How about using a GM mask to specific signal space? I guess the most probable reason for signals in ventricle voxels would be registration errors, which might be confounded with the different blocks of conditions. SG. On Sep 25, 2012, at 7:22 PM, Moritz Lehne wrote: > Dear fMRI experts, > > In an fMRI study using music stimuli I consistently find activations in the ventricles (especially in the posterior horn of the lateral ventricles) when contrasting music vs. silence (in addition to the typical areas related to auditory processing). > > Does anyone have an idea what this may be related to and how to get rid of it? (I already tried adding the mean BOLD signal of CSF voxels as a nuisance regressor into the model - without success. I also tried different cutoff frequencies for the highpass filter. Unfortunately, I didn't record any physiological measures like heart rate or respiration that I could add to the model.) > > Thank you for your answers (and sorry for bothering you with questions that are not exactly SPM related). > > Cheers, > Moritz --- Seung-Goo KIM