Elekta grad mag scaling

Discussions
1

Dear collegues,

Anybody has an idea how gradiometer and magentometer are scaled? I have seen in the brainstorm code a factor of 0.4, but I would like to understand why this factor has been chosen. Further, in the code there is a comment that this factor has been chosen somehow arbritary (in the 3.0 version). Anybody can give me a hint, please?

Thanks a lot in advance,

Best,

Stephan

2

Hi Stephan,

Working with mixed magnetometers and gradiometers on Elekta systems is complex. You have to deal with the simultaneous representation of two types of measures that have completely different meanings and amplitude ranges. The magnetometers record magnetic fields and the gradiometers record spatial derivatives of the same fields.

The values of the two types of sensors should NOT be compared directly. But there are a few cases where we want to use the two types of information at the same time. Typically when displaying the time series for all the sensors (double click on the “recordings” file in the database explorer):

=> The gradiometers values (in T/m) are converted into the magnetometers units (T) by multiplying them by the size of the sensor (~4cm). This operation generates a warning message in Brainstorm: “MEG Gradiometers values are multiplied by: 0.040 (axial factor)”, in function bst_scale_gradmag.m.
Please note that this scaling is not a mathematical truth, it’s just an artificial way to represent the two types of information on the same figure. This 0.04 value happens to work well in most cases.

Does that answer your question?
Francois

3

Hi Francios,

Thanks a lot for your awnser. I awnsered my question perfectly! I only have one last question with regard of the scaling. When calculating a common leadfield, do you have to scale the data the same way? I guess that a unit activation of a dipole gives different scales for the forward solution with respect to grad and mags. I am asking that, as I like the idea to use the grad and mag (and maybe EEG) during the same inversion process.

Many thanks and great job done with Brainstrom !!!

Best,

Stephan

4

The leadfield matrix computed with Brainstorm already contains the information about how each dipole projects onto each sensor. You do not have to worry about scaling magnetometers and gradiometers at this point.
The information about the geometry of the different types of MEG sensors is defined in the file: brainstorm3/toolbox/io/private/coil_def.dat (text file). And integrated in the channel files (definition of the sensors, fields Loc, Orient and Weight).

You can do the combined source reconstruction of gradiometers and magnetometers safely. As for the combination of MEG and EEG, I am not sure it’s a good idea for now. At least not with Brainstorm (I will add a warning to discourage the users from doing that). The question of the balance between the two modalities in the source reconstruction, plus the fact that we might not observe the same things with the two techniques, makes it a very complex problem that we do not handle properly in Brainstorm now.

Best,
Francois

5

Hi Francios,

Thanks a lot for the informativa awnser! Now things are much more clearer to me. Before I have been working with system that only had magnetometers, and now I work with a Elekta system. Thus, I was a little bit unsure how to handle the different sensor types.

Thanks a lot,

Best,

Stephan