search

wiheto / esfmri_connectivity

Apache License 2.0
2 stars 1 forks source link

Stim site properties #32

Closed wiheto closed 4 years ago

wiheto commented 4 years ago

[PROC] Using pre-op resting-state data, we will calculate the P of stimulation sites and community partitions. IF we conclude that there is a problem with making contrasts between the pre-op and post-op resting-state data (e.g., due to different levels of noise), we will report the following two analyses: (a) once excluded the problematic subjects; (b) with P of the stimulation sites calculated on an external dataset using healthy controls.

wiheto commented 4 years ago

Note: Considering https://www.biorxiv.org/content/10.1101/659508v3 we need to consider how to relate the communities from rest to task when there are different parcellations and community vectors.

If the stimsite is included in the parcellatoin, this is not a problem. Otherwise we are going to have to use the preop community partition, the template partition (but only works for cortical sites) or infer the stimsite community from the preop community partition but still use the postop community partition.

wiheto commented 4 years ago

It was decided instead in #39 that it was best to use the community detection. Many runs still had the most overlapping parcel close to the stimulation site to be marked as "good". Since this was possible, and considering the problems above regarding inferring properties from other datasets, it was decided it was best to use the es/post op dataset to find the stimulation site and the topographic properties. This is detailed in #33 as it is a deviation from the preregistration.