Dear FLICA experts,
I am using FLICA to fuse 4 vertex-wise structural cortical metrics outputted from freesurfer (thickness, surface area, curvature and grey/white-contrast (GWC)) (n≈9000). All datasets are mapped to fsaverage and smoothed with an identical kernel (albeit with innate smoothing discrepancies).
I successfully ran FLICA using 1000 iterations and a model order of 60, and both fusing and correlations between ICs and other phenotypes including age and mean MRI metrics all appear reasonable. Still, the GWC metric alone shows extremely low standard deviations from noise, with all values within the relevant column in noiseStedvOut.txt being below 0.4 and often by a lot. Turning off the per-subject noise estimation by adding “opts.lambda_dims = '' did not improve estimates, but fixed the GWC noiseStedvOut values to 0.08.
I tested running FLICA on a separate sample which also had slightly different preprocessing steps, but I got highly similar disfavouring noise estimates for GWC only, perhaps signifying that this is metric-specific.
Any ideas on how to troubleshoot this issue is much appreciated!
Kind regards Linn
########################################################################
To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=FSL&A=1
This message was issued to members of www.jiscmail.ac.uk/FSL, a mailing list hosted by www.jiscmail.ac.uk, terms & conditions are available at https://www.jiscmail.ac.uk/policyandsecurity/
|