Hey Statisticians!

I have a bit of a conundrum.

I have a 2x3 model with factors Drug (E0,E1) and Music (No Music 1, Music, No Music 2). I have 12 subjects, so 72 individual values per metric.

The music factor is somewhat unbalanced due to repeating the No Music factor, but not the music factor.

However, these two No Music scans do clearly show different effects on several fMRI functional connectivity graph metrics, and our working hypothesis is that music has a sustained effect on brain activity while under the drug, but not under placebo.

Am I allowed to treat this as a repeated measures (No music before, Music, No Music after), or does it have to be a LME? Am I then allowed to conduct paired t-tests on the contrasts to explain the significant interaction in the repeated measures? I've already ran these tests and found significant effects, but since the LME does not show significant interactions, I can't really explain them using that approach.

Thanks in advance.