My fMRI pages

You can find on this pages various information related to MRI and fMRI.

Quite a long time ago now, I wrote a presentation with my friend Seppo Matthilla (a great astronomer, now in Turku) on the basis of MRI physics presenting the basis physical principles (MNR, magnetism, gradient) underlying MRI acquisition and image reconstruction.

Others pages:

Guidelines for fMRI articles: during the first months of 2005, a discussion started on the SPM-list on how fMRI/PET methods and results should be described. Usually, article descriptions are 'deficient' leading to the inability to reproduce results. I report on this page, proposals made several persons, noticeably Max Gunther, Matthew Brett, Karsten Specht, Russ Poldrack, Jesper Andersson and Thomas Nichols. For each part of an article (materiel, method, result), a minimum of information should be reported. An article as been publish in 2008 by Russ and coll. in NeuroImage on this topic. There is now as well the COBIDAS report from which I extracted some tables to reuse for conviniance.

fMRI designs: neuroimaging experimental designs are more complicated than behavioral experimental designs as, in same time as thinking about experimental effects, we have to think about fMRI acquisition parameters. I present here key notions on the different types of fMRI designs (blocks, event related, mixed) and clues to 'optimize' these designs. Many information can also be found on Rick Henson website, part of this is presented here.

Statistical designs: one major issues with hiearchical linear models is how to combine parameters from the 1st level analysis and especially what contrasts are valid - Jan Glascher and Darren Gitelman wrote a note a while back (SPM5 style) distributed on the SPM webmail on how to use the 'flexible factorial' option. For an up-to-date a well documented way on performing repeated measures analysis using univariate models see this paper by Martyn McFarquhar. There is even better, you can perform repeated measures analysis the right way, i.e. using multivariate models, again check Martyn's paper and toolbox.

MR principles of the BOLD contrast: the BOLD contrast is the most used neuroimaging technique over the world. Beyond the blobs, it's important to understand that it relies on the magnetic properties of blood cells and vasculature to interpretation these blobs. I summarized here the discoveries I found the most relevant - I try to keep this part up-to-date but many publications appears every months so if I missed something really important thanks for letting me know.


My Favorite links


MRC-CBU website
Joseph Hornak's website
Thomas Nichols' website