help
help > RE: Bandpass filtering in task-based analyses
Nov 13, 2014 06:11 AM | Alfonso Nieto-Castanon - Boston University
RE: Bandpass filtering in task-based analyses
Hi Roger,
In my opinion for task-based connectivity the main concern with filtering is the potential cross-talk/spillage of the BOLD signal between blocks/conditions (e.g. the level of activation within one condition block after filtering will be affected by the original level of activation within another block). Depending on your design I would probably recommend either not filtering or using only a high-pass filter (e.g. [0.008 inf] which implements the standard 128 seconds high-pass used in SPM for regular task analyses), since that could seem like an acceptable compromise between minimizing cross-talk while still benefiting from the increased SNR afforded by filtering, but I would be interested to hear others thoughts as well.
Best
Alfonso
Originally posted by Roger Beaty:
In my opinion for task-based connectivity the main concern with filtering is the potential cross-talk/spillage of the BOLD signal between blocks/conditions (e.g. the level of activation within one condition block after filtering will be affected by the original level of activation within another block). Depending on your design I would probably recommend either not filtering or using only a high-pass filter (e.g. [0.008 inf] which implements the standard 128 seconds high-pass used in SPM for regular task analyses), since that could seem like an acceptable compromise between minimizing cross-talk while still benefiting from the increased SNR afforded by filtering, but I would be interested to hear others thoughts as well.
Best
Alfonso
Originally posted by Roger Beaty:
Hi all,
I'm running a task-based connectivity analysis with two conditions in CONN. For resting-state analyses, one would typically include a bandpass filter (0.008Hz – 0.09Hz) as a preprocessing step; however, it seems that such filtering is not as common with task-based analyses because the task-induced signal may be filtered out of the data. Does anyone have any thoughts on whether bandpass filtering should be applied to task-based data?
Thanks,
Roger
I'm running a task-based connectivity analysis with two conditions in CONN. For resting-state analyses, one would typically include a bandpass filter (0.008Hz – 0.09Hz) as a preprocessing step; however, it seems that such filtering is not as common with task-based analyses because the task-induced signal may be filtered out of the data. Does anyone have any thoughts on whether bandpass filtering should be applied to task-based data?
Thanks,
Roger
Threaded View
| Title | Author | Date |
|---|---|---|
| Roger Beaty | Nov 12, 2014 | |
| Alfonso Nieto-Castanon | Nov 13, 2014 | |
| dspets | Dec 14, 2023 | |