Mixed factorial design (Nilearn)

Dear NeuroStars community,

I’m using Nilearn to compare brain activation in responders vs non-responders patients across two sessions (T1, before treatment vs T2, after treatment). My design is a 2×2 mixed factorial design:

  • Between-subject factor: Group (responders vs non-responders)
  • Within-subject factor: Session (T1 vs T2)

What is the best way to implement this design using Nilearn?

Thank you!

Best,

Chlöé

very quick answer: because Nilearn does not handle non-sphericity correction at the group level, in general I recommend making the contrast you are interested in at the subject level first (T1 vs T2 for example) and then bring the results of those contrasts only at the group level (either in a one sample or 2 sample t-test).

let me know it, if you need more details