Intersubject correlation (ISC) measures the extent of BOLD response similarity or synchronization between two subjects who are scanned under the same experience such as movie watching, music learning, etc. The new program 3dISC (available in the next AFNI release) performs voxelwise analysis at the population level with linear mixed-effects (LME) modeling that is laid out in the following paper:
Chen, G., Taylor, P.A., Shin, Y.W., Reynolds, R.C., Cox, R.W., 2017. Untangling the Relatedness among Correlations, Part II: Inter-Subject Correlation Group Analysis through Linear Mixed-Effects Modeling. Neuroimage 147:825-840.
The LME platform allows for the incorporation of various types of explanatory variables including categorical (between- and within-subject factors) and quantitative variables (e.g., age, behavioral data). The details of running 3dISC can be found in the help
3dISC -help
A few demonstrative examples are shown in the help as well.
This is great news! Will be very helpful in my analysis.
I was wondering if 3dISC has the capability to do an ISC analysis based on ROIs rather than voxels? I was hoping to parcel my functional images into networks with ICA and to calculate the network-wise, rather than voxel-wise, ISC between subjects. Would I be able to use the images that have undergone ICA as inputs for 3dISC? Or do I need to re-consider my analysis if I want to use the program?
I was wondering if 3dISC has the capability to do an ISC analysis based on ROIs rather than voxels?
I was hoping to parcel my functional images into networks with ICA and to calculate the network-wise,
rather than voxel-wise, ISC between subjects.
Would it be something like what’s described in the following paper?
Would I be able to use the images that have undergone ICA as inputs for 3dISC? Or do I need to
re-consider my analysis if I want to use the program?
The input for 3dISC is typically the voxelwise ISC values from each pair of subjects. You would have to decide how your ICA results fit into this current context.
Hi Gang, sorry, I did not see this post until now. I read it over and yes - that is pretty much exactly what I was hoping to do. I would be creating the parcellations myself through ICA, rather than using pre-defined set, but other than that, the methods described are the same as I was imagining. Good to know that that is possible. Also – Part III was very interesting – excited to see the future updates that will be incorporating the BML.
Thank you!
Ryann
Gang Wrote:
I was wondering if 3dISC has the capability to
do an ISC analysis based on ROIs rather than
voxels?
I was hoping to parcel my functional images into
networks with ICA and to calculate the
network-wise,
rather than voxel-wise, ISC between subjects.
Would it be something like what’s described in the
following paper?
Would I be able to use the images that have
undergone ICA as inputs for 3dISC? Or do I need to
re-consider my analysis if I want to use the
program?
The input for 3dISC is typically the voxelwise ISC
values from each pair of subjects. You would have
to decide how your ICA results fit into this
current context.
The
National Institute of Mental Health (NIMH) is part of the National Institutes of
Health (NIH), a component of the U.S. Department of Health and Human
Services.