Running F-tests (or similar) in FSL-MRS

Hi everyone!

Does anyone have experience with running F-tests in FSL-MRS? I currently have fMRS data, which I obtained before and during an infusion of medication (~40 min). I divided this into 9 regressors: 1x baseline, 1x bolus, 7x infusion (see example design matrix below).


To run the first and second level, I would like to determine the change from baseline for each of the bolus and infusion regressors for the first-level contrast and then do an F-test for each of these contrasts to determine if the change is significantly non-zero. However, I realised I am not sure if this is possible (or the correct way to approach this) in FSL-MRS. Does anyone know if this is possible and, if not, does anyone have suggestions on alternative approaches?
Many thanks!


Hi @dboucherie,

I realised I didn’t expose the flameo interface for f-tests in the fmrs_stats tool. I’ll do that in an update (and address your other request too).

Regarding the design, do I understand right that there is an initial (rapidly administered) bolus of the medication and then continuous infusion? Are the blocks just splitting the infusion up, or are they different in any way?

Hi @wclarke,

Great to hear! You are absolutely correct about the design. The 7 infusion blocks are indeed just used to split the very long infusion up.

Hi @dboucherie , Sorry for the delay 2.1.13 is now out which implements f-tests for the group level (first level is a WIP). You can provide a f-test vector/matrix with the --hl-ftests command line argument. As with the other inputs this needs to be in the VEST format.