[Mne_analysis] Forward solution and inverse operator gain for deep discrete dipoles

[Matti Hamalainen]

On Jun 3, 2009, at 3:13 PM, Andres Felipe Salazar Gomez wrote:

> Hi MNE users,
>
> With mne_volume_source_space I got a model with 6 discrete sources,  
> 4 in
> deep brain structures (brain stem) and 2 at the cortex. After  
> getting the
> forward solution and the inverse operator I was able to find the  
> estimates
> for these 6 sources but those at the cortex seem to be two orders of
> magnitude smaller, which is not what I was expecting to get.
>
> My question is: when MNE calculates the forward solution and the  
> inverse
> operator, does it take into account how deep are the dipoles? I am
> wondering if somehow the gain in the G matrix (fwd.sol.data) or the  
> noise
> covariance matrix (inv.noise_cov.data) is smaller for the cortical  
> dipoles
> and therefore that is why I am seeing this differences in magnitude.  
> Any
> idea?
>
> Thank you very much in advance for any help regarding this issue.
> Sincerely,

Hi,

The deeper dipoles are not weighted any differently from the cortical  
ones.

The difference in magnitude is because on the cortex the estimates  
have a chance (and will) spread out thus creating a weak distributed  
source in MNE whereas there are only few brain stem dipoles have to  
account for the data and will, therefore, be stronger. How to best  
deal with this is subject to further investigation.

- Matti