 |
 |
[log in to unmask]" type="cite"> I apologize, my original message to the spm list serve was not included. I attached those files to represent the differences between the 128 and 400 filter. What I attached was the "explore design" for the 400 filter for the 3rd scanning session high painful stimulus regressor, which is why you see the regressor onset 10 minutes in. The first stimulus occurs 50 seconds into the scan and lasts for 40 seconds, alternating with rest. The participant experiences four scanning sessions (A, B, C, D) with each session having 6 blocks of intermixed 3 high and 3 low painful stimuli with 40 seconds of rest between painful stimuli. My initial concern was that the default filter of 128 was removing much of the variance (see Design attachments). My concern now is if I am interpreting the explore design right for the 128 filter, and if adjusted as suggested with the "doubled longest mean interval difference" (e.g., 400seconds) am I doing it right? What is the right way to handle if the default high-pass filter in SPM is removing variance as the SPM8 manual states (p.69): "The frequency domain graph is useful for checking that experimental variance is not removed by high-pass filtering. The grayed out section of the frequency plot shows those frequencies which are removed"
Here was my original question with a description of the design.Dear SPM,This is my first go at an fMRI analysis and I need your expert help interpreting whether or not the high-pass filter default(128s) in SPM8 is removing too much variance from the active regressors? It appears the frequencies are spiking in the gray shaded area (or below the .008 hz threshold), which indicate frequencies removed by the high pass filter. I have attached for you 3 examples that have been consistent across subjects. Is it recommended to adjust the high-pass filter at this point? If, so what is the correct and most appropriate way of adjusting this high-pass filter?
This is a block design fMRI study. It encompasses 4 sessions of 6 blocks of stimulus. Blocks are intermixed low painful and high painful stimulus. The block design is 40 seconds active and 40 seconds rest. The participants are only receiving these two different painful physical stimuli. They are not performing any type of task. I have attached the explore design output for session 1 high and low active regressors and a second session high active regressor for one participant from the level 1 model specification.
Much appreciation!Jess
On Wed, Aug 22, 2012 at 2:57 PM, Chris Watson <[log in to unmask]> wrote:
The first stimulus for that participant was almost 10 minutes into the scan?
Also, you mention that your blocks are 40s on - 40s off; I don't understand why you would need an hpf of 400 in that case. Can you explain the design in more detail?
On 08/22/2012 03:44 PM, Jessica Wojtalik wrote:Hi Gabor,I apologize for the delayed response, wanted to make sure I had understood your response which I greatly appreciate. Just to followup, the longest mean interval between subsequent onsets is 200. The onsets for this high stimulus regressor for this particular participant are 587.5 seconds, 827.5 seconds, 987.5 seconds (240+160)/2=200. Therefore, a new high-pass filter of 400 is appropriate then? How acceptable is it in the field to adjust the high-pass filter in SPM for block designs with blocks with longer task times (e.g., 40 seconds on and 40 off in my case)? Should an adjusted high-pass filter become a concern for publications? Are there publications out there providing support to adjust the high-pass filter for longer block periods as you have suggested?Thank you so much again for your helpful response,
Jess
On Tue, Aug 14, 2012 at 2:21 PM, Gabor Oederland <[log in to unmask]> wrote:
Hello Jess,
this is to the Nyquist sampling theorem, see http://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem . http://spm.martinpyka.de/?p=51 might also give you an idea what happens if the high-pass filter is "too short".
You can determine an appropriate high-pass filter by calculating the mean interval between subsequent onsets of one regressor. For "HighA" in Design_1_high-1 the onsets seem to be approximately 40 seconds, 160 seconds, and 240 seconds (in case your TR = 2s). The mean difference of subsequent trials is then (120 + 80)/2 = 100 seconds. For "LowA", the onsets seem to correspond to 100 seconds, 300 seconds and 360 seconds, so you get (200 + 60)/2 = 120 seconds. For your high-pass filter, enter a value that is at least double the size of the mean difference, that is 240 seconds in this case, or maybe e.g. three times the size which would be 360 seconds then. The high-pass filter should be the same for different subjects, so if there are different onsets for different subjects calculate the mean intervals for all of them and take the "longest" .
You might also want to take a filter which is twice as large as the longest interval between trials of the same condition. This would be 400 seconds in your case. At least I have already read both options in papers. Also have a look at an older thread, which is a collection of various postings https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind06&L=spm&D=0&1=spm&9=A&J=on&K=4&X=35D0701792D24A3126&Y=oederland%40gmx.ch&d=No+Match%3BMatch%3BMatches&z=4&P=5990336
One problem might be that you pick up noise depending on whether you suffer from scanner drifts or not.
Hope this helps,
Gabor