Eye tracker

[mohammad abuzari]

Hello friends,I have a question about using an eye tracker dataset. First of all, has anyone worked with this dataset? Secondly, for instance, for Movie 1, I need 921 seconds of data, but in the eye tracker dataset, I have 1015 seconds. Therefore, I don't know how to extract my functional 921 seconds. Could anyone help me with this? Let me know if you need any further assistance!

[Keith Jamison]

There are events embedded in the eyetracking data referring to the start and end of the scan, and for most datasets an event for each 1-second TR as well. 

I previously did some conversion and some minimal preprocessing/cleaning/QA on the movie eyetracking data, which you can find in this folder: https://wcm.app.box.com/folder/104164180881?s=vobtvkhf1l5l436n4pquy0ysl8uukxsp

Each file contains the eyetracker data for all available subjects for a given movie (ranging 161-163), downsampled to 24fps, including saccade, fixation, and blink masks (which were computed by eyelink).The data have been truncated to the stop/start of the scan.

So in python you might do:

M=loadmat('eyetracker_24fps_MOVIE1_7T_AP_allsubj_data.mat', simplify_cells=True)
subj0_xy=M['gazecoods'][:,:,0] #22104x2 array of (x,y) pixel coords for subject 0 (921*24=22104)
subj0_id=M['subjects'][0] #subjects[0] for this movie is '100610'

One detail that could be relevant is some scans were run with the eyetracker configured for a screen width of 1024 and some for 1280. That info is in the 'dispsize' field in the mat file. If you don't take it into account your gaze model will be inconsistent.

Let me know if you have any questions.

-Keith

[Mohammad Ebrahim Katebi]

Dear Dr. Jamison,

Although the ID_7T_MOV$_eyetrack_summary.csv file contains rough information about the eyetracking–fMRI delay, I have not been able to identify the events embedded in the EyeLink data lines that can be used to synchronize the gaze data with the scan TRs.

Would it be possible to reshare the preprocessed movie eyetracking dataset mentioned here? It is a valuable contribution; however, the current link does not work for me and returns a “permission required” error.

Thank you in advance for your time and help.

Best regards,
Ebrahim