to the taxo-

nomy of the regions in brain associated with supe-

rior temporal gyrus (see Figure 3). Moreover, the

user can see the abstract to the article connected

with the above mentioned experiment.

4. The user wants to reference some of the articles

from the retrieved results in a manuscript. He

has selected the bibliographic format, which he

is going to use (in this example case ‘BibTeX’),

pressed Query button in the Query Brede database

[export to ﬁle(s)] panel (shown on Figure 1) and

the BibTeX ﬁle with the references was obtained.

5. The user has discovered a missing feature in the

BredeQuery plugin. He thus has pressed the Feed-

back button (Figure 1) and sent a comment to the

develop team.

6 FUTURE WORK

The presented plugin for SPM5 is still under develop-

ment and more features are planned.

It was recently emphasized that there are many

separated research communities in neuroscience,

which do not want to share or exchange the ex-

perimental data (Ascoli, 2006). Researchers have

expressed concerns that sharing of data can lead

to unfair use (Teeters et al., 2008). However,

data sharing is an important current issue in neu-

roscience (Kennedy, 2007; Liu and Ascoli, 2007)

and it is believed that broad data sharing could lead

to breaktroughs in our understanding of brain func-

tion (Van Horn and Ball, 2008). Invoking online

social networks and computer-based communication

can support closer relationships and trust (Lampe

COORDINATE-BASED META-ANALYTIC SEARCH FOR THE SPM NEUROIMAGING PIPELINE - The BredeQuery

Plugin for SPM5

Figure 5: Brede Database query – result displayed in a web browser. List of nearby coordinates to a queried coordinate,

displaying distance, the three-dimensional coordinates, the paper identiﬁer, the anatomical label for the retrieved coordinates

and short description of the experiment.

et al., 2006) hence, reduce the resistance to data sha-

ring.

Consequently, an interesting extension of the

functionality of the plugin can be a direct connec-

tion from the SPM environment to a neuroscientiﬁc

research community, web service or social network.

The user would be able to upload the coordinates,

results of the analysis, to his own account and save

in the assigned server disk space in order to process

them later. He can decide whether he wants to keep

it private, share only with his research group or alter-

nately release it as a public resource to all users of the

service.

It is also possible to employ the BredeQuery

plugin to expand the Brede Database. The increment

in number of the articles stored in the database could

cause bigger interest from the neuroscientists. They

could then be encouraged to register their published

or unpublished publications in the database via the

BredeQuery plugin together with the reported coor-

dinates and keywords.

7 CONCLUSIONS

In this paper we presented the BredeQuery plugin for

SPM5 - an application which offers a direct link from

SPM5 to the Brede Database. We provide a mech-

anism which allows the SPM user to ﬁnd references

to articles which relate to the similar brain activation

areas through so-called coordinate-based searching.

Moreover, the BredeQuery plugin facilitates the cre-

ation of the bibliography ﬁles in popular formats.

HEALTHINF 2009 - International Conference on Health Informatics

ACKNOWLEDGEMENTS

We would like to thank Torben Lund and Julian Ma-

coveanu for very constructive comments and feed-

back. This work is supported by Lundbeckfonden

through the Center for Integrated Molecular Brain

Imaging (CIMBI) – www.cimbi.org.

REFERENCES

Ascoli, G. A. (September 2006). The Ups and Downs

of Neuroscience Shares. Neuroinformatics, 4:213–

216(4).

Brett, M. (1999). The MNI brain and the Talairach atlas.

MRC Cognition and Brain Sciences Unit.

Cui, X. and Li, J. (2007). xjView - a viewing program for

SPM. http://people.hnl.bcm.tmc.edu/cuixu/xjView/.

Fox, P. T., Mikiten, S., Davis, G., and Lancaster, J. L.

(1994). BrainMap: A database of human function

brain mapping. In Thatcher, R. W., Hallett, M., Zef-

ﬁro, T., John, E. R., and Huerta, M., editors, Func-

tional Neuroimaging: Technical Foundations, chap-

ter 9, pages 95–105. Academic Press, San Diego, Cal-

ifornia.

Friston, K., Ashburner, J., Kiebel, S., Nichols, T., and

Penny, W., editors (2007). Statistical Parametric

Mapping: The Analysis of Functional Brain Images.

Academic Press.

Hamilton, A. (2005). AMAT - a meta-analysis toolbox.

http://www.antoniahamilton.com/amat.html.

Keator, D. B., Gadde, S., Grethe, J. S., Taylor, D. V., and

Potkin, S. G. a. (2006). A general XML schema and

SPM toolbox for storage of neuro-imaging results and

anatomical labels. Neuroinformatics, 4(2):199–212.

Kennedy, D. N. (2007). Neuroinformatics and the Society

for Neuroscience. Neuroinformatics, 5:141–142.

Laird, A. R., Lancaster, J. L., and Fox, P. T. (2005). Brain-

Map: The Social Evolution of a Human Brain Map-

ping Database. Neuroinformatics, 5:65–78.

Lampe, C., Ellison, N., and Steinﬁeld, C. (2006). A

face(book) in the crowd: social searching vs. social

browsing. In CSCW ’06: Proceedings of the 2006

20th anniversary conference on Computer supported

cooperative work, pages 167–170, New York, NY,

USA. ACM Press.

Lancaster, J. L., Tordesillas-Guti´errez, D., Martinez, M.,

Salinas, F., Evans, A., Zilles, K., Mazziotta, J. C., and

Fox, P. T. (2007). Bias between MNI and Talairach

coordinates analyzed using the ICBM-152 brain tem-

plate. Human Brain Mapping.

Liu, Y. and Ascoli, G. A. (September 2007). Value Added

by Data Sharing: Long-Term Potentiation of Neuro-

science Research: A Commentary on the 2007 SfN

Satellite Symposium on Data Sharing. Neuroinfor-

matics, 5:143–145(3).

Nielsen, F.

A. (2003). The Brede database: a small database

for functional neuroimaging. In NeuroImage, vol-

ume 19. Elsevier. Presented at the 9th International

Conference on Functional Mapping of the Human

Brain, June 19-22, 2003, New York, NY.

Nielsen, F.

A., Christensen, M. S., Madsen, K. M., Lund,

T. E., and Hansen, L. K. (2006). fMRI Neuroinfor-

matics. IEEE Engineering in Medicine and Biology

Magazine, 25(2):112–119.

Nielsen, F.

A. and Hansen, L. K. (2002). Modeling of acti-

vation data in the BrainMap(TM) database: Detection

of outliers. Human Brain Mapping, 15(3):146–156.

Nielsen, F.

A. and Hansen, L. K. (2004). Finding related

functional neuroimaging volumes. Artiﬁcial Intelli-

gence in Medicine, 30(2):141–151.

Nielsen, F.

A., Hansen, L. K., and Balslev, D. (2004). Min-

ing for associations between text and brain activation

in a functional neuroimaging database. Neuroinfor-

matics, 2(4):369–380.

Pekar, J. (2006). A brief introduction to functional MRI.

IEEE Engineering in Medicine and Biology Maga-

zine, 25(2):24–26.

Talairach, J. and Tournoux, P. (1988). Co-planar Stereo-

taxic Atlas of the Human Brain. Thieme Medical Pub-

lisher Inc, New York.

Teeters, J. L., Harris, K. D., Millman, K. J., Olshausen,

B. A., and Sommer, F. T. (2008). Data Sharing for

Computational Neuroscience. Neuroinformatics.

Van Horn, J. D. and Ball, C. A. (2008). Domain-Speciﬁc

Data Sharing in Neuroscience: What Do We Have to

Learn from Each Other? Neuroinformatics.

Wager, T. D., Lindquist, M., and Kaplan, L. (2007). Meta-

analysis of functional neuroimaging data: current and

future directions. Social Cognitive and Affective Neu-

roscience, 2(2):150–158.

COORDINATE-BASED META-ANALYTIC SEARCH FOR THE SPM NEUROIMAGING PIPELINE - The BredeQuery

Plugin for SPM5