help
help > RE: Gentle Reminder: Setting filter values in denoising step
Apr 15, 2023 12:04 AM | Alfonso Nieto-Castanon - Boston University
RE: Gentle Reminder: Setting filter values in denoising step
Dear Ramesh,
I encourage others with more experience in brainstem imaging to chime in. From my limited experience the main recommendations for denoising are similar in the case of brainstem regions vs. other regions, with the caveat that you can expect larger amounts of physiological noise and movement-related noise in the brainstem due to due to its proximity to CSF and the relatively small size of brainstem nuclei, so if possible/reasonable trying a more aggressive denoising strategy may be warranted. Regarding filtering, unless you have a rather fast acquisition (e.g. TR below 200ms) filtering does not usually play a large role in removing physiological noise (since physiological noise components are aliased into the entire frequency range of the sampled BOLD signal when TR is relatively slow), so using a more restrictive filter may not be very helpful, on the contrary if you are planning to use more aggressive denoising strategies (which may require removing a higher number of components from the BOLD signal; e.g. using Friston24 or higher motion-related regressors) you may want to consider using a less-restrictive filter (e.g. a high-pass filter only: 0.01 Hz - inf) mainly for the increase in the residual degrees of freedom that this would offer you.
Hope this helps
Alfonso
Originally posted by Ramesh Babu MG:
I encourage others with more experience in brainstem imaging to chime in. From my limited experience the main recommendations for denoising are similar in the case of brainstem regions vs. other regions, with the caveat that you can expect larger amounts of physiological noise and movement-related noise in the brainstem due to due to its proximity to CSF and the relatively small size of brainstem nuclei, so if possible/reasonable trying a more aggressive denoising strategy may be warranted. Regarding filtering, unless you have a rather fast acquisition (e.g. TR below 200ms) filtering does not usually play a large role in removing physiological noise (since physiological noise components are aliased into the entire frequency range of the sampled BOLD signal when TR is relatively slow), so using a more restrictive filter may not be very helpful, on the contrary if you are planning to use more aggressive denoising strategies (which may require removing a higher number of components from the BOLD signal; e.g. using Friston24 or higher motion-related regressors) you may want to consider using a less-restrictive filter (e.g. a high-pass filter only: 0.01 Hz - inf) mainly for the increase in the residual degrees of freedom that this would offer you.
Hope this helps
Alfonso
Originally posted by Ramesh Babu MG:
Dear Experts,
I have a question regarding setting the filter values in denoising step for ROI analysis. I selected some cortical regions from the default atlas provided by the CONN and also defined ROIs based on MNI coordinates for brainstem regions including dorsolateral pons and PAG regions which have high amount of physiological noise. For setting the filter values in denoising step, what would be the ideal value to filter out the physiological noises from the brainstem regions also? Currently, I used 0.01 - 0.1. will it be ideal to use for cortical and brainstem regions? it okay to continue with this filter value?
Expecting your suggestions.
Thanks in advance,
Ramesh
I have a question regarding setting the filter values in denoising step for ROI analysis. I selected some cortical regions from the default atlas provided by the CONN and also defined ROIs based on MNI coordinates for brainstem regions including dorsolateral pons and PAG regions which have high amount of physiological noise. For setting the filter values in denoising step, what would be the ideal value to filter out the physiological noises from the brainstem regions also? Currently, I used 0.01 - 0.1. will it be ideal to use for cortical and brainstem regions? it okay to continue with this filter value?
Expecting your suggestions.
Thanks in advance,
Ramesh
Threaded View
| Title | Author | Date |
|---|---|---|
| Ramesh Babu MG | Apr 4, 2023 | |
| Ramesh Babu MG | Apr 13, 2023 | |
| Alfonso Nieto-Castanon | Apr 15, 2023 | |
| Ramesh Babu MG | Apr 17, 2023 | |