Hi Ella,

On 6/16/13 9:27 AM, Ella Hinton wrote:

Hi all,

I have been advised to use optseq to aid my event related fMRI design, specifically to help calculate stimulus order and timing. I've read the help in some detail but am still a little unclear on some of the appropriate values to input and indeed what values I am aiming for in the output. It'd be great if anyone can help out...

Briefly, my study is a simple food cue reactivity study in which three different types of pictures (50 each) are presented for 3s each, in pseudo-random sequence. I now intend to include null events to optimise the spacing of the events to detect a difference between conditions, but also to provide a baseline for a 'sanity check' just in case.

I am uncertain as to the value of psdwin and associated variables (PSD, min, max dPSD) - should I include a post-stimulus delay or should that be set to 0 as in most examples?

You always have to have non-zero post-stimulus delay window. The PSD starts at PSDmin and goes to PSDmax by steps of dPSD. The min is usually set to 0 (stimulus onset), unless you want to create FIR plots starting prior to stimulus onset (eg, to show where the baseline is or make a prettier plot). There should be no response prior to stimulus onset.

Also what is the recommended psdwinMAX value for events of 3s? 12 or as much as 18 seconds to capture the whole haemodynamic response?

I would set it to 21 or even 24.

Also I am not certain how to judge the resulting output. I understand I should maximise 'efficiency' but I can't find any advice as to what is a 'good' value of efficiency to aim for. Any advice here? Also should I be judging the sequence on any other values?

There is no real way to judge it just from this information. What you are really asking about depends on how big your effect is and is it answered by a power analysis. If you have no effect, the an infinite efficiency is not good enough to show an effect. Having said that, my rule of thumb is that the variance reduction factor (VRF) be 20-40 when summed over all runs, but it is just a rule of thumb. You will really need to pilot the study to see how many stimuli you need.

So I have inputted the following and got a value of 2.586 for efficiency for the best sequence. Is this reasonable/sufficient?

optseq2 --ntp 200 --tr 3 ---psdwin 0 12 3 --ev ev1 3 50 --ev ev2 3 50 --ev ev3 3 50 --nkeep 3 --o test6 --nsearch 10000

I would add --focb 100
doug

Many thanks in advance,
Ella

_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer