Hi Alfonso.
I am working with the ICA method implemented in CONN and I have some doubts.
Regarding the function incorporated in CONN to employ the subject-specific spatial-component maps resulting from ICA as a seed:
Does this method allow to evaluate the differences in functional connectivity within the network (evaluate differences within the mask/IC used as seed) or only how the rest of the brain is connected to the IC (between-network connectivity analyses)?
Excerpt of the CONN manual with the mentioned method:
3) weighted/unthresholded group-level maps: in Setup.ROIs select 'ROI tools. Add ICA-network ROIs' to define a new set of ROIs directly form the group-level spatial masks, using these maps as weighted/probabilistic-ROIs (one ROI per network).
Another doubt related to this procedure: I'm trying to find other papers that have used the same procedure and I'm having problems to find them, basically because I don't know very well which criteria i have to use to find them. Is this procedure known as Hybrid ICA-Seed-Based? Is this procedure shown in this article ( https://doi.org/10.1155/2010/868976 ) ? If not, would you please suggest any work that employed this method?
Finally, a question related to the "import values" function at the results viewer of the spatial properties section at the ICA network second level (using the standard method implemented in CONN, Calhoun's group-ICA 2001 approach and GICA3 for back-projection). Do the values obtained through this function correspond to the z-Fisher-transformed values (just as in the seed to voxel method)? I guess so, but I would like to be sure since I obtain values that are quite high compared to the values I used to obtain in the seed to voxel mode (I attach a graph of the effect size for the ICA results with the supposedly high values)
thank you very much in advance
best regards
Benxa
I am working with the ICA method implemented in CONN and I have some doubts.
Regarding the function incorporated in CONN to employ the subject-specific spatial-component maps resulting from ICA as a seed:
Does this method allow to evaluate the differences in functional connectivity within the network (evaluate differences within the mask/IC used as seed) or only how the rest of the brain is connected to the IC (between-network connectivity analyses)?
Excerpt of the CONN manual with the mentioned method:
3) weighted/unthresholded group-level maps: in Setup.ROIs select 'ROI tools. Add ICA-network ROIs' to define a new set of ROIs directly form the group-level spatial masks, using these maps as weighted/probabilistic-ROIs (one ROI per network).
Another doubt related to this procedure: I'm trying to find other papers that have used the same procedure and I'm having problems to find them, basically because I don't know very well which criteria i have to use to find them. Is this procedure known as Hybrid ICA-Seed-Based? Is this procedure shown in this article ( https://doi.org/10.1155/2010/868976 ) ? If not, would you please suggest any work that employed this method?
Finally, a question related to the "import values" function at the results viewer of the spatial properties section at the ICA network second level (using the standard method implemented in CONN, Calhoun's group-ICA 2001 approach and GICA3 for back-projection). Do the values obtained through this function correspond to the z-Fisher-transformed values (just as in the seed to voxel method)? I guess so, but I would like to be sure since I obtain values that are quite high compared to the values I used to obtain in the seed to voxel mode (I attach a graph of the effect size for the ICA results with the supposedly high values)
thank you very much in advance
best regards
Benxa
