Hello DSI-Studio colleagues,
I was running into a potential issue and wanted to see if others in the group had thoughts about best next steps for troubleshooting.
In brief, my group reconstructed ~90 subjects with QSDR using all the default/standard settings. We completed some group connectometry using a recent version of DSI-studio (July 13th build). We opened the group connectometry database, loaded variables of interest (just one independent variable) and ran analyses as shown in different tutorials (q<.1, t=2.5, length=20; we bumped up the permutations and the pruning to 6000 and 6 but everything was fairly standard).
There's a few tracts that emerged (see the attached map) and then I attempted to isolate significant tracts for analysis "outside" of DSI-studio (and this is where I may be doing something incorrect). I loaded the .tt.gz results, went to (Tracts Misc) -> (Recognize and Cluster), and then output individual tract files of the results. My hope was to use those tracts analogous to an ROI in functional imaging. I then extracted individual subject values for each tract (using QA and NQA) and attempted to "replicate" the results I find from the group connectometry analysis of interest.
Interestingly, the relations drop a good bit (when data is extracted and analyzed in R). Outside of DSI-studio, the associations have a t-range of 1-1.9; however, the connectometry analyses in DSI-studio had a t-threshold of 2.5 (so I was expecting the effects to be a bit stronger). I know I am aggregating signal across the whole tract, but I thought the statistical relations would have been a bit higher. The isolated tracts were significant with whole-brain analyses (at a higher threshold).
Have others run into this? Is (Recognize and Cluster) a major issue? Is there another/better way to isolate and reduce down group connectometry for use outside of DSI-studio (like in R or python, etc.)? I basically want to create tract of interest files based on the whole connectometry analyses (and hope to have roughly similar results inside or outside of DSI-studio). Any advice is much appreciated!
Thanks much, and all the best,
Jamie.