BrainMapTM


DTU Informatics >>> HBP > Projects > BrainMap

Icon with BrainMap kernel density estimate

Functional volumes modeling

Functional volumes modeling models the distribution of stereotaxic coordinates. Different models can be applied and we have applied Gaussian mixture models [1] and kernel density estimates [2] with data from the original BrainMapTM database at the Research Imaging Center in San Antonio.

Visualization in VRML of the estimated probability densities are available for

  1. Kernel density estimation of coordinates marked with as "audition" (red wireframe) and "vision" (green surfaces) in the BrainMap annotation. The VRML corresponds to the figure in [2].
  2. Conditional Gaussian mixture models of coordinates marked as motion ("M" textured surfaces), perception (wireframe) or cognition (green surfaces) by the BrainMap "behavioral domain" annotation. The VRML corresponds to Figure 1 in [1].

References

  1. Modeling of BrainMap data
    Finn ┼rup Nielsen and Lars Kai Hansen.
    Manuscript 1999 May (Submitted to NIPS'99 but not accepted)
    [ Entry in department archive | PDF from department archive | Gzipped Postscript ]
  2. Functional Volumes Modeling using Kernel Density Estimation
    Finn ┼rup Nielsen, Lars Kai Hansen.
    Abstract, 2000 January.
    [ Entry in department database | PDF from department database ]
Finn Årup Nielsen, 2009.

Icon for BrainMap modeling

Outlier detection.

Cerebellum VRMLs: [ Probability density volume | Talairach Atlas | ICBM Atlas ]

Novelty in the BrainMaptm database can be spotted using mathematical modeling of the activation foci [1]. The novelty is detected as a discrepancy between the 3D coordinates and the anatomical label. The method finds a few entry errors in BrainMaptm together with a few errors in the original articles. The result is available as an automatic generated list of BrainMap outliers (Note the functional interface to the BrainMap database has been changed so the URL for the database does not work).

Technical: The mathematical modeling is based on probability density modeling using the Specht kernel density estimate (Parzen window) applied on the 3D Talairach space conditioned on the words and phrases of the anatomical label. Novelty detection is implemented by directly ranking the likelihood.

  1. Modeling of activation data in the BrainMapTM: Detection of outliers
    Finn Årup Nielsen, Lars Kai Hansen
    Human Brain Mapping, 15(3):146-156, 2002 March. PubMed
    [ HTML Publisher abstract | Submitted, PostScript | ResearchIndex ]
  2. Mining the BrainMapTM database: Detection of outliers
    Finn ┼rup Nielsen, Lars Kai Hansen, Ulrik Kjems,
    Manuscript, 2000
    [ Postscript ]
  3. Modeling of locations in the BrainMap database: Detection of outliers
    Finn ┼rup Nielsen, Lars Kai Hansen, Ulrik Kjems, NeuroImage, 2001 supplement, 13(6):S211
    [ Abstract in PostScript ]
Finn Årup Nielsen, 2000 May, November, 2001

Icon for BrainMap visualization

Icon for BrainMap visualization

V5/MT - visual motion.

Comparison of different functional neuroimaging studies that reference the visual area V5 supposedly involved in visual motion processing. There are two studies from BrainMap [1] [2] and one study from 1998 [3].
  1. Area V5 of the human brain: Evidence from a combined study using positron emission tomography and magnetic resoniance imaging
    J. D. G. Watson, R. S. J. Frackowiak and J. V. Haynal.
    Cerebral Cortex, 3, 79-94.
    [ BrainMap (link no longer available) | PubMed ]
  2. Different perceptual tasks performed with the same visual stimulus attribute activate different regions of the human brain: A positron emission tomography study
    P. Dupont, Orban GA, Vogels R, Bormans G, Nuyts J, Schiepers C, De Roo M and Mortelmans L.
    Proceedings of the National Academy of Sciences, 90: 10927-10931.
    [ BrainMap (link no longer available) | PubMed ]
  3. The functional anatomy of attention to visual motion A functional MRI study
    Christian Buchel, Oliver Josephs, Geraint Rees, Robert Turner, Chris D. Frith and Karl J. Frison
    Brain, 121: 1281-1294, 1998.
    [ Oxford University Press | Brain | Abstract ]
Finn Årup Nielsen, 1999 June.

Icon for visualization with Ian Law's data

Automatic labeling

Automatic labeling of activation foci of a new study is possible if a probability density model of the Talairach coordinates is constructed. We constructed such a model with a new kind of EM-type algorithm "Generalizable Gaussian Mixtures". As the new study we used a saccadic eye movement study [1]. The results have been presented in a paper [2].
  1. Parieto-occipetal cortex activation during self-generated eye movements in the dark
    Ian Law, Claus Svarer, Egill Rostrup and Olaf B. Paulson
    Brain, 121(11): 2189-2200, 1998 November.
    [ Oxford University Press | Brain | Abstract ]
  2. Modeling of BrainMap data
    Finn ┼rup Nielsen and Lars Kai Hansen.
    Manuscript 1999 May (Submitted to NIPS'99 but not accepted)
    [ Entry in department archive | PDF from department archive | Gzipped Postscript ]
Finn Årup Nielsen, 1999 May.

Icon for human brain HBP THOR Center for Neuroinformatics, Human Brain Project Repository (This server)
IMM DTU Informatics
DTU Technical University of Denmark

© DTU, 1995-2000, 2002, 2009
http://neuro.imm.dtu.dk/projects/brainmap/brainmap.html
$Id: brainmap.html,v 1.8 2009/06/30 17:59:35 fn Exp $