open-discussion
open-discussion > RE: Filtering data
Aug 17, 2015 06:08 PM | Alfonso Nieto-Castanon - Boston University
RE: Filtering data
Dear Giannarita,
Band-pass filtering in CONN (as part of the Denoising step) is applied, separately for each session, to both the voxel-level timeseries (the preprocessed BOLD-signal functional volumes) and the ROI-level timeseries (the extracted BOLD-signal for each specified ROI). In general CONN will filter your data as part of the Denoising step using the specific band-pass window specified there (you can change the choice of band-pass window in the GUI Denoising tab 'band-pass filter' field, or if using batch scripts in the batch.Denoising.filter field)
The implementation uses the same general procedure as SPM which simply removes the out-of-band components in a DCT (discrete cosine transform) of the timeseries. The function conn_filter.m implements this (it uses a FFT of the even-symmetry-extended original timeseries; there are no additional default parameters there, so let me know if you would like to use a different form of band-pass filtering -e.g. using a finite-impulse filter- so I can perhaps guide you on how to change that in CONN).
Hope this helps
Alfonso
ps. for future reference, when posting CONN-related questions please consider using the 'NITRC CONN' forum (go tohttp://www.nitrc.org/projects/conn and select 'support' or 'forum' there) instead of this more general 'NITRC community' forum (posts there have a better chance of standing out to some of the CONN developers or users, and you can more easily search there for previous questions that may relate to your current CONN-related problem/question/thoughts). Hope this helps
Originally posted by Giannarita Iannotti:
Band-pass filtering in CONN (as part of the Denoising step) is applied, separately for each session, to both the voxel-level timeseries (the preprocessed BOLD-signal functional volumes) and the ROI-level timeseries (the extracted BOLD-signal for each specified ROI). In general CONN will filter your data as part of the Denoising step using the specific band-pass window specified there (you can change the choice of band-pass window in the GUI Denoising tab 'band-pass filter' field, or if using batch scripts in the batch.Denoising.filter field)
The implementation uses the same general procedure as SPM which simply removes the out-of-band components in a DCT (discrete cosine transform) of the timeseries. The function conn_filter.m implements this (it uses a FFT of the even-symmetry-extended original timeseries; there are no additional default parameters there, so let me know if you would like to use a different form of band-pass filtering -e.g. using a finite-impulse filter- so I can perhaps guide you on how to change that in CONN).
Hope this helps
Alfonso
ps. for future reference, when posting CONN-related questions please consider using the 'NITRC CONN' forum (go tohttp://www.nitrc.org/projects/conn and select 'support' or 'forum' there) instead of this more general 'NITRC community' forum (posts there have a better chance of standing out to some of the CONN developers or users, and you can more easily search there for previous questions that may relate to your current CONN-related problem/question/thoughts). Hope this helps
Originally posted by Giannarita Iannotti:
Dear all,
I would like to know better how the band pass filter in CONN is performed.
More precisely:
-Are all the volumes of interest filtered or the average time-course extracted from each specified ROI is filtered?
-How the filtering is performed? Is there a way to edit the function associated to the filter step to know the defaul paremeters?
Thank you very much for your help
Giannarita
I would like to know better how the band pass filter in CONN is performed.
More precisely:
-Are all the volumes of interest filtered or the average time-course extracted from each specified ROI is filtered?
-How the filtering is performed? Is there a way to edit the function associated to the filter step to know the defaul paremeters?
Thank you very much for your help
Giannarita
Threaded View
| Title | Author | Date |
|---|---|---|
| Giannarita Iannotti | Aug 17, 2015 | |
| Daniel Rogers | Nov 3, 2015 | |
| Alfonso Nieto-Castanon | Aug 17, 2015 | |
| Jean-Baptiste Poline | Aug 17, 2015 | |
| Alfonso Nieto-Castanon | Aug 18, 2015 | |