help
help > RE: Extracting voxel-time-series for ROI-voxels
Nov 26, 2014 03:11 AM | Alfonso Nieto-Castanon - Boston University
RE: Extracting voxel-time-series for ROI-voxels
Dear Tobias,
Yes, the Denoising step is processed separately as two independent pipelines for voxel-level data and ROI-level data, so one possible reason for the <1 correlation when checking 'despiking' would be if you choose the 'do not overwrite' option (or select for example only the 'ROI-to-ROI' pipeline right after pressing 'Done' in the Denoising tab) so only the new ROI gets processed with the modified Denoising options (while the rest of the ROIs as well as voxel-level timeseries remain unchanged using the previously defined values). If this is the reason for the observed discrepancy it should go away if you re-run the Denoising step making sure both the 'ROI-to-ROI' and 'seed-to-voxel' checkboxes are checked, and also leaving the default 'proceed for all subjects' option selected as well, since that will run all of your subjects ROI-level data as well as voxel-level data through the same Denoising settings. Let me know if this seems to explain the observed differences.
Best
Alfonso
Originally posted by Tobias Meindl:
Yes, the Denoising step is processed separately as two independent pipelines for voxel-level data and ROI-level data, so one possible reason for the <1 correlation when checking 'despiking' would be if you choose the 'do not overwrite' option (or select for example only the 'ROI-to-ROI' pipeline right after pressing 'Done' in the Denoising tab) so only the new ROI gets processed with the modified Denoising options (while the rest of the ROIs as well as voxel-level timeseries remain unchanged using the previously defined values). If this is the reason for the observed discrepancy it should go away if you re-run the Denoising step making sure both the 'ROI-to-ROI' and 'seed-to-voxel' checkboxes are checked, and also leaving the default 'proceed for all subjects' option selected as well, since that will run all of your subjects ROI-level data as well as voxel-level data through the same Denoising settings. Let me know if this seems to explain the observed differences.
Best
Alfonso
Originally posted by Tobias Meindl:
Dear Alfonso,
thank you very much for your last answer. It made me feel a lot more confident about my data-extraction approach.
As you suggested I checked the approach and now got a correlation coefficient of 1, at least if I unchecked the "despiking"-option in the preprocessing step.
The time-series still do not exactly match for a scaling problem which I assume is due to a scaling within the ROI vs. scaling with the entire ROI. If this is true the difference is meaningless for further work with data extracted from the niftiDATA_Subject001_Condition001.nii-file.
The next thing I encountered is that the correlation for the time courses (my extracted mean) vs. the ROI-mean shown by conn and contained in the ROI_Subject001_Condition001.mat-file is 1 if the desping option in the preprocessing step is unchecked whereas is it is slightly less if the "despiking"-option is checked.
Is it correct that the despinking procedure is also done seperately for ROI-Data and whole-brain-Data? This would again mean that inspite of the slight differences with the conn-output I could proceed.
Again I would like to mention how grateful I am for your last answers!
Kind regards
Tobias
thank you very much for your last answer. It made me feel a lot more confident about my data-extraction approach.
As you suggested I checked the approach and now got a correlation coefficient of 1, at least if I unchecked the "despiking"-option in the preprocessing step.
The time-series still do not exactly match for a scaling problem which I assume is due to a scaling within the ROI vs. scaling with the entire ROI. If this is true the difference is meaningless for further work with data extracted from the niftiDATA_Subject001_Condition001.nii-file.
The next thing I encountered is that the correlation for the time courses (my extracted mean) vs. the ROI-mean shown by conn and contained in the ROI_Subject001_Condition001.mat-file is 1 if the desping option in the preprocessing step is unchecked whereas is it is slightly less if the "despiking"-option is checked.
Is it correct that the despinking procedure is also done seperately for ROI-Data and whole-brain-Data? This would again mean that inspite of the slight differences with the conn-output I could proceed.
Again I would like to mention how grateful I am for your last answers!
Kind regards
Tobias
Threaded View
| Title | Author | Date |
|---|---|---|
| Tobias Meindl | Oct 25, 2014 | |
| Tobias Meindl | Nov 18, 2014 | |
| Alfonso Nieto-Castanon | Nov 18, 2014 | |
| Tobias Meindl | Nov 25, 2014 | |
| Alfonso Nieto-Castanon | Nov 26, 2014 | |
| Alfonso Nieto-Castanon | Nov 15, 2014 | |