 |
 |
Thanks Vladimir,
I have taken a longer time window (0-300 ms) but the bizarre
reconstruction did not disappear.
I have inverted the two original conditions as you suggested, and this
time it disappeared, bingo!
Now I am trying to batch all the source reconstruction analysis and I
have a question:
How can I perform a second inversion using the same forward model
without writing on the first one? I thought to put the value 2 in the
inversion index, but it does not work, I guess because there is only
one forward model.
Thanks,
Marco
On 7 November 2012 20:58, Vladimir Litvak <[log in to unmask]> wrote:
> Dear Marco,
>
> The source reconstruction algorithm in SPM relies on certain assumptions
> that I suspect your data violates. As a result your data looks much like
> noise for the algorithm. I would suggest you to invert your two original
> conditions together rather than the difference and then export two images
> per subject and put them in paired t-test design. Furthermore, 20 ms is a
> very short window, depending on your sampling rate it can contain just a few
> samples. You should perhaps invert a much longer window especially if the
> peak you are interested in is quite prominent and then focus on your short
> window when summarizing your result as an image. Perhaps look at the
> multimodal face evoked responses tutorial in the manual and use that as the
> basis for your analysis.
>
> Regarding noise covariance, in the Bayesian framework noise is whatever is
> not explained by your model (the residual) and it the essence of the
> inversion procedure to figure out what part of the data is noise and what is
> not.
>
> Best,
>
> Vladimir
>
>
>
> On Wed, Nov 7, 2012 at 3:25 PM, Marco Buiatti <[log in to unmask]>
> wrote:
>>
>> Dear SPM masters,
>>
>> I am performing source reconstruction on the difference between two
>> ERPs using group inversion (20 subjects), standard reconstruction
>> parameters on a narrow temporal window (20 ms) on a well-known ERP
>> wave (N170). Standard MNI brain.
>>
>> I have two questions:
>> 1) Why do I have spurious reconstructions like the one attached
>> (source reconstruction on one subject, but this appears to be
>> relatively frequent in my analyses)? Is there a way to avoid such
>> artifacts?
>> 2) A more general question. As far as I know, algorithms of
>> reconstruction of distributed sources use a noise covariance matrix
>> either on a baseline or on a separate set of data where no task
>> -related activation occurs. How does it work with SPM? Is this
>> computed implicitly somewhere or does it work differently? Since I am
>> computing a difference between ERPs, I expect results independent from
>> baseline correction. Am I right or does it influence the noise
>> covariance matrix? I could not find such information in the SPM
>> manual.
>>
>> Thanks for your feedback,
>>
>> Marco
>>
>> --
>> Marco Buiatti, PhD
>>
>> CEA/DSV/I2BM / NeuroSpin
>> INSERM U992 - Cognitive Neuroimaging Unit
>> Bāt 145 - Point Courrier 156
>> Gif sur Yvette F-91191 FRANCE
>> Ph: +33(0)169.08.65.21
>> Fax: +33(0)169.08.79.73
>> E-mail: [log in to unmask]
>> http://www.unicog.org/pm/pmwiki.php/Main/MarcoBuiatti
>>
>> ***********************************************
>
>
--
Marco Buiatti, PhD
CEA/DSV/I2BM / NeuroSpin
INSERM U992 - Cognitive Neuroimaging Unit
Bāt 145 - Point Courrier 156
Gif sur Yvette F-91191 FRANCE
Ph: +33(0)169.08.65.21
Fax: +33(0)169.08.79.73
E-mail: [log in to unmask]
http://www.unicog.org/pm/pmwiki.php/Main/MarcoBuiatti
***********************************************