In other signal processing fields, such as the earth sciences (my background is geophysics) it is reasonable and common to take a spherical coordinate system and related raw signals and project them onto an appropriately chosen 2 d plane to support analysis. Examples may include mapping seismic data recorded on the surface of the earth.
Is it clinically reasonable to map EEG channel data in a similar way such as mapping montages that cross the naison/inion plane onto the naison/inion plane?
If it is reasonable then are there standard methods for doing so?
are you suggesting to plot a sensor that is located “below” the Nasion-Inion plane in 3D space by orthogonally projecting it onto this plane, thereby placing it closer to the center of the circle formed by the Nasion-Inion plane?
Though i am first asking if this is an appropriate thing to do from a clinical perspective. Different montages may cover different parts of the brain and I don’t know enough to know if it is reasonable to treat the brain as a homogenous entity from a signal processing perspective.
On the other hand when working with vectors in other physics problems such projection is quite common - i.e determination of a vector’s contribution to each of the axes in the target coordinate system.
when MNE produces the 2D projections, sensors below the Nasion-Inion plane are moved farther away from the head circle center. I believe this makes the most sense and provides reasonable interpretability.
Of course, avoiding such a dimensionality reduction altogether is probably the cleanest way to deal with the data … hence, I suggest you also take a look at the plot_field() method of evoked objects, which I believe is an awesome way to visualize the data without having to worry about the question of how to best project 3D to 2D … that is, if you have evoked data.
looks a lot like vector projects used in the earth sciences such as this one for polar atmospheric flow. In the earth sciences people use a library called NCAR.