Dear Stefanie and Yury,
Thank you for your emails. At this point I would like to respond to your
concerns from the methodological point of view. If more detailed
information about MaxFilter software is needed, we may still have to
wait until Jukka Nenonen returns to the office next week.
In principle, the head position transformation can be done separately
after tSSS. The command should be of the form "maxfilter -f
<data_tsss.fif> -trans default -force". The "force" option is needed
because otherwise the program refuses to proceed as it discovers from
the file that the data have already been processed by SSS or tSSS. If
this does not help, could you please give us the exact error messages so
that we can find out the reason of the problem.
The head position transformation may alter the apparent noise level and
this can be understood by the fact that the signal-to-noise ratio (SNR)
cannot be virtually increased by transforming the position closer to the
sensors than in the actual measurement. Consider, for example, a case
where during the measurement the head is in a much lower position than
normally giving data with poor SNR because of the large distance from
the brain to the sensors. The brain signal amplitudes can be
artificially increased by virtually transforming the head to an upper
position but then we will also increase the noise level at least in some
parts of the sensor array. That is, we cannot improve the original SNR
just by virtually transforming the head closer to the sensors. It should
be noted that the noise considered here is random sensor noise and
should be distinguished from external interference.
Typically, this noise effect is not very significant when transforming
positions over reasonably small distances, as in standardization of the
head positions among different subjects or measurement sessions. In
addition, tSSS is less sensitive than SSS to these effects. The noise
can also be damped by giving less weight to the spatial features of the
magnetic field that were not measured from the original head position.
We are currently investigating an optimal way to do the weighting and
this will also affect future versions of MaxFilter.
With best regards,
Samu
Yury Shtyrov wrote:
> Hi Stefanie -
>
> What we heard previously from ENM is that there is a bug in the
> maxfilter utility - perhaps this is preventing it from doing what you
> want. Also, I suspect that running tsss may automatically remove all
> the information about the head movements as it contaminates the
> signal, so a later movement correction may not be possible.
>
> We have not noticed the problem with noise after movement correction
> (although we have not had much experience with this so far), but this
> sounds serious - I am afraid we have to wait until one of the
> developers gets back to us about this when the holidays season in
> Helsinki is over ;-(
>
> cheers,
> yury
>
> 2007/7/26, Hassel, Stefanie <[log in to unmask]>:
>
>> Hi Yury,
>>
>> Thanks for getting back on this … yes, I think you are right it would be
>> best to do artefact correction and head transform in one step.
>>
>> The problem is twofold:
>>
>> 1) I already have a number of datasets where I only applied tsss. As it is
>> time-consuming (it seems to run a fair while at least on our machine), I had
>> hoped to do the transformation separately on the already maxfiltered data.
>>
>>
>>
>> 2) - this I find far more concerning: for a recent subject I applied both
>> steps in one go, but the output file was still quite noisy. So, I run tsss
>> without head-transform and the data was a lot better … that I found odd and
>> wanted to compare it to the output I would get from running head transform
>> on an already maxfiltered dataset.
>>
>> I would need the –nosss option (it does not work when input file is already
>> maxfiltered) plus the –trans option and that somehow does not work. The
>> error messages are from can't read input file to command not known …
>>
>>
>>
>> I hope this makes sense. I wonder have you noticed difference in the
>> "noise-level" of your data when running tsss + head transform vs tsss alone?
>>
>>
>>
>> stefanie
>>
>>
>>
>> Stefanie Hassel, PhD
>>
>> Functional Neuroimaging in Emotional Disorders,
>>
>> Department of Psychiatry,
>>
>> Western Psychiatric Institute and Clinic,
>>
>> University of Pittsburgh
>>
>> Loeffler Building,
>>
>> 121, Meyran Avenue,
>>
>> Pittsburgh. PA 15213.
>>
>>
>>
>> Phone: (+1) 412-383-8193
>>
>> Fax: (+1) 412-383-8335
>>
>>
>>
>>
>>
>>
>> -----Original Message-----
>> From: [log in to unmask] [mailto:[log in to unmask]] On
>> Behalf Of Yury Shtyrov
>> Sent: Thursday, July 26, 2007 8:56 AM
>> To: Hassel, Stefanie
>> Subject: Re: maxfilter: head transform after tsss
>>
>>
>>
>> Wouldn't it be better to run it on the original raw data along with
>>
>> tsss extension? Otherwise, if it attempts to apply sss for the second
>>
>> time, it is probably not good. What is the program's feedback? It
>>
>> probably sees the file has been sss-ed and stops there. You could
>>
>> probably try it with nosss option, but then it will probably refuse to
>>
>> do the transform.
>>
>>
>>
>> yury
>>
>>
>>
>> 2007/7/25, Hassel, Stefanie <[log in to unmask]>:
>>
>>
>>> dear all,
>>>
>>> i wondered if it is possible - using maxfilter - to do the Move --> head
>>>
>>> transform (to default) after having already applied tsss.
>>>
>>> i have tried running it separately from the command line
>>>
>>> maxfilter -trans <data_tsss.fif> | default
>>>
>>> but it does not seem to work.
>>>
>>> any advice would be very much appreciated.
>>>
>>> many thanks,
>>>
>>> stefanie
>>>
--
-----------------------------------------------------------
Samu Taulu
Elekta Neuromag Oy
Street address: Elimäenkatu 22, Helsinki, Finland
Mailing address: P.O. Box 68, FIN-00511 HELSINKI, Finland
Tel: +358 9 756 240 83 (direct), +358 9 756 2400 (operator)
Fax: +358 9 756 240 11
E-mail: [log in to unmask]
|