help > importing missing / incomplete
May 14, 2018  09:05 AM | Harrison Fisher - Martinos Center
importing missing / incomplete

I am working on a dataset where participants were either controls or patient population. Controls came in for 1-3 fMRI scans and patients came in for 2 visits (pre and post treatment). 

During the scan, we collected a resting state scan initial (6'), and then after some more sequences, went through two more of the same "resting state" scan but this time while receiving a form of neuromodulation. These two scans were performed back to back with a few minutes between. 
An anatomical T1 was collected at each session. 

So each healthy volunteer could have up to 9 relevant "resting state" runs (3 sessions x (rest + neuromod1 + neurmod2))
and each patient could have up to 6 (pre vs post x (rest + neuromod1 + neurmod2)). 

My questions concern how to deal with grouping/organizing the data in conn and what do with missing data. 

Often something went wrong with one of the neuromodulation scans so that there is some missing data.

For example : a patient might have pre_rest, pre_neuromod1, post_rest, post_neuromod2 scans only. 

Additionally, some healthy volunteers came in 3x , where as others only had 1 study visit. 

How should I deal with the missing data? Should each scan be considered a separate "session" in conn and grouped together using conditions and second level covariates for study group? 

I already tried once loading in all the data specifying the number of "sessions" for each subject individually. But this also creates the same number of sessions for the structural images. So I might have a healthy volunteer who came for 2 visits and received all 3 scans at each (6 functional runs), but they have only 2 anatomical images. Is there a way to specify the mappings of the sessions to the specific anatomical? 

Thanks in advance for any input! 

- Harris Fisher

P.S. The CONN workshop was incredibly helpful and I thoroughly enjoyed it!