Hi All,
We are working with using 3dToutCount to assess the quality of our 3d+t datasets, and potentially using the output results as input to a censor file in 3dDeconvolve. If this is an appropriate use of this program, we are trying to establish a criteria for excluding volumes (maybe an arbitrary cutoff, like a certain number of voxels, or percent of voxels?). Hopefully you can address the concerns we've listed below.
1) Is this an appropriate way to use 3dToutCount, as a censor to 3dDeconvolve?
2) How to determine an appropriate cutoff, from the 3dTcount output, for censoring voulmes.
3) Should this cutoff be based upon a standard number (1000 voxels, 10% or voxels), or should it be individualized per subject.
4) The max number of voxels that are considered outliers at any volume seem to differ greatly across subjects, so what is the best way to use this information to exclude volumes that should be considered outliers?
4) How are the outliers determined, is it based upon motion alone? Or are outliers also based upon the quality of the volume (i.e aquisition issues). Many of our subjects that moved very little (less than half voxel width displacement across all runs), sometimes were identified as having many outliers.
Thanks!
Tim Souza