[Mne_analysis] differing weight_norm defaults for make_lcmv and make_dics

[Marijn van Vliet]

        External Email - Use Caution        

Hello Jeff,

there is currently no consensus for the best choice of parameters for beamformers. It seems that every work of literature on the topic has its own way of doing things, hence the enormous number of parameters. The LCMV beamformer is developed by Britta Westner and reflects the practices of the Center of Functionally Integrative Neuroscience, Aarhus University. The DICS beamformer is developed by me and reflects the practices of the Imaging Language Group, Aalto University. We are together with Christian Kiefer in the process of performing an extensive comparison study to map out the effects of all the parameters in practise.

Here is what the docstring of make_dics has to say about it:

The DICS beamformer is very similar to the LCMV (make_lcmv()) beamformer and many of the parameters are shared. However, make_dics() and make_lcmv() currently have different defaults for these parameters, which were settled on separately through extensive practical use case testing (but not necessarily exhaustive parameter space searching), and it remains to be seen how functionally interchangeable they could be.

The default setting reproduce the DICS beamformer as described in [5]:

inversion='single', weight_norm=None, normalize_fwd=True
To use the make_lcmv() defaults, use:

inversion='matrix', weight_norm='unit-gain', normalize_fwd=False

References:
[5] van Vliet, et al. (2018) Analysis of functional connectivity and oscillatory power using DICS: from raw MEG data to group-level statistics in Python. bioRxiv, 245530. https://doi.org/10.1101/245530

best,
Marijn.

--
Marijn van Vliet
Postdoctoral Researcher
Department of Neuroscience and Biomedical Engineering
Aalto University

> On 7 Nov 2019, at 16:55, Jeff Hanna <jeff.hanna at gmail.com> wrote:
> 
>         External Email - Use Caution        
> 
> 
> Hello,
> 
> The default value for weight_norm in mne.beamformer.make_lcmv is "unit-noise-gain," but is None in .make_dics. Is there a reason behind this, i.e. are there potential issues with the use of unit-noise-gain in DICS that aren't a problem with LCMV?
> 
> Best,
> Jeff Hanna
> _______________________________________________
> Mne_analysis mailing list
> Mne_analysis at nmr.mgh.harvard.edu
> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/mne_analysis