Dear all, 

We are carrying out a PET analysis in SPM. When we do the second level analysis we apply a binary mask in MNI space to avoid the signal from the skull and the peripheral CSF, however the second level analysis shows a problematic signal in the ventricles which the mask is not able to avoid. 

 

The other question is about segmentation. Could the segmentation of GM and WM only (w/o CSF) avoid the signal into the ventricles in the second level analysis or is it only useful for the normalisation process? In other words, can segmentation act as a mask?

Does anyone have any ideas of how to solve this, either by creating a more defined mask that excludes the ventricles or by modifying the pipeline?

Any help would be appreciated, 

Best regards, 

Adriana Roca.