You can localise power but the source inversion step should come before computing power in your pipeline.
Vladimir
On 20 Apr 2015, at 17:47, "Rivolta, Davide" <[log in to unmask]> wrote:
> Dear Vladmir,
>
> many thanks for your email. I am glad I did not make mistakes so far.
> However, from what you said concerning source localization, I should only use epoched time-domain data (ERFs?). Thus, I should NOT be able to localise, for instance, gamma band power using SPM when contrasting two gorups (controls vs. patients)?
>
> Apologies if I misunderstood.
>
> Bests,
> Davide
> ________________________________________
> From: Vladimir Litvak [[log in to unmask]]
> Sent: Monday, 20 April 2015 5:46 PM
> To: Rivolta, Davide
> Cc: [log in to unmask]
> Subject: Re: [SPM] MEG: FieldTrip to SPM
>
> Dear Davide,
>
> SPM behaved correctly in this case because 33 Hz is the sampling rate of your time-frequency data. However, it is not possible to localise sensor-level power directly because computing power is a non-linear transformation of the original data. For source localisation you should use epoched time domain data. You can find an example of how to do source localisation in SPM in the tutorial chapter on multimodal face data (see p. 413 in the manual).
>
> Best,
>
> Vladimir
>
> On Mon, Apr 20, 2015 at 2:59 PM, Rivolta, Davide <[log in to unmask]<mailto:[log in to unmask]>> wrote:
> Dear all,
>
> I am approaching SPM12 (version: 20140806) for the first time.
> I have CTF MEG data that I have analysed (using multitapers) at the sensor-level with Fieldtrip (FT). I now wish to run source statistics with SPM. I hope you can help me.
>
> The first thing I should do is to convert. As such, I am using the spm_eeg_ft2SPM function.
>
> The data I have (as after FT processing) has the following fields:
> - Label
> - Dimord
> - Freq
> - Time
> - Powspctrm
> - cumtapcnt
> - Hdr
>
> Hdr has the following fields:
> - nChans
> - label
> - Fs (600 hz)
> - nSamples
> - nSamplesPre
> - nTrials
> - grad
>
> Hdr.grad has the following fields:
> - Pnt
> - Ori
> - Tra
> - Label
> - Unit (cm)
> - Balance
>
> When I use the batch to run spm_eeg_ft2spm on these data:
> D = spm_eeg_ft2spm(DATA, outputfile)
>
> I get the warning:
> Data type is missing or incorrect, assigning default
> Warning: could not determine channel type from CTF header
>
> It does not give an error, and the resulting D has these fields:
> Type: continuous
> Transform: TF
> 1 conditions
> 265 channels
> 31 frequencies
> 111 samples/trial
> 1 trials
> Sampling frequency: 33.3333 Hz
>
> Few questions:
> 1. I am not sure why my sampling rate, which originally was 600 Hz, has been changed to 33.3333 Hz.
> 2. May you suggest a reference (tutorial?) I should follow to proceed without (many) problems?
>
> Any advice would be greatly appreciated.
>
> Many thanks and best regards,
> Davide
>
|