I think a 6 second video could reasonably still be considered “event related”. If you model the HRF across several duration values, you will find that it does not plateau until events of about 20 seconds. While you specified you are not interested in modelling the HRF shape per se, its still a reasonably good model of how you expect signal to change, and I would argue our knowledge of the linear changes in BOLD signal over time suggests that a 6 second video would be no worse than a 4 second video. behaviorally, it should also be well within most people’s attention spans to maintain vigilance for six seconds.
I think option 2 is the best.
While people may often think of event-related as 0-4 secs duration, and blocked at 20, 30 or more, I think it is important to not get caught up in such distinctions. The important questions are, will your design enable you to test what you want? Can you model it in a reasonable way? I think the answer is clearly that we can adequately model a 6 second event, we know what the BOLD response should look like, and it will probably enable you to answer your questions of interest without the need to start changing your videos.
However, all options you suggests are viable.
Best of luck,
Colin Hawco, PhD
Neuranalysis Consulting
Neuroimaging analysis and consultation
www.neuranalysis.com
[log in to unmask]
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf Of Chirag Limbachia
Sent: July-09-17 11:05 PM
To: [log in to unmask]
Subject: Re: [FSL] optimal length of block/event (using video stimuli)
Hi Anna,
I personally prefer event-related design as it produces comparable results if not as good as or better than block designs. Using jittered event-related designs may yield less power (as per lit), but they produce results that are not confounded by habituation or prediction as you already mentioned. More importantly, it allows you to sort responses to each trial post hoc if you are using eye-tracking, button press or something like that.
Anyhow, the order of my preference to the options you suggested is 3, 1, and 2 if editing the video is not going to hamper the goal of your research. Option 2 is worth an experiment. Option 1 is the safest as it will definitely yield results which will have to be caveated with the downsides of using a traditional block design that you already mentioned.
Thank you,
Chirag
Sent from my iPhone
On Jul 5, 2017, at 9:27 AM, Anna van 't Veer <[log in to unmask]> wrote:
Hi Chirag and other fsl users,
Thanks so much for your thorough reply. We have some follow-up questions.
Our goal is to find a difference in the overall response to experimental vs control videos (we are not interested in estimating shapes for individual videos per se). We could go with several options, which would be the most powerful?
1. we could compile several videos to make a block length of 18 seconds (so three videos), or 24, or 30 but we would like to stay on the short side in order to prevent habituation. This would be like a traditional block design that generally has greater power than an event-related design (although we still would get habituation because the experimental and control videos are not presented randomly, thus creating predictability).
2. we could present the videos individually, this could be seen as a short 6 second block or a long event, however as we are not interested in estimating the shape of the HRF response, it may not be a problem that our trials are somewhat longer. And by presenting the videos this way we could randomise their presentation and add jitter, adding to power.
3. we could edit out videos to be 4 seconds, and make it an event-related design (again with randomisation and jitter).
We would love to hear from anybody who has thoughts on this (does anybody have experience with video stimuli lasting 4 or 6 seconds?),
Greatly appreciate it,
Thanks,
Dr. Anna van 't Veer
\ Child and Family Studies
\ Leiden Institute for Brain and Cognition (LIBC)
\ Social and Behavioural Sciences
\ Leiden University
On 29 Jun 2017, at 18:59, Chirag Limbachia <[log in to unmask]> wrote:
Hi Anna,
For block design, I would suggest you to keep the epoch lengths at least 20 secs long. In your case, 30 secs will be best as 6*5 = 30 sec. The epoch length for a block design is usually 20-60 secs. Also, to get highest optimality for your block design, mask sure you distribute your task epochs and rest epochs evenly throughout the fMRI run. Best way to do this is to keep 30 secs for task and 30 secs for rest (or a different task).
For event-related design, it is advisable to have trails (events) no more than 4 secs because longer trial duration give a block design like effect to your event-related design. Your randomized ITIs look fine. Also, for highest optimality of event-related design, make sure your task and rest duration is evenly distributed for the entire length of the event-related experiment.
To make a choice between an event-related and block design really depends on how you plan to assess activation: by 1) estimating the overall amplitude of the hemodynamic response to the task (rise of BOLD signal from baseline) or by 2) estimating the shape of the hemodynamic response to every individual trail?
For 1), block is more appropriate.
For 2), event-related is more appropriate.
Please feel free to ask any question you may have.
Thank you
Chirag
On Wed, Jun 28, 2017 at 9:34 AM, Anna van 't Veer <[log in to unmask]> wrote:
Dear FSL users,
Could you kindly advise me on a question regarding the optimal length of epochs in our task design?
We would like to design our task using (experimental and control) videos that have a length of 6 seconds.
For block designs I can find documentation about optimal block lengths of 15-20 seconds, and for event related designs lengths seem to vary between 2 and 4 seconds. Our videos seem to fit in neither category.
Would you (A) advise to present our videos—comparable to an event related design—as individual trials of 6 seconds (with truncated exponential ITI’s with minimum, maximum and average ITI of 3, 8 and 4.5 respectively), or (B) would you advise to present several videos of 6 seconds together, comparable to a block design (if yes, how many videos in one block)?
Very curious to hear your thoughts!
Cheers,
Dr. Anna van 't Veer
\ Child and Family Studies
\ Leiden Institute for Brain and Cognition (LIBC)
\ Social and Behavioural Sciences
\ Leiden University
