Hi Doug,
Thanks for your response. I've been trying to build the sequence with your advice but I am still stuck.
For the study we want to replicate, we will use TR=2s and we have 3x stimulus type (N=20) presented for 1s each. We gave a window of 1s for participant to respond. On top of this, we add jittered ITI (M= 1000ms; Range 750-1250ms, randomized psychological jitter to prevent participant to anticipate the trial). To deconvolve the MRI signal we want to have null blocks (length=2s per block) that are scattered in the sequence.
I used the tnullmin tnullmax to build in the psychological ITI but then realize this was not a good idea b/c tnullmin and max has to be integer multiple of the dPSD. So the only way I thought of achieving the above is to remove the psychological jitter, and include the response time window into tEv and write the below line. I thought I will add the randomized psychological jitter after the sequence is built.
optseq2 --ntp 80 --tr 2 --psdwin 0 12 2 --ev ConA 1 20 --ev ConB 1 20 --ev ConC 1 20 --evc 1 -1 0 --focb nCB1Opt --nkeep 3 --nsearch 100000 --o MRIsequence
While the sequence is written fine, I suppose I no longer have an optimum sequence b/c my tEv is not correctly defined. Also the additional psychological jitter will change the assumption upon which the optimum sequence is built. Would really appreciate your thoughts on this.
Best, Steve
Hi Steve, a couple of points:
1. Don't get too stuck on the sequence being "optimal". When you go to analyze, there will be inevitable differences between how you optimized and how you analyze. For one, it is optimized assuming an FIR model. If you assume a shape to the HRF, then it no longer optimum. It will still be good and the FIR assures that you are statistically robust.
2. Your PSD window probably needs to be more than 12sec. For a 2sec stimulus, I would use 24sec.
3. The tNullMin and Max need to be integer multiples of the dPSD, not necessarily the TR. You can set the dPSD to anything that is an integer divisor of the TR (eg, 1 sec), though if you set it too low you will get problems.
4. Alternatively, you can set the TR to 1 and ntp to 160.
doug
On 8/16/13 6:42 AM, Lukito, Steve wrote:
Hi Doug,
Thanks for your response. I've been trying to build the sequence with your advice but I am still stuck.
For the study we want to replicate, we will use TR=2s and we have 3x stimulus type (N=20) presented for 1s each. We gave a window of 1s for participant to respond. On top of this, we add jittered ITI (M= 1000ms; Range 750-1250ms, randomized psychological jitter to prevent participant to anticipate the trial). To deconvolve the MRI signal we want to have null blocks (length=2s per block) that are scattered in the sequence.
I used the tnullmin tnullmax to build in the psychological ITI but then realize this was not a good idea b/c tnullmin and max has to be integer multiple of the dPSD. So the only way I thought of achieving the above is to remove the psychological jitter, and include the response time window into tEv and write the below line. I thought I will add the randomized psychological jitter after the sequence is built.
optseq2 --ntp 80 --tr 2 --psdwin 0 12 2 --ev ConA 1 20 --ev ConB 1 20 --ev ConC 1 20 --evc 1 -1 0 --focb nCB1Opt --nkeep 3 --nsearch 100000 --o MRIsequence
While the sequence is written fine, I suppose I no longer have an optimum sequence b/c my tEv is not correctly defined. Also the additional psychological jitter will change the assumption upon which the optimum sequence is built. Would really appreciate your thoughts on this.
Best,
Steve
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
freesurfer@nmr.mgh.harvard.edu
-
Douglas Greve -
Lukito, Steve