Different numbers of volumes between participants and runs

1

I have fMRI BOLD series for 50+ subjects, each with between 141 and 155 volumes. Each participant has 3 runs, which are usually the same number of volumes but sometimes do differ. Is this a conceptual or practical issue for analysis in SPM (or elsewhere)?

2

Hi @ZCLFreeman,

Without knowing more about your analysis goals it is hard to say one way or another.

Best,
Steven

3

Sorry Steven, that wasn’t very informative of me!

This is an event-related task design where participants perform 3 runs of the same task, across three scanning sites. The aim is to see if there are differences between the two types of stimuli people see in the scanner in a within-participant analysis. We have the onsets and durations for from the playlists created and used for each run already, which are all of the same length, but the number of volumes are different in one scanning site.

I’m concerned that a) it’s hard to tell whether volume 1 is always related to the start of the playlist file, and b) whether (if ‘a’ is resolvable) these participants can be grouped together given the different number of volumes.

I hope that’s a little clearer, thanks for your patience.

4

Hi @ZCLFreeman,

Do all the data occupy the same time length, and its just different TR values across sites leading to the volume count mismatch? Were the across-site sequences harmonized in any specific ways? Were the same presentation software (PsychoPy, Psychtoolbox, etc) used across the sites? Same scanner vendor?

Best,
Steven

5

Thanks Steven, my basic understanding was that volumes indicate time length but I will look to see whether there is another marker of this. The across-site sequences were harmonised from what I’ve been told but I’m not familiar with what to look for to check if this is the case - one scanner was GE and the other two Philips, and the presentation software was the same for all sites.

6

Hi @ZCLFreeman,

The total time is number of volumes * TR. I have only used Siemens so I am not as familiar with how GE and Phillips trigger presentation software, but I can only imagine that it should trigger properly.

I would confirm that, in the site with lower volumes, no events were left off, based on the total scan time and onsets/durations of events. I would check that the number of volumes per stimuli type is consistent across sites, such that the variance/confidence in beta estimates should be consistent. I would check that within site you get the expected activations from the dummy contrasts for the events, which would indicate the onsets/durations line up with what is expected.

Best,
Steven

7

Thank you Steven I will look into these, much appreciated!