We haven't had this in MF 2, I believe, it normally finds too many bad
channels. New version - new bugs? ;)
yury
2009/10/3 Tony W. Wilson <[log in to unmask]>:
> Hi Everyone,
> Sorry to beat this topic to death, but I noticed a potential bug in MF 2.1
> when performing the "dummy" autobad runs. No matter what one sets the
> badlimit at, it never finds a bad channel (according to the log). I have
> tried various values from 30 to 0.1 and never had a bad channel occurrence.
> I have used several different raw.fif files, each about 5 mins at 1kHz.
> Anyone else have this problem?
> Maybe my data is just that good! (hehe)
> Thanks
> Tony
>
> On Fri, Oct 2, 2009 at 7:58 AM, Daniel Wakeman
> <[log in to unmask]> wrote:
>>
>> Hi Tony,
>>
>> The current version of MNE i.e. 2.6.0 will allow you to browse raw
>> maxshield data see section 4.2.2 for more details (--allowmaxshield).
>>
>> Dan
>>
>> Tony W. Wilson wrote:
>>>
>>> Matti and Yury,
>>> Thanks again for the input. I believe the 'dummy' MF pass is the way to
>>> go. I cannot open/browse the raw file in ENM or MNE without some pass
>>> through MF because we use active shielding (smartshield) for all
>>> acquisitions. Our room is only a 1-layer and our environment is quite
>>> noisy. On a related note, is there any data (or opinions) on the degree to
>>> which bad channels affects MF performance (normal or tsss)? I presume that
>>> including only one or two clearly bad channels would affect the accuracy of
>>> the MF results at least moderately, but maybe not.
>>>
>>> Tony
>>>
>>> On Fri, Oct 2, 2009 at 6:22 AM, Yury Shtyrov
>>> <[log in to unmask] <mailto:[log in to unmask]>>
>>> wrote:
>>>
>>> I think the new MF 2.1 should recognise MNE marking as well as ENM
>>> one (at least it was promised at some point) , but I have not had a
>>> chance to test this.
>>>
>>> In mark_bad_fiff you just specify a list of channels that you don't
>>> like, it does not do any detection itself. You can use mne_browse_raw
>>> or anything else (that can show raw data) to look through the channels
>>> to see which ones are bad, or do a 'dummy' pass of plain MF without
>>> tsss and see which ones are detected by autobad.
>>>
>>> y.
>>>
>>>
>>> 2009/10/2 Matti Hamalainen <[log in to unmask]
>>> <mailto:[log in to unmask]>>:
>>> >
>>> > Hi Tony,
>>> >does not involve any criteria. It just marks channels bad
>>> > irrespective of whether they really are bad or not.
>>> > As a side not, the MNE software employs a different way to mark
>>> channels bad
>>> > in a fif file. MNE for sure does not recognize the bad channel
>>> markings made
>>> > with mark_bad_fiff and I think Neuromag software does not
>>> recognize the bad
>>> > channels indicated by mne_mark_bad_channels.
>>> > - Matti
>>> > On Oct 1, 2009, at 3:08 PM, Tony W. Wilson wrote:
>>> >
>>> > Thanks Yury.
>>> > Do you know how mark_bad_fiff defines a bad channel? Is it the
>>> same
>>> > criteria described in the manual for the autobad feature?
>>> > Tony
>>> >
>>> > On Thu, Oct 1, 2009 at 9:47 AM, Yury Shtyrov
>>> > <[log in to unmask]
>>> <mailto:[log in to unmask]>> wrote:
>>> >>
>>> >> This is certainly something that many people are doing in the
>>> previous
>>> >> version of MF here, and I can't see why you wouldn't do it in MF
>>> 2.1.
>>> >> It's indeed either mark_bad_fiff or by entering them as bad
>>> channels
>>> >> using MF command line options.
>>> >>
>>> >> yury
>>> >>
>>> >> 2009/10/1 Tony W. Wilson <[log in to unmask]
>>> <mailto:[log in to unmask]>>:
>>> >> > I have been confused about whether Maxfilter 2.1 is excluding
>>> bad
>>> >> > channels
>>> >> > prior to computing the tsss correction. I understand from
>>> watching the
>>> >> > program and reading the manual (Oct 2008 revision) that tsss
>>> switches
>>> >> > off
>>> >> > the automated bad channel detection, but does detect and exclude
>>> >> > saturated
>>> >> > channels and static bad channels from the computation. To me,
>>> it seems
>>> >> > there could be additional channels one would want to exclude.
>>> For
>>> >> > example,
>>> >> > sensors that were noisy in a particular run, or on that day,
>>> but were
>>> >> > not
>>> >> > excluded during acquisition (due to an oversight or whatever).
>>> To
>>> >> > ensure
>>> >> > such channels are excluded, I'm guessing one needs to run
>>> mark_bad_fiff
>>> >> > on
>>> >> > each raw file prior to tsss. Is my understanding correct? Is
>>> anyone
>>> >> > else
>>> >> > doing this (ie., mark_bad_fiff, then tsss)?
>>> >> > All the best,
>>> >> > Tony
>>> >> > ___________________________________________________
>>> >>
>>> >>
>>> >> Yury Shtyrov, Dr.Phil., Prof.
>>> >> Senior Scientist (PLT)
>>> >> Manager, MEG Laboratory
>>> >> Medical Research Council (MRC)
>>> >> Cognition and Brain Sciences Unit
>>> >> 15 Chaucer Rd, CB2 7EF
>>> >> Cambridge, United Kingdom
>>> >> tel +44 1223 273703 (office)
>>> >> tel +44 1223 355294 (reception), ext 832
>>> >> fax +44 1223 359062
>>> >> e-mail [log in to unmask]
>>> <mailto:[log in to unmask]>
>>> >> http://www.mrc-cbu.cam.ac.uk/~yury
>>> >
>>> >
>>> >
>>> >
>>> > ---------
>>> > Matti Hamalainen, Ph.D.
>>> > Athinoula A. Martinos Center for Biomedical Imaging
>>> > Massachusetts General Hospital
>>> > [log in to unmask] <mailto:[log in to unmask]>
>>>
>>>
>>
>>
>> --
>> BEGIN:VCARD
>> VERSION:3.0
>> N:Wakeman;Daniel;Gary;;
>> FN:Daniel Gary Wakeman
>> ORG:University of Cambridge;
>> TITLE:Student 2010
>> EMAIL;type=INTERNET;type=WORK;type=pref:[log in to unmask]
>> EMAIL;type=INTERNET;type=HOME:[log in to unmask]
>> item1.EMAIL;type=INTERNET:[log in to unmask]
>> item1.X-ABLabel:_$!<Other>!$_
>> TEL;type=WORK;type=pref:+44 (0) 1223 355 294 ex. 595
>> TEL;type=CELL:+44 (0) 7877 043 797
>> item2.TEL:+1 518-291-4347
>> item2.X-ABLabel:US
>> TEL;type=WORK;type=FAX:+44 (0) 1223 359 062
>> item3.ADR;type=HOME;type=pref:;;Wolfson College\nBarton
>> Road;Cambridge;Cambridgshire;CB3 9BB;United Kingdom
>> item3.X-ABADR:us
>> item4.X-AIM;type=pref:[log in to unmask]
>> item4.X-ABLabel:Video
>> X-AIM;type=HOME:WakerCrzy0
>> X-ABUID:F98E7710-43F5-4D9A-992F-E4FD6FFE5B6B\:ABPerson
>> END:VCARD
>
>
|