I have the following problem: we conducted an event-related fMRI experiment using Presentation software to display the experiment. Usually each fMRI pulse/scan is logged in the presentation log file so the times for event-related analysis can be derived later on. Unfortunately, for about half of the subjects these triggers in the log files are missing, so I don't know when the experimental events occurred relative to the first scan.

Do you have any ideas how I might still try to analyze that part of the data? All suggestions are welcome.

  • $\begingroup$ I am stricken by your positive enthusiasm after finding such a nightmarish gap in your data set! Good luck. +1 $\endgroup$ – AliceD Aug 31 '16 at 9:56
  • $\begingroup$ Thank you! To be precise, I did not start the project, I just happen to be the only person left working with fMRI in our group, so I kind of "inherited" the project and discovered this problem... $\endgroup$ – Natalie Sep 1 '16 at 7:37
  • $\begingroup$ Pheww, Good for you! That likely turns it from a disaster into a bummer :) $\endgroup$ – AliceD Sep 1 '16 at 7:42

I can see one case in which you still might analyze the data.

This only works under three conditions:

  1. your paradigm has no variable durations of either stimulus presentations or breaks or others (or you can determine all the times retrospectively) AND

  2. all your fmri scans have timestamps in their header information (which I think is always the case, at least in the raw file) AND

  3. the pulse sequences and the presentation were always started at the same time (or the logfile has at least a starting time and the difference between both starts was consistent).

I guess you can answer immediately whether the first condition is met. For the second one, any appropriate software for opening the raw files should be able to show you the timestamps if they exist. For dicom files there are timestamps for study time, series time and acquisition time where the acquisition time is different for every single file (according to this help site from the AFNI program). The third one you might also immediately know but if not you could look into the timestamps of the subjects where you do have the triggers in the logfile and compare it with the starting time of the logfile (if that exists). If you calculate that difference for all subjects and it's always the same you are lucky and can calculate the trigger times for the subjects where they are missing.

  • $\begingroup$ Thank you very much for this detailed answer! I've also thought in a similar direction. The log file has all stimulus presentation times and durations relative to the start of the paradigm and the fMRI scans also have time stamps in their header information. The third condition is the real problem, though: the scanner and the paradigm were run from two different machines, thus a certain time on one machine does not correspond to the same time on the other machine (afaik neither of the machines is connected to the internet, so there's no time sync). $\endgroup$ – Natalie Sep 5 '16 at 8:32

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.