[FieldTrip] EEG Inverse Solution assumption: Expected Dipole Orientation

[Lukas Hecker]

Dear fellow fieldtrippers,

I am a student from the university of Freiburg (germany) and I am currently working on a comparison of various EEG inverse solutions as implemented in fieldtrip and SPM. 

I have simulated a large number (10k) of smooth neural source configurations using a regular grid (20mm spacing) as source model. These source simulations were then projected on 32 scalp electrodes using a BEM headmodel and the ft_dipolesimulation function. 

In 10k Simulations I simulated varying numbers of sources (between 1 and 5) with random dipole orientations. 

After getting nice results with the source reconstruction of inverse solutions implemented in fieldtrip (e.g. eLORETA) I tried to reconstruct sources using „multiple sparse priors“ (MSP) in SPM. 

However, I came across the problem that in the current implementation they assume a surface-based source model with fixed dipole orientations perpendicular to the respective surface area (triangle). In my simulations, however, the orientation of each individual source patch was chosen randomly. Source reconstruction success using MSP was very limited. 

Here are my questions: 

[1] Can I expect MSP to reconstruct sources when the assumptions about dipole orientations are not met in my simulation of neural activity?

[2] Since I read that structures of forward-models in fieldtrip and SPM are shared: Is it possible to use the leadfield that I used for the simulations in fieldtrip (with free dipole orientations) 
for SPM inverse solutions?

Thank you very much for your help.



With best regards,
Lukas Hecker