Dear Sandro,
Sorry for the delay, but the FSL Course prevented me from looking into this data sooner.
The timings in the text file look sensible to me.
So I suspect that either there is a problem in some of the other files, or that the matlab output is not comparable.
The latter could easily happen if you were assuming that the order of the regressors would be the same (as they probably are not) and what really counts is whether the signals removed from the data with the different sets of regressors (looking at each set, not individual regressors) is similar or not. Note that subtle differences in filtering and other things might also mean that the regressors appear different yet span very similar spaces when it comes to regressing out signals from the data. Therefore it is probably worth investigating whether the PNM output is actually correcting for the physiological noise well or not, rather than getting too fixed on comparing with another version of RETROICOR/PNM.
I hope this helps.
All the best,
Mark
On 27 Mar 2014, at 17:02, Sandro Nunes <[log in to unmask]> wrote:
> Hello,
>
> I am sorry, when I attached the .txt file, the values in it got pasted into the body of the message instead of being sent as a file. Here's a link for the file: https://www.dropbox.com/s/uzyq9f0r356g8mt/PNM_slice_timings.txt.
>
> The matrix is, in fact, 123x138 and I checked in notepad++ for any possible wrong formation and it seems to be fine.
>
> I also compared the values of the cardiac regressors for the first slice of each volume (in which slice timing correction should have no effect ) output by the PNM and the regressors obtained by using the matlab RETROICOR script provided in here http://cbi.nyu.edu/software/ and they don't match, which makes me think that the error is probably on how the cardiac data is being input. I tried inputting in 2 different ways:
>
> - Two columns, the first with the value of each time sample (in seconds) and the second with the corresponding phase (computed using the matlab script): https://www.dropbox.com/s/9jp61v53w5agb1e/PNM_card_acq.txt
> - Two columns, the first with the value of the time sample of each TR (in seconds) and the second with the corresponding phase (computed using the matlab script): https://www.dropbox.com/s/ykyhw3hjqdqh620/PNM_card_TR.txt
>
> I then input the data to pnm_evs as follows:
>
> pnm_evs --slicetiming=../PNM_slice_timings.txt --oc=2 --or=0 --tr=2.5 -i ../fmri_data.nii -o ../evs -c ../PNM_card.txt
>
> I obtained similar results when trying to compute respiratory regressors in addition to cardiac regressors.
>
> Again, thank you for your help. If there is something important I did not mention or if something is not clear, please let me know.
>
> Best regards,
>
> Sandro Nunes
|