[color=#000000]Dear Lisa,[/color]
[color=#000000]The R2R clustering procedure uses a functional-similarity measure defined as the euclidean distance between the connectivity patterns of different seeds and averaged across the [b]included conditions[/b] and [b]included subjects[/b] in each specific 2nd-level analysis/test (i.e. data from conditions and subjects that are not part of the 2nd-level analysis will not be used by the clustering algorithm). For example, if you have two groups, looking at the results within each group separately (e.g. one-sample t-test) will result in different ROI orders and clusters. You can always then export/import those orders and clusters from one set of analyses to another (e.g. if you prefer all of your second-level analyses to be using the clusters that resulted from a particular/reference analysis)[/color]
[color=#000000]Hope this helps[/color]
[color=#000000]Alfonso[/color]
[i]Originally posted by Lisa MP:[/i][quote]Dear Conn developpers,
First thanks for all your hard work on the Conn toolbox, it is a great tool!
I have a question about the clustering of ROIs in the Spatial Pairwise Clustering method.
If the clustering is based on functional connectivity similarity (as it is by default with a weight of 0.95) is this functional similarity computed using all the subjects in the Conn project or only the ones selected for the test?
I am actually doing tests on a subset of subjects in the Conn project who were scanned at one specific center and would like the ROIs to be clustered using only these subjects data.
Similarly, when you have multiple timepoints is the clustering based on the average bold timeseries between the different timepoints?
I have, finally, one last question : is functional similarity defined as the average connectivity value on all the population?
Thanks a lot for your help,
Best regards,
Lisa[/quote]
[color=#000000]The R2R clustering procedure uses a functional-similarity measure defined as the euclidean distance between the connectivity patterns of different seeds and averaged across the [b]included conditions[/b] and [b]included subjects[/b] in each specific 2nd-level analysis/test (i.e. data from conditions and subjects that are not part of the 2nd-level analysis will not be used by the clustering algorithm). For example, if you have two groups, looking at the results within each group separately (e.g. one-sample t-test) will result in different ROI orders and clusters. You can always then export/import those orders and clusters from one set of analyses to another (e.g. if you prefer all of your second-level analyses to be using the clusters that resulted from a particular/reference analysis)[/color]
[color=#000000]Hope this helps[/color]
[color=#000000]Alfonso[/color]
[i]Originally posted by Lisa MP:[/i][quote]Dear Conn developpers,
First thanks for all your hard work on the Conn toolbox, it is a great tool!
I have a question about the clustering of ROIs in the Spatial Pairwise Clustering method.
If the clustering is based on functional connectivity similarity (as it is by default with a weight of 0.95) is this functional similarity computed using all the subjects in the Conn project or only the ones selected for the test?
I am actually doing tests on a subset of subjects in the Conn project who were scanned at one specific center and would like the ROIs to be clustered using only these subjects data.
Similarly, when you have multiple timepoints is the clustering based on the average bold timeseries between the different timepoints?
I have, finally, one last question : is functional similarity defined as the average connectivity value on all the population?
Thanks a lot for your help,
Best regards,
Lisa[/quote]