It seems like this hasn’t been implemented, yet, but I’m wondering if there is a principled way of dealing with voxel-wise missing data in 3dLME or 3dMVM. This is largely due to a few subjects that have incomplete full-brain coverage. It seems plausible that I could still use their scans as they have data for many, but not all voxels. Thoughts?
In principle that can be implemented by tracking each voxel individually in terms of missing data, but I just haven’t got time pondering about the details. If the number of subjects with missing data is relatively small (e.g., less than 25%) or if you still have large coverage in the regions with missing data for those few subjects, you should be fine.
I know this is an old thread now, but just wondering if there were updates on this front.
Particularly, I am have a dataset where I have 3dAutomask-ed each individual’s image to remove dropout regions. I was hoping to have this excluded person/voxel-wise in the group-level 3dLME - but it seems like it may be reading this in as zero values. I tried using the -bounds options but am getting an error (tested this a few ways but was aiming to use a lower bound of ~0.1 to remove any zeros where all valid values are positive in this case).
3dLMEr -prefix lme
-jobs 4 -bounds .01 10
-qVars ‘age’ -qVarCenters ‘0,’
-gltCode ocd ‘Dx : 1OCD -1Healthy’
***** End of data structure information *****
Reading input files now…
Reading input files for effect estimates: Done!
Range of input data: [0.000, 3.748]
Input data confined within [0.01, 10]
If the program hangs here for more than, for example, half an hour,
kill the process because the model specification or something else
is likely inappropriate.
Loading required package: lmerTest
Loading required package: lme4
Loading required package: Matrix
Attaching package: ‘lmerTest’
The following object is masked from ‘package:lme4’:
The following object is masked from ‘package:stats’:
Loading required package: phia
Loading required package: car
Loading required package: carData
Registered S3 methods overwritten by ‘car’:
boundary (singular) fit: see ?isSingular
 “Great, test run passed at voxel (33, 33, 42)!”
 “Start to compute 91 slices along Z axis. You can monitor the progress”
 “and estimate the total run time as shown below.”
 “05/13/21 14:19:17.469”
Loading required package: snow
Package snow loaded successfully!
Error in checkForRemoteErrors(val) :
4 nodes produced errors; first error: missing value where TRUE/FALSE needed
Calls: aperm … clusterApply → staticClusterApply → checkForRemoteErrors
Some voxels have 0 value, and thus are converted with missing values when you set the range as [0.01, 10]. I’ll make some changes to accommodate the situation.