Manesh Girn

Hi there, I'm trying to compute ICA on a mask which is a nifti image with all 1s within the default network and all 0s elsewhere (I've also tried defining the mask with NaNs instead on 0s), and the ICA outputs I get are very sparse salt and pepper spatial maps. The spatial ICA is computed using 2 reduction steps (100 and then 10 PCs with standard PCA) , and the ICA is with infomax. I've inspected my input data and there are no abnormalities, and it has been used before. The data is 2 runs of 10 min...

Hi there, I used the DFNC toolbox on my GIFT-calculated ICs, and followed all of the steps done in past research (e.g., as in Allen et al 2014). However, clustering for both 4 and 6 states, many subjects are only spending time in 2-3 of the states. Computing a histogram on 'fraction of time in state' values, 0 represents around half of the total number of values across subjects. Is this something that is common? What might lead this to be the case? I'm not sure if these results suggest that something...

sorry. posted in wrong forum. please see my post in the help forum.

Hi there, I used the DFNC toolbox on my GIFT-calculated ICs, and followed all of the steps done in past research (e.g., as in Allen et al 2014). However, clustering for both 4 and 6 states, many subjects are only spending time in 2-3 of the states. Computing a histogram on 'fraction of time in state' values, 0 represents around half of the total number of values across subjects. Is this something that is common? What might lead this to be the case? I'm not sure if these results suggest that something...

resolved

<resolved></resolved>

Hi there, This may be a relatively basic question, but I'm unclear on the answer. I'm interested in group differences in IC x IC connectivity. In addition to computing two sample t-tests, I want to compute the correlation matrix (static FC) for each group (computed following despiking, detrending, and low-pass filtering). Currently, when I try to run MANCOVAN to do this, I am only able to get the matrices for all subjects (across both groups). Is there a way to use MANCOVAN to get correlation matrices...