There appears to be a mismatch between how fMRI results are presented graphically and how we know the analyses are normally done.

Say we do a group-level analysis for some contrast of interest, and we thus obtain voxelwise t-values that we plot in a typical 'brain activation' map. Prior to the analysis however, the images typically undergo Gaussian smoothing (blurring), and this would mean that the voxelwise information is itself only as spatially-accurate as the images were after they were smoothed.

This would mean, their spatial resolution should be considerably lower than that of an anatomical (T1-weighted) image. Indeed, in many statistical-parametric-maps, the activated voxels do appear of a lower resolution, but in some (see below) they seem to be of the same resolution as the unsmoothed voxels around them (normally: 1mm). How can this be? Is it an error on the part of the scientists who make these plots, or a glitch in my understanding of how smoothing is done in fMRI analyses?

enter image description here


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.