Before you apply SSS, it is a very good idea to mark bad channels. These can be detected with the "autobad" feature of Maxfilter (though this does not work for timeperiods in which continuous HPI is active, which is why we often only start cHPI after about 20s, and see below for bugs associated with this option). We recommend to scan the channels by eye during acquisition, keep a list of bad channels, and mark them during acquisition. Afterwards, you can use an MEG analysis program to investigate you data. The best option available is mne_browse_raw which is a part of MNE suit available on the linux cluster. BESA can be used on windows, and EEGLab toolbox on matlab but you need to change the data format before reading it into these two. Once you know the bad channels, you can then tell this to maxfilter (-bad option, see below), or you can mark them in the actual FIFfile using a commandline utility mark_bad_fiff under /neuro/bin/util/. Eg:
/neuro/bin/util/mark_bad_fiff 632 741 2211 <filename.fif>
Maxfilter will identify the bad channels you marked in the fiff file, and will not use them in calculations. If you don't want a permanent mark in the files, then use -bad option in maxfilter itself.