help
help > Number of consecutive volumes needed for stable connectivity results
Jun 5, 2018 02:06 PM | Martijn Teuchies - University of Florida
Number of consecutive volumes needed for stable connectivity results
Dear fellow CONN users,
To get stable functional connectivity results you need at least 5 minutes of data (with a TR of 2,5 seconds this equates to 120 volumes, so I assume this is the minimum number of volumes you need depending on the TR you use). See e.g. the article by Van Dijk et al (2010) Intrinsic functional connectivity as a tool for human connectomics: theory, properties, and optimization.
What I could not find is to what extent these 5 minutes of data or 120 volumes have to be consecutive / back-to-back volumes, or whether you could for instance have 60 volumes at the beginning, lose a substantial number of volumes in the middle (say 15 or 20 if the participant moved a lot in the middle of the session) and then have another 60 volumes at the end. Or alternatively have a lot of outlier volumes scattered across the session (say 50 outliers out of 170 volumes), so you still end up with at least 120 good volumes in total, would the above two scenario's still be sufficient to look at functional connectivity results and have meaningful output?
Related to this, if you have a lot of scattered outliers using strict ART parameters, would it make sense to use a more liberal threshold so you end up with less outlier volumes and thus a higher number of consecutive volumes? Assuming also that the aCompCor method will also take care of motion to some extent? see e.g. Muschelli 2014; Reduction of motion-related artifacts in resting state fMRI using aCompCor.
Thanks,
Martyn
To get stable functional connectivity results you need at least 5 minutes of data (with a TR of 2,5 seconds this equates to 120 volumes, so I assume this is the minimum number of volumes you need depending on the TR you use). See e.g. the article by Van Dijk et al (2010) Intrinsic functional connectivity as a tool for human connectomics: theory, properties, and optimization.
What I could not find is to what extent these 5 minutes of data or 120 volumes have to be consecutive / back-to-back volumes, or whether you could for instance have 60 volumes at the beginning, lose a substantial number of volumes in the middle (say 15 or 20 if the participant moved a lot in the middle of the session) and then have another 60 volumes at the end. Or alternatively have a lot of outlier volumes scattered across the session (say 50 outliers out of 170 volumes), so you still end up with at least 120 good volumes in total, would the above two scenario's still be sufficient to look at functional connectivity results and have meaningful output?
Related to this, if you have a lot of scattered outliers using strict ART parameters, would it make sense to use a more liberal threshold so you end up with less outlier volumes and thus a higher number of consecutive volumes? Assuming also that the aCompCor method will also take care of motion to some extent? see e.g. Muschelli 2014; Reduction of motion-related artifacts in resting state fMRI using aCompCor.
Thanks,
Martyn