help
help > RE: ICA Spatial Match to template?
Jul 5, 2022 11:07 AM | Federico Zanella
RE: ICA Spatial Match to template?
Originally posted by Alfonso Nieto-Castanon:
thank you for your time and the provided information.
Is the r value stored in a variable? Is it possible to derive it?
Thank you very much
Federico
Hi Kevin and
Federico
The spatial match to template in CONN computes two measures of overlap /similarity between an arbitrary ROI and the ICA spatial map associated with each individual network component: a) the spatial correlation measure is a simple Pearson correlation coefficient across all voxels comparing the ICA spatial map scores at each voxel to the values from an apriori ROI (typically 0/1 values characterizing a simple deterministic ROI); and b) the spatial overlap measure computes the dice coefficient across all voxels comparing the thresholded ICA spatial map scores at each voxel (thresholded at the value chosen in the GUI) to the values from an apriori ROI (thresholded at the same threshold value if the ROI is defined using continuous values, or at the maximum value if the ROI is defined using 0/1 or other dichotomous values)
Regarding the choice of a priori ROIs, by default CONN will use its own network ROIs (defined in conn/rois/networks.nii) characterizing common networks, but you can also simply select in the GUI any alternative choice of a priori ROIs to be used by the measures above for comparison with your ICA results
Hope this helps
Alfonso
Originally posted by Federico Zanella:
Hi Alfonso, The spatial match to template in CONN computes two measures of overlap /similarity between an arbitrary ROI and the ICA spatial map associated with each individual network component: a) the spatial correlation measure is a simple Pearson correlation coefficient across all voxels comparing the ICA spatial map scores at each voxel to the values from an apriori ROI (typically 0/1 values characterizing a simple deterministic ROI); and b) the spatial overlap measure computes the dice coefficient across all voxels comparing the thresholded ICA spatial map scores at each voxel (thresholded at the value chosen in the GUI) to the values from an apriori ROI (thresholded at the same threshold value if the ROI is defined using continuous values, or at the maximum value if the ROI is defined using 0/1 or other dichotomous values)
Regarding the choice of a priori ROIs, by default CONN will use its own network ROIs (defined in conn/rois/networks.nii) characterizing common networks, but you can also simply select in the GUI any alternative choice of a priori ROIs to be used by the measures above for comparison with your ICA results
Hope this helps
Alfonso
Originally posted by Federico Zanella:
Originally posted by Kevin T:
Hi,
I am in the same situation. Never found this info and now the rev 2 asked to be more specific. Have you ever found an answer?
many thanks,
Federico
Hi Conn Experts
A quick question that have been asked before but I've found no substantial answers.
I'm wondering what was used to compute the spatial match to template in the ICA tools (ie SPM toolbox? like FSLnets)?
Also what template was used? Also are there any other templates (ie with more than 8 networks)? Thank you. I'm trying to reference this work.
Kevin
A quick question that have been asked before but I've found no substantial answers.
I'm wondering what was used to compute the spatial match to template in the ICA tools (ie SPM toolbox? like FSLnets)?
Also what template was used? Also are there any other templates (ie with more than 8 networks)? Thank you. I'm trying to reference this work.
Kevin
Hi,
I am in the same situation. Never found this info and now the rev 2 asked to be more specific. Have you ever found an answer?
many thanks,
Federico
thank you for your time and the provided information.
Is the r value stored in a variable? Is it possible to derive it?
Thank you very much
Federico
Threaded View
| Title | Author | Date |
|---|---|---|
| Kevin T | May 17, 2021 | |
| Federico Zanella | Jul 1, 2022 | |
| Alfonso Nieto-Castanon | Jul 5, 2022 | |
| Federico Zanella | Jul 5, 2022 | |