Dear Urs,
This is a design decision of Karl who thought that physiological
features of ERP cannot be shorter than that. There is something called
'matched filter theorem' that suggests that your time resolution
should match the extent of the features you are interested in. So
although you have very high sampling rate, the brain doesn't get any
faster because of that.
Vladimir
On Wed, Jul 20, 2011 at 10:53 PM, Urs Bachofner <[log in to unmask]> wrote:
> Dear Vladimir,
>
> Thank you a lot for the dozenth time.
>
> Yes, since I know the exact milisecond of the peak I'm interested in I will
> rewrite all of my Nifti Images in the way you told me. (Thankfully this is a
> step after the Inversion so it will not take a lot of time)
>
> Thanks for clearing that up - still remains the question: If the time span
> of the Nifti images is non-existant, why is there an 8 ms time span there?
> Is it because I am spoiled to work with a sampling rate of 1000/s?
>
> Best,
>
> Urs
>
>
>
>
> Am 21.07.2011 00:05, schrieb Vladimir Litvak:
>>
>> Dear Urs,
>>
>> You are right, sorry. I meant the 'Window' button. If you specify the
>> same time point twice you'll get the shortest possible time window
>> which is Gaussian with FHWM of 8 ms. But usually activations are not
>> that short so perhaps you should define a longer window around your
>> peak of interest. Also note that if you want to compare several time
>> windows in your stats you should specify them all at once (separated
>> by semicolon).
>>
>> Best,
>>
>> Vladimir
>>
>> On Wed, Jul 20, 2011 at 10:36 PM, Urs Bachofner<[log in to unmask]>
>> wrote:
>>>
>>> Dear Vladimir,
>>>
>>> I think if I agree with you we'd both be wrong! :)
>>>
>>> I'm asked to specify the time span and frequency I'm interested in when
>>> pressing the "Window" button. When I press the "Image" button, I'm not
>>> asked
>>> to specify anything and it just writes the Nifti Image (and I updated to
>>> the
>>> newest versions some weeks ago).
>>> I'm lucky to know exactly where the peaks i'm interested in are since I
>>> use
>>> a matlab script that displays me the exact time point of the peak of my
>>> interest.
>>> So did I get you correctly that you advice me to enter that exact instant
>>> of
>>> time after pressing the "Window" button (and leave the frequency band of
>>> interest to zero, default) to get the Nifti image at the time point of
>>> interest?
>>> But then again, I think after pressing the Window button it clearly
>>> demands
>>> two time points, which I thought was the span of one trial.
>>>
>>> Looking forward to your answer!
>>>
>>> Thank you a lot, again,
>>>
>>> Urs
>>>
>>>
>>> Am 20.07.2011 23:20, schrieb Vladimir Litvak:
>>>>
>>>> Dear Urs,
>>>>
>>>> When you press the Image button you are asked to specify time and
>>>> frequency window and the MIP image that you see then is the summary
>>>> over that time and frequency window and it will correspond to the
>>>> exported image (if there are several conditions or time windows only
>>>> one will be shown, but all will be exported). There is another MIP
>>>> that you get just after the inversion or when you press the MIP button
>>>> without specifying time or coordinates. This MIP corresponds to the
>>>> time of maximal activation which may or may not be what you are
>>>> interested in. But this is just done to display something and you are
>>>> always asked to specify explicitly the parameters for image
>>>> generation.
>>>>
>>>> Best,
>>>>
>>>> Vladimir
>>>>
>>>> On Wed, Jul 20, 2011 at 9:47 PM, Urs Bachofner<[log in to unmask]>
>>>> wrote:
>>>>>
>>>>> Dear Vladimir,
>>>>>
>>>>> thank you very much for your patient answer.
>>>>>
>>>>> I will have a look at the sections you pointed me to tomorrow again.
>>>>> I had the feeling that by generating the Nifti Image the whole length
>>>>> of
>>>>> the
>>>>> Trials would be involved. But basically did I understand you correctly
>>>>> that:
>>>>>
>>>>> - When I press the Image button to generate the Nifti Image, the
>>>>> MIP-picture
>>>>> should be already in the time span which I'm interested in (which is
>>>>> usually
>>>>> indeed the maximum activation in MY data)? So that the generated Nifti
>>>>> File
>>>>> shows the activation within the milisecond I'm interested in, so the
>>>>> dimension of time will be obsolete in the file?
>>>>>
>>>>> Thank you a lot again,
>>>>>
>>>>> Urs
>>>>>
>>>>>
>>>>> Am 20.07.2011 20:05, schrieb Vladimir Litvak:
>>>>>>
>>>>>> Dear Urs,
>>>>>>
>>>>>> You sound quite confused indeed. Perhaps you should take a look at the
>>>>>> SPM EEG paper again (http://www.hindawi.com/journals/cin/2011/852961/)
>>>>>> particularly the parts about sensor-level statistics (Section 3,
>>>>>> Figure 2) and section 4.6. "Convert to images" menu has to do with
>>>>>> sensor-level statistics and what you are trying to do is source-level.
>>>>>> Analyze (.img + .hdr) and NIFTI (.nii) formats are very similar and
>>>>>> are used interchangeably in SPM. Either can be used for statistics. It
>>>>>> so happens that the sensor-level images are created in Analyze format
>>>>>> and source-level images in NIFTI. So the bottom line is, use source
>>>>>> images such as the one you sent me for statistics. You don't need to
>>>>>> do anything else.
>>>>>>
>>>>>> Regarding the time dimension, SPM does not support statistics for more
>>>>>> than 3 dimensions so it would not be possible to further analyze
>>>>>> images with both 3D space in time. You can only do analysis on both
>>>>>> time and space only at the sensor level because scalp maps can be
>>>>>> reduced to 2D. So you must summarize the time and frequency dimensions
>>>>>> as explained in the paper. If you want to look at multiple time
>>>>>> windows you can generate multiple images at the same time, but the
>>>>>> time windows will be at least ~8ms.
>>>>>>
>>>>>> Best,
>>>>>>
>>>>>> Vladimir
>>>>>>
>>>>>> On Wed, Jul 20, 2011 at 1:23 PM, Urs Bachofner<[log in to unmask]>
>>>>>> wrote:
>>>>>>>
>>>>>>> Hi Vladimir,
>>>>>>>
>>>>>>> sorry to bother you again with an amateur's question but I would
>>>>>>> appreciate
>>>>>>> it if you could point me into the right direction.
>>>>>>>
>>>>>>> I'm still working with my EEG Files (induced) that are now converted
>>>>>>> into
>>>>>>> Nifti Files (by clicking the Image button in the Source
>>>>>>> Reconstruction
>>>>>>> menu). However I now read that for statistics you should choose
>>>>>>> "Convert
>>>>>>> to
>>>>>>> Images" in the drop down menu and after I did this to the .mat file I
>>>>>>> didnt
>>>>>>> get one Nifti File but two files for every good trial, a .hdr and a
>>>>>>> .img.
>>>>>>>
>>>>>>> Which of the files (the .nii or the .hdr/.img pair) am I supposed to
>>>>>>> use
>>>>>>> for
>>>>>>> statistical analysis?
>>>>>>>
>>>>>>> If I should use the .hdr/.img files, how can I get SPM to produce one
>>>>>>> averaged pair (induced) instead a pair for all good trials?
>>>>>>>
>>>>>>> If I should use the .nii file: Why is there no time scale in the
>>>>>>> image
>>>>>>> as
>>>>>>> I
>>>>>>> display it? I have trials of 600ms and would like to see the
>>>>>>> activation
>>>>>>> in
>>>>>>> each time point?
>>>>>>>
>>>>>>> (There is one nifti Image in the attachment)
>>>>>>>
>>>>>>>
>>>>>>> Thanks a lot for every bit of help
>>>>>>>
>>>>>>> Urs
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> --
>>>>>>> NEU: FreePhone - 0ct/min Handyspartarif mit Geld-zurück-Garantie!
>>>>>>> Jetzt informieren: http://www.gmx.net/de/go/freephone
>>>
>
>
|