help
487 Subscribers 
help > Determining Best Parameters for CONN
Showing 1-1 of 1 posts
Nov 4, 2016 05:11 PM | Emily Belleau - McLean Hospital/Harvard Medical School
Determining Best Parameters for CONN
Hi Alfonso,
I am brand new to using CONN. I want to make sure that I am using the best parameters for optimizing my resting state functional connectivity analysis. I have preprocessed all of my data in SPM and used ART to regress outliers outside of CONN.
1.) Structural Files: I should be using the normalized structural (ending in wm?)
2.) Functional: Should I choose the smoothed files, then in the ROI section, I should choose just the normalized functional files to decrease the chances that the ROI will spill into other structures?
3.) Bandpass settings: I have typically used 0.01 to 0.1. However, I noticed the default in CONN is a bit different. Am I better off leaving the default. I have seen some decent variability in bandfilter settings across studies? Any guidance on how to choose the best settings? If I wanted to do both static and dynamic rsFC, I have read that a bandpass of 0.0278-0.10 might be best?
4.) Bandpass: Is it better to do simultaneous? Would you recommend this? I have been told It is recommendable that all regressions are carried out in one step, i.e. including the CSF and WM and bandpass regressors together. In such a way, the fitting can properly deal with possible correlations between the regressors.
5.) Despiking....I typically do before regression, would you recommend? Then also do linear detrending.
6.) Motion as 1st level covariates: I have the ART outliers and motion parameters files: containing the outliers based on global mean intensity, motion outliers, and then the residual motion parameters (three translation and three rotation parameters, plus one composite motion parameter reflecting the maximum scan-to-scan movement) I can also get the temporal derivatives by changing the derivative order to 1 in the preprocessing step? I assume I should also out pub the subject level aggregates to later look at maximum movement per sub?
7.) Is there any way to change the labeling of subjects so that they may reflect the actual subject number used in the study (and labeled on the structural and functional files imported into CONN)? I'd like to be able to do this to avoid any confusion with determining which subject is which and linking subject to their second level covariates?
Sorry for all of the questions! I very much appreciate any guidance!
Emily
I am brand new to using CONN. I want to make sure that I am using the best parameters for optimizing my resting state functional connectivity analysis. I have preprocessed all of my data in SPM and used ART to regress outliers outside of CONN.
1.) Structural Files: I should be using the normalized structural (ending in wm?)
2.) Functional: Should I choose the smoothed files, then in the ROI section, I should choose just the normalized functional files to decrease the chances that the ROI will spill into other structures?
3.) Bandpass settings: I have typically used 0.01 to 0.1. However, I noticed the default in CONN is a bit different. Am I better off leaving the default. I have seen some decent variability in bandfilter settings across studies? Any guidance on how to choose the best settings? If I wanted to do both static and dynamic rsFC, I have read that a bandpass of 0.0278-0.10 might be best?
4.) Bandpass: Is it better to do simultaneous? Would you recommend this? I have been told It is recommendable that all regressions are carried out in one step, i.e. including the CSF and WM and bandpass regressors together. In such a way, the fitting can properly deal with possible correlations between the regressors.
5.) Despiking....I typically do before regression, would you recommend? Then also do linear detrending.
6.) Motion as 1st level covariates: I have the ART outliers and motion parameters files: containing the outliers based on global mean intensity, motion outliers, and then the residual motion parameters (three translation and three rotation parameters, plus one composite motion parameter reflecting the maximum scan-to-scan movement) I can also get the temporal derivatives by changing the derivative order to 1 in the preprocessing step? I assume I should also out pub the subject level aggregates to later look at maximum movement per sub?
7.) Is there any way to change the labeling of subjects so that they may reflect the actual subject number used in the study (and labeled on the structural and functional files imported into CONN)? I'd like to be able to do this to avoid any confusion with determining which subject is which and linking subject to their second level covariates?
Sorry for all of the questions! I very much appreciate any guidance!
Emily