help
help > RE: Graph-theory in conn
Jul 13, 2012 08:07 AM | Alfonso Nieto-Castanon - Boston University
RE: Graph-theory in conn
Hi Igor
The statistical test comparing the global efficiency of the entire network across your two subject groups is reported in the top-right list (in the row labeled as 'network'); in addition, the global efficiency of the entire network can be represented as the average of the global efficiency of each node of the network (global efficiency of a node in a graph is defined as the average inverse shortest-path distance from this node to all other nodes in the graph). So yes, in addition to comparing between groups the global efficiency of the 'whole-brain' network, you can also compare between groups the global efficiency of individual nodes (ROIs) within this network (this is sometimes thought of as a measure of centrality or relative importance of a given node within a graph); the rest of the result rows shown in the top-right list correspond to the statistical test on the global efficiency measure for each of the nodes/ROIs (only shown those nodes that survive the chosen false positive threshold; the labels of these nodes are displayed in the bottom-right list). In your example attached it would seem that the increased global efficiency (of the entire brain network) in groupA (from the positive beta value and significant p-value in the 'network' row) could be perhaps attributed to an increased efficiency/centrality within the network of the particular ROIs shown in the results display (from their positive beta values and significant FDR-corrected p-values in the corresponding rows for each ROI). Let me know if this clarifies
Best
Alfonso
Originally posted by Igor M:
The statistical test comparing the global efficiency of the entire network across your two subject groups is reported in the top-right list (in the row labeled as 'network'); in addition, the global efficiency of the entire network can be represented as the average of the global efficiency of each node of the network (global efficiency of a node in a graph is defined as the average inverse shortest-path distance from this node to all other nodes in the graph). So yes, in addition to comparing between groups the global efficiency of the 'whole-brain' network, you can also compare between groups the global efficiency of individual nodes (ROIs) within this network (this is sometimes thought of as a measure of centrality or relative importance of a given node within a graph); the rest of the result rows shown in the top-right list correspond to the statistical test on the global efficiency measure for each of the nodes/ROIs (only shown those nodes that survive the chosen false positive threshold; the labels of these nodes are displayed in the bottom-right list). In your example attached it would seem that the increased global efficiency (of the entire brain network) in groupA (from the positive beta value and significant p-value in the 'network' row) could be perhaps attributed to an increased efficiency/centrality within the network of the particular ROIs shown in the results display (from their positive beta values and significant FDR-corrected p-values in the corresponding rows for each ROI). Let me know if this clarifies
Best
Alfonso
Originally posted by Igor M:
Thanks, Alfonso!
Using pre-formatted AAL ROI files has been a real time saver.
The analysis seems to work fine and I can retrieve all measures (local efficiency, global efficiency, betweenness, so on) from the *.network file.
My question now is about interpreting the group contrast ( [1 -1] ) results.
In other programs (e.g. FSL) this contrast would highlight areas of the cortex where a measure is greater in "group A" than in "group B".
I ran this contrast and several nodes became highlighted in the graph-theory window. My question is this: what metric is being compared at those nodes? "GlobalEfficiency" is selected as the "Analysis measure" on the left side of the window, but GlobalEfficiency pertains to an entire network, not a single node. In other words, I didn't think that it could be compared between groups at single nodes.
In the "network nodes" tab "network of all ROIs" is selected. A screenshot is attached to this post.
IM
Using pre-formatted AAL ROI files has been a real time saver.
The analysis seems to work fine and I can retrieve all measures (local efficiency, global efficiency, betweenness, so on) from the *.network file.
My question now is about interpreting the group contrast ( [1 -1] ) results.
In other programs (e.g. FSL) this contrast would highlight areas of the cortex where a measure is greater in "group A" than in "group B".
I ran this contrast and several nodes became highlighted in the graph-theory window. My question is this: what metric is being compared at those nodes? "GlobalEfficiency" is selected as the "Analysis measure" on the left side of the window, but GlobalEfficiency pertains to an entire network, not a single node. In other words, I didn't think that it could be compared between groups at single nodes.
In the "network nodes" tab "network of all ROIs" is selected. A screenshot is attached to this post.
IM
Threaded View
| Title | Author | Date |
|---|---|---|
| Igor M | Jul 9, 2012 | |
| Igor M | Jul 9, 2012 | |
| Alfonso Nieto-Castanon | Jul 10, 2012 | |
| Igor M | Jul 10, 2012 | |
| Alfonso Nieto-Castanon | Jul 11, 2012 | |
| Saman Sarraf | Jan 21, 2014 | |
| Igor M | Jul 12, 2012 | |
| Alfonso Nieto-Castanon | Jul 13, 2012 | |
| Igor M | Jul 21, 2012 | |
| Alfonso Nieto-Castanon | Jul 21, 2012 | |