Dear Brainstorm users,
I am performing time-frequency decomposition of EEG recordings in specific frequency bands (e.g., delta, theta, alpha) and for specific surface (e.g., left/right hippocampus, cerebellum) and volumes scouts (e.g., left/right thalamus, pallidum). I only want to save the mean power values of the specified surface/volume scouts by doing so: “Hilbert transform > Use scouts > Scout function: Mean” and “Hilbert transform options > Edit… > Group in frequency bands”.
This works perfectly but results are somewhat curious. For instance, when checking the time-frequency maps of the left/right cerebellum I get three TF maps: Cerebellum L.1, Cerebellum L.2, Cerebellum L.3 and Cerebellum R.1, Cerebellum R.2, Cerebellum R.3. This is the same for some volumes (e.g., thalamus L.1, thalamus L.2, thalamus L.3 and thalamus R.1, thalamus R.2, thalamus R.3).
However, for some reasons, when checking the hippocampus I only get one TF map for each of the left and right hemisphere: Hippocampus L and Hippocampus R. This is the same for some volumes such as the pallidum: Pallidum L and Pallidum R.
Note that this is consistent across conditions and subjects. I just do not understand why do I get three different TF maps.
Thank you.
Arnaud
Hi Arnaud,
What you’re observing is due to a limitation of the time-frequency decomposition functions. They can handle correction either constrained (1 value per grid point) OR unconstrained sources (3 values per grid point).
However, in the case of mixed head models, you are able to select simultaneously scouts that a coming regions with orientation constrains (eg. hippocampus) and regions without orientation constrains (eg. cerebellum, thalamus). In this case, it processes them as if they were all constrained, which prevents the grouping of the three orientations (1=x,2=y,3=z) into one value that usually occurs when processing unconstrained sources alone.
If you select only the unconstrained scouts (cerebellum, thalamus, etc) and run the process, it should sum the power along the three orientations correctly and produce only one value per scout (hippocampus L, hippocamus R, thalamus L, thalamus R). Then run it separately for the constrained scouts.
Otherwise, put your TF file (with the Cerebellum L.1, Cerebellum L.2, Cerebellum L.3, etc) in the Process1 box and run process “Average > Average signals”: Average by scout, arithmetic average.
It would compute a mean of the 3 orientations instead of a sum, therefore you would get the same values but with divided by 3.
I’d recommend you go with the other option to avoid any confusion due to this factor 3 (computing unconstrained scouts separately).
I added this issue to the todo list, but it will take a long time before we reach this item (http://neuroimage.usc.edu/brainstorm/Next#Processes).
Cheers,
Francois
Hi François,
Thanks a lot for the clear and informative reply. This is really helpful.
Cheers,
Arnaud