Functional Brain Signal Processing: EEG

1
Functional Brain Signal Processing EEG
fMRILesson 16
M.Tech. (CS), Semester III, Course B50

• Kaushik Majumdar
• Indian Statistical Institute Bangalore Center
• kmajumdar_at_isibang.ac.in

2
Multi-Voxel Pattern Analysis
Poldrack et al., 2011
MVPA is concerned about simultaneous activation
patterns across multiple voxels in different
parts of the brain with the help of machine
learning algorithms.
3
A Task Specific MVPA
Haynes Rees, 2006 Haxby et al., 2001
FFA fusiform face area (red). PPA
parahippocampal place area. During face image
presentation (red arrow) signal is enhanced in
FFA and during building image presentation (blue
arrow) signal is enhanced in PPA.
Activation patterns in temporal lobe during
visualization of chair and shoe. r is correlation
coefficient between activation patterns on same
and different objects.
4
A Hypothetical Scheme for MVPA Computation
Norman et al., 2006
A cortical activation pattern coding scheme
during chair and shoe visualization.
Two dimensional projection of high dimensional
feature space where binary classification has
been accomplished by the red-dashed line.
5
MVPA by Statistical Pattern Recognition
Haynes Rees, 2006
6
Four Basic Steps of MVPA

• Feature selection which voxels will be involved
  in classification analysis.
• Pattern assembly sorting the data into discrete
  brain patterns corresponding to the patterns of
  activity across selected voxels at a particular
  time in the experiment.

7
Basic Steps (cont)

• Classifier training feeding a subset of the
  leveled patterns into a multivariate pattern
  classification algorithm. The algorithm learns a
  function that maps a voxel activity pattern into
  an experimental condition.
• Generalization testing putting the
  classification algorithm to test on hitherto
  un-presented data.

8
Nature of Classifiers

• Most MVPA studies used linear classifiers
  including correlation based classifiers.
• Neural networks.
• Support vector machine.
• Bayesian classifiers.

9
Correlation Based Classifier
Haxby et al., 2001
10
References

• K. A. Norman, S. M. Polyn, G. E. Detre and J. V.
  Haxby, Beyond mind-reading multi-voxel pattern
  analysis of fMRI data, Trends Cog. Sc., 10(9)
  424 430, 2006.
• R. A. Poldrack, J. A. Mumford and T. E. Nichols,
  Handbook of Functional MRI Data Analysis,
  Cambridge University Press, Cambridge, New York,
  2011. Chapter 10.

11
References (cont)

• J.-D. Haynes and G. Rees, Decoding mental states
  from brain activity in humans, Nat. Rev.
  Neurosci., 7 523 534, 2006.
• J. V. Haxby et al., Distributed and overlapping
  representations of faces and objects in ventral
  temporal cortex, Science, 293 2425 2430, 2001.

12
THANK YOUThis lecture is available at
http//www.isibang.ac.in/kaushik