This web-site contains volumes representing brain regions.
You can browse them from the alphabetic index.
The volumes are established by probability density estimates of locations
from the BrainMap database.
The locations have been symmetrized and converted by Matthew
Brett's nonlinear transformation
to get from Talairach-space to approximately the MNI-space.
Original data: From BrainMap
The basis for the volume are the data recorded in the "original"
This is not our database but a database setup
by Peter T. Fox and Jack L. Lancaster at the Research Imaging Center, San
So if you use the volumes in a scientific publication please cite their work
with one of their references listed below.
The "original" BrainMap database is no longer available but a
new BrainMap database by the same group is accessible from http://brainmap.org/.
(The volumes are not constructed from information in this database)
The information used from the original BrainMap database are the
3-dimensional Talairach coordinates as well as the 'lobar anatomy'
field for each coordinate.
Method: Kernel density modeling of Talairach coordinates
For each sub-phrase in the 'lobar anatomy' field all Talairach
coordinates are found that have the sub-phrase in the field.
Each set of coordinates is then modeled with a kernel density
modeling to construct a probability density conditioned on the
The method is described a bit more in the Modeling of
activation data in the BrainMapTM database article below.
The coordinates have be copied between the right and left
hemisphere to make the volume symmetric.
The coordinates have also been converted from Talairach space to
(approximately) MNI-space with Matthew Brett's transformation.
The individual steps in the procedure are implemented in the
It is programed in Matlab and available for download.
Warning: Approximative volumes
Since the volumes are based on annotation of Talairach coordinates,
that represents brain activation hot spots reported in the
scientific literature, the volumes are inherently coarse
approximations to the "real" brain regions.
You should not take the volumes to define precise borders of
They are simply regions were other researchers did (with a certain
probability) mark it with a specific neuroanatomical label.
They are not constructed by a neuroanatomist carefully delineating
the brain regions.
If you want more precisely delineated volumes you might want to
look into the AAL or
atlases. The WFU
Pickatlas Matlab program has features to work with these atlases.
Also there exist probabilistic atlases by Alexander Hammers and
Claus Svarer has a program that allows for the probabilistic atlas
to be fitted to MRs from individual subjects, see pvelab.
If you use the data in a scientific work you might cite it as:
The above reference is just a short article. The original BrainMap
database is described in depth in the following book chapter:
BrainMap: A database of human function brain mapping
Peter T. Fox, Shawn Mikiten, Gwendolyn Davis, Jack
In Functional Neuroimaging: Technical Foundations,
Editors: Robert W. Thatcher and M. Hallett and T. Zeffiro and
E. R. John and Michael Huerta, Academic Press (San Diego,
California), chapter 9, pp. 95-105, 1994. ISBN 0126858454.
The pages are constructed with the
Toolbox programed in Matlab.
It can be referenced as
The kernel density estimation is described a bit more in the
Modeling of activation data in the BrainMapTM
database: Detection of outliers
Finn Årup Nielsen, Lars Kai Hansen.
Human Brain Mapping, 15(3):146-156, 2002 March.