|
|
Post by Ivan Alekseichuk on Jun 2, 2021 18:55:21 GMT
Neurocircuitry-Guided TMS in Psychiatry: Personalization, Precision & Clinical Response
R Cash, L Cocchi, J Lv, P Fitzgerald, A Zalesky
University of Melbourne
Contact: robin.cash@unimelb.edu.au
|
|
|
|
Post by Max on Jun 3, 2021 18:37:59 GMT
Hi Robin, thanks for you presentation, really interesting results! I was wondering if, by any chance, you have also applied your approach to surface-based data, i.e. the CIFTI data provided by the HCP. Best, Max
|
|
|
|
Post by Robin on Jun 4, 2021 3:30:00 GMT
Hi Max,
Thanks!!
Yes we did try that - but the results didn't seem to be better - although this could be tested and developed further.
I think there are a few complexities with surface space - for example:
1) ideally the search space within surface space should be first restricted to the proximity of gyral crowns.
2) the distance between targets will appear to be artificially inflated, purely because surface space spreads the cortex out into a sheet. This will give the impression of reduced reproducibility.
3) It seems to make more sense to mimic a decaying TMS e-field (e.g. via the weighted sphere) in volume space compared to surface space - although it is possible.
4) For clinical/research implementation (and future adjustments), I was thinking that volume space might be more user friendly.
Do you have any particular thoughts on this. It is interesting because there are a few advantages of moving towards surface space and the imaging field more generally seems to be doing so.
|
|
|
|
Post by Robin on Jun 4, 2021 3:32:31 GMT
Also sorry for the delayed reply (the conference was overnight in Australian time).
|
|
|
|
Post by Max on Jun 4, 2021 17:41:18 GMT
Thanks for your answer, Robin! Interesting that it actually didn't seem to be better. And I see your points, there are definitely some more (or different) challenges when working with surface data. But there might also be some advantages, especially in relation to surface-based E-field simulations. I might contact you via email to further discuss this.
|
|
|
|
Post by Desmond Oathes on Jun 4, 2021 19:29:55 GMT
Great work, Robin. Any chance you've done a similar analysis for distance to POSITIVE FC peak with sgACC? Working on some papers now suggesting positive FC spots do even better in engaging sgACC (negative BOLD fMRI) and do well clinically in MDD and PTSD.
|
|
|
|
Post by Robin on Jun 5, 2021 5:06:41 GMT
Hi Desmond,
That's interesting and surprising - looking forward to hearing more when it's out.
We have done something related - switching the code to look for sites of positive FC, not for SGC-FC, but for coordinates derived from a different RSN. We can share the code if you like (just email me: robin.cash@unimelb.edu.au).
|
|
|
|
Post by Robin on Jun 5, 2021 5:12:59 GMT
Hi Max,
Yes - there are likely to be some advantages and it is worth exploring further. We could provide the code and maybe collaborate if you like.
We decided not to go further in part because we considered that (i) in terms of reproducibility, 2mm variation/single voxel is hard to improve upon and (ii) we were seeing a solid relation to clinical outcome in the retrospective data analysis. I guess we were wondering whether we were getting to a point of diminishing returns. Nonetheless, you are right and it is worth exploring.
|
|
