HI @leoallves
In general EEG is less accurate for assessing deep brain sources ( compared to MEG) due to many factors such as the sensitivity of EEG to volume conduction effects and variations in tissue conductivity, as well as the inverse methods.
You can refer to some previous discussions in the forum on similar topic :
- Inverse Algorithm from Sensor to Deep Areas (hippocampus, thalamus, amygdala)
- https://neuroimage.usc.edu/forums/search?q=deep%20source%20thalamus
Once you have your source map, you can then apply the connectivity at the source space that you have defined for your model.