Print

Print

This is occurring at the subject level, specifically during the second level analysis that combines individual within-subject runs.

And yes I may have initially included null EVs in the f-contrasts, and am now going to try only including existing EVs in within-run f-tests.

Will I then be able to combine contrasts across runs (within-subject), i.e. a contrast in one run that contains a given EV that may not exist in the second run, or vice versa?

Thanks,

Binyam

 

From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf Of Matthew Webster
Sent: Monday, August 13, 2018 5:37 AM
To: [log in to unmask]
Subject: Re: [FSL] modelling null or missing EVs in a feat fsf file

 

Hello,

  Is this problem occurring during the subject-level or higher-level processing? Do any of the f-contrasts at subject-level involve these null EVs?

 

Kind Regards

Matthew

--------------------------------
Dr Matthew Webster
FMRIB Centre 
John Radcliffe Hospital
University of Oxford



On 8 Aug 2018, at 22:09, Binyam Nardos <[log in to unmask]> wrote:

 

Hello,

I am using the HCP pipeline’s task fMRI scripts to process task fMRI data in CIFTI format.

My task is loosely structured as such:

                BOLD 1: contains EV1, EV2, EV3

                BOLD 2: contains EV1, EV3, EV4

                BOLD 3: contains: EV2, EV5, EV6

 

As you can see, not every EV is present in every run. My initial approach when creating the fsf files was to model existing EVs in a given run using the 3 column file input format, and model EVs that don’t exist in a particular run (e.g. EV6 in BOLD 1) as a null EV using the “Empty (all zeros) option. Using this setup, I am unable to create copes nor conduct ftests. I get an error during level 2 of the task fMRI processing script saying some of the contrasts are not linearly independent (or are a linear sum of other contrasts). 

Can you please advise whether all EVs (1 to 6 in the above case) need to be modeled in the design matrix/fsf to enable subsequent second level analyses across runs then subjects etc? If so, what is the appropriate way to model an EV that is not present in a given run?

 

Thank you.,

Binyam

 

Binyam Nardos, PhD
Oregon Health and Science University
Postdoctoral Researcher, Fair Neuroimaging Lab

Mackenzie Hall 2198
3181 SW Sam Jackson Park Road L470
Portland, Oregon 97239

503-418-1895 (o)
314-315-1510 (c)

<image001.png>

YES! Youth Engaged in Science Initiative

NW NOGGIN: Neuroscience Outreach Group (growing in networks)

Resource Council member

 

 

To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1

 

 

To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1


To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1