Tsss algorithm and bad channels

SaarK · January 9, 2023, 1:20am

Hey all!
I am running tsss from brainstorm. I have already configured python to work with brainstorm, and it seems to run. Thank you.
I just want to make sure the tsss does not include bad channels when running. Meaning, that channels that I defined as bad channels are not passed to the tsss, and that the tsss doesnt include their data when running. Is this the case?
Thank you!
Saar.

Francois · January 9, 2023, 9:40am

Brainstorm sets the field bads of the Info object with the list of bad channels and then calls the MNE-Python function mne.preprocessing.maxwell_filter:

github.com

brainstorm-tools/brainstorm3/blob/master/toolbox/io/out_mne_data.m

function [mneObj, DataMat, ChannelMat, iChannels] = out_mne_data(DataFiles, ObjType, ChannelFile, SensorTypes)
% OUT_MNE_DATA: Converts a data file or a set of data files into a MNE-Python object (Raw, Epoched or Evoked)
% 
% USAGE:  [mneObj, DataMat, ChannelMat, iChannels] = out_mne_data( DataFiles, ObjType, ChannelFile=[], SensorTypes/iChannels=[]);
%         [mneObj, DataMat, ChannelMat, iChannels] = out_mne_data( DataMat,   ObjType, ChannelMat=[],  SensorTypes/iChannels=[]);
%
% INPUTS:
%    - DataFiles    : String or cell-array of strings, relative path(s) to data file(s) available in the database
%    - DataMat      : Brainstorm data file structure (if multiple files, the F matrix matrix must be [nEpochs, nChans, nTime])
%    - ObjType      : Class of the returned MNE-Python object {'Raw', 'Epoched', 'Evoked'}
%    - ChannelFile  : Relative path to a channel file available in the database (if not provided: look for it based on the DataFile)
%    - ChannelMat   : Brainstorm channel file structure
%    - iChannels    : Vector of selected channel indices
%    - SensorTypes  : Names or types of channels, separated with commas

% @=============================================================================
% This function is part of the Brainstorm software:
% https://neuroimage.usc.edu/brainstorm
% 
% Copyright (c) University of Southern California & McGill University

This file has been truncated. show original

Whether this is sufficient to exclude completely the bad channels from the tsss computation, I'm not sure...
@Alexandre @Eric89GXL @Luc ?

Eric89GXL · January 9, 2023, 2:01pm

Indeed adding bad channels to info['bads'] should exclude them from tSSS (and standard SSS) computation. The bad (MEG) channels should then be interpolated by SSS and automatically be removed from info['bads']

SaarK · January 10, 2023, 2:30am

I see. OK great, thank you guys !!

Francois · January 10, 2023, 1:02pm

@Eric89GXL
Thanks for the confirmation!
Would you recommend we always use the re-interpolated channels, for source estimation for example?
Is it what happens when using MaxFilter (all the bad channels are marked as good)?

Eric89GXL · January 10, 2023, 10:26pm

Yeah you can, and might as well (we do in MNE-Python) -- it makes things like group studies easier, and shouldn't have much effect on things like inversion (effective data rank stays the same, e.g., < 80 somewhere).