Thoughts on fMRI

Researchers at the ATR Computational Neuroscience Lab in Japan have made some interesting progress using fMRI for visual letter recognition and has me thinking about the field in more depth

(http://www.medgadget.com/archives/2008/12/fmri_extracts_images_from_the_brain.html).
They also discuss the possibilities using it for doing dream analysis. Check out their site at
http://www.cns.atr.jp/indexE.html. Still researching and learning, I thought I would examine packages for the R language. The fmri package in R is a library of functions for performing fMRI analysis based on Tabelow, et. al. (2006). The web site for their work is at

(http://www.wias-berlin.de/people/tabelow/doc/tsb2008.pdf).

The reference manual can be found at (http://cran.r-project.org/web/packages/fmri/fmri.pdf). The package has routines for Non-Gaussian Component Analysis based on Blanchard et. al. (2005) and Independent Component Analysis. Also, the expected BOLD response can be created from task indicator functions as show in the manual. A good paper on ICA by Bai, Shen and Truong (2006) is at (
http://www.samsi.info/200607/ranmat/workinggroup/rc/ranmat-rc/Truong_main.pdf). They use the R package called AnalyzeFMRI. The manual is at

http://cran.r-project.org/web/packages/AnalyzeFMRI/AnalyzeFMRI.pdf

To get started using this packages there are instructions at

https://mri.radiology.uiowa.edu/mediawiki/index.php/FMR_Analysis_in_R

that involves converting an AFNI Image into a single 4D Analyze image and then creating a
mask. AFNI (Analysis of Functional Neuroimages) is an open source application for fMRI data
analysis at http://afni.nimh.nih.gov/afni/. Currently, there are no ports for the application to the Windows environment.

I have done some initial experimentations with both packages using simulation data and
will discuss that more in coming posts. In order to put these packages in perspective, there is a good, but dated article by Gold et.al. (1998) on "Functional MRI Statistical Software Packages: A Comparative Analysis". As far as the teaching aspect, I like this paper by Lai, et. al. "Teaching Statistical Analysis of fMRI Data at http://www.vanth.org/docs/Lai_Gollub_Greenberg_ASEE03.pdf; however, they use Matlab software for their work.

References

Blanchard, G., Kawanabe, M., Sugiyama, M., Spokoiny, V. and Müller K.-R. (2005). In Searchof Non-Gaussian Components of a High-Dimensional Distribution. Journal of Machine LearningResearch. pp. 1-48.

Tabelow, K., Polzehl, J., Voss, H.U., and Spokoiny, V.(2006). Analysing fMRI experiments with
structure adaptive smoothing procedures, NeuroImage, 33:55-62.