New preprint...and this one is truly EPIC šØ. Using a large group of patients with depression and healthy controls (N>800), we show differences in the functional segregation of insular subnetworks. And we can use it to classify! Led by @glassybrain.bsky.social #neuroskyence š©ŗ osf.io/preprints/ps...
Wow - incredible work!ššš
Great work. It would be interesting to know if you get similar results (as in MDD) for healthy controls if you ask them to think sad thoughts, think of sad memories,or listen to sad music etc.
Thank you. It is a great idea to estimate the state component in healthy participants. In our fMRI studies, we usually collect mood ratings before and after the scan, but we typically manipulate mood only indirectly. Will keep this in mind (and maybe someone has good data on it).
I say this because when Drevets showed abnormal activation in the subgenual cortex in depression, my memory (and it may be wrong) is that someone showed you could get the same results by asking healthy people to think sad thoughts.
John Teasdale worked a long time ago on manipulating mood in healthy people by playing some slowed up music. This was not in fMRI. But my point is: are the results on MDD simply due to the fact that they are thinking sad thoughts? Of course they may not be, but it would be worth ruling it out.
Yeah, I agree. There are many mood induction studies, but to the best of my knowledge, I am not aware of a similar analysis, so it is definitely worth looking into this.
this is so cool! I wanted to try exactly this kind of analysis since the postdoc days. Well done!
Thanks! I don't want to say how long it took us to finish the paper š« . It would be nice to see how well it replicates across samples as well.
We should definitely try this in the VMP, Iām sure we can benefit from your experience!
Definitely need to take a look, did you parcellate based on gyri??
We use (normalized) voxels and the Hammers atlas, not individual vertices. There are concerns about spurious correlations in surface-based analysis, so ours is comparatively straightforward (using the default CONN pipeline). direct.mit.edu/imag/article...