help
help > RE: Colorbar values
Oct 25, 2016 08:10 PM | Alfonso Nieto-Castanon - Boston University
RE: Colorbar values
Hi Athena,
Regarding (1) the values there represent the same statistics as reported in the original "results explorer" window (typically T-values, but perhaps F- values or chi-square depending on your specific second-level design). And regarding (2) the discrepancy is likely due to different ways of projecting the voxel-level stats to the surface. Unless your original analyses where surface-based then the stastics are computed separately for each voxel. When displaying those same results on the surface there is some level of aggregation/approximation going on depending on the exact position of the surface that one assumes as well as the methods used for projection. In order to minimize a bit those effects CONN's approach is to extract from three different surfaces, placed at the middle and the two extremes of the cortical ribbon from freesurfer fsaverage surface, and then display the average of those three values. This typically produces smoother results than simply extracting from the mid-position only. Other software packages might have a different approach or they might use different surfaces so the projected values might differ. If you want to change CONN's approach you may edit the file conn_mesh_display.m (e.g. comment the two lines that say "use this for smoother display" and uncomment the adjacent lines that say "use this for accurate values" if you want CONN to simply extract the actual mid-ribbon values)
Hope this helps
Alfonso
Originally posted by Athena Demertzi:
Regarding (1) the values there represent the same statistics as reported in the original "results explorer" window (typically T-values, but perhaps F- values or chi-square depending on your specific second-level design). And regarding (2) the discrepancy is likely due to different ways of projecting the voxel-level stats to the surface. Unless your original analyses where surface-based then the stastics are computed separately for each voxel. When displaying those same results on the surface there is some level of aggregation/approximation going on depending on the exact position of the surface that one assumes as well as the methods used for projection. In order to minimize a bit those effects CONN's approach is to extract from three different surfaces, placed at the middle and the two extremes of the cortical ribbon from freesurfer fsaverage surface, and then display the average of those three values. This typically produces smoother results than simply extracting from the mid-position only. Other software packages might have a different approach or they might use different surfaces so the projected values might differ. If you want to change CONN's approach you may edit the file conn_mesh_display.m (e.g. comment the two lines that say "use this for smoother display" and uncomment the adjacent lines that say "use this for accurate values" if you want CONN to simply extract the actual mid-ribbon values)
Hope this helps
Alfonso
Originally posted by Athena Demertzi:
Hello,
at the 2nd level analysis I have results thresholded after correcting for multiple comparisons. Can you pls help me with the following questions?
1. When I use the surface display, I call the colorbar and I see the range of values. Are these T or z?
2. When I load the (exported) mask on BrainNet, the range on the colorbar is different. Why?
Thanks a lot!
Athena
at the 2nd level analysis I have results thresholded after correcting for multiple comparisons. Can you pls help me with the following questions?
1. When I use the surface display, I call the colorbar and I see the range of values. Are these T or z?
2. When I load the (exported) mask on BrainNet, the range on the colorbar is different. Why?
Thanks a lot!
Athena
Threaded View
| Title | Author | Date |
|---|---|---|
| Athena Demertzi | Oct 23, 2016 | |
| Alfonso Nieto-Castanon | Oct 25, 2016 | |
| Athena Demertzi | Oct 31, 2016 | |