Hello, I am Muskaan from National Brain Research Center, working on my masters dissertation.
I am working on fMRI data where I have 2 sessions, one is resting state scan and another is scanned while participant is reading sentences, I want to have sentence>resting contrast with resting as baseline.
But I am having problem during first level analysis when I am trying to make GLM, it’s not showing anything, as resting scan has TR = 2s and task(sentence reading) has TR = 0.4s.
How to downsample to the sentence scan, or any other method to solve the problem?
I would recommend not trying to contrast across runs. When you have different TRs, your scan parameters are going to be sufficiently different as to be a massive confound, regardless of whether you match sampling rates and perform any other operations to statistically align the runs.
I assume that the subject was not reading constantly, so you’ll have some periods of “rest” within the reading task, and you can construct a regressor that looks something like
___|‾‾‾‾‾|____|‾‾‾‾‾|____
If you fit that regressor to your BOLD series, that beta will already be your task > baseline contrast. Now, you may have several conditions such as “read words” and “read non-words”. In which case you might get two regressors like
___|‾‾‾‾‾|______________
_____________|‾‾‾‾‾|____
Now, you’ll get two beta estmates words > baseline and non-words > baseline. To get reading > baseline, you create a contrast of these two betas with weights [1/2, 1/2]. To get words > non-words you would use weights [1, -1].
Hope this helps get you started in the right direction.