Calculating Connectivity between 2 ROIs

Submitted by yhcha@mednet.uc... on Thu, 08/25/2011 - 04:57

Dear Rest MRI Forum,
I have recently completed a PET scan study between a patient group and controls and have computed some ROIs which are more active in the patients or more active in the controls. I'd like to see whether these ROIs are functionally connected. eg. whether a region that is more active in the patients is actually functionally connected to a region that is more active in the controls.
I am wondering which REST function I can use for this. Theoretically, I'd like to input two ROI .mat or .nii files and then have an output of the degree of connectivity.

Thank you very much for this and all the prior great help.
Yoon-Hee

8-25-11: Addendum.
I used the Functional Connectivity button on the REST GUI and entered the coordinates of two ROIs. It gives me 3 outputs. The first is called FCMap_ALFF.txt that has a number like 0.85-0.97.
The second is file called ROI_FCMap_ALFF, which has a long list of numbers, which I believe is a bunch of correlation coefficients. The third file is zFCMAP_ALFF.txt which gives a single value of something like 1.22-2.13. I am wondering which of these is the true representation of the functional connectivity, ie what should be reported. Thank you.

Submitted by YAN Chao-Gan on Thu, 09/01/2011 - 05:20

Hi!

FCMap_ALFF.txt: this is the Pearson's correlation coefficient "R" value
zFCMap_ALFF.txt: this is Fisher's z value. Before statistical analysis, people usually convert the raw "R" value into "Z" value to improve normality by Fisher's R-to-Z transformation.
ROI_FCMap_ALFF.txt: this are the time series for your two ROIs. Each column represents one ROI.

If you want to perform statistical analysis, you may need to use the zFCMap_ALFF.txt.

Best wishes!

Submitted by yhcha@mednet.uc... on Sun, 09/11/2011 - 14:29

Thank you so much Chao-Gan.
I think that I was confused because the z values I get are very small. I am using a single voxel seed (spherical voxel with radius of 1mm) with 110 volumes of data. I am using the connectivity of the two frontal eye fields as a control. These connectivities are about z=2.0. When I used motion sensitive area MT, the connectivity is z=1.52. All other connectivity values from my ROIs range from about 1.0 to 1.7. In papers I have read, the z scores can get up to 15. Thus, I was concerned that perhaps I was not doing the right thing. But, could these low values be because there are few data points in a single voxel?
Thanks so much. You've been life-saver!
Yoon-Hee

Submitted by YAN Chao-Gan on Thu, 09/15/2011 - 10:10

Hi!
Actually, z=2.0 is a very strong connectivity, it corresponds to r=0.964!!!

I guess you might be confused the statistical z value of 15 with this Fishers' z. After Statistical analysis on Fisher's z, you will get T values. And somebody will report this T value as Z value (since Z is statistical value without degree of freedom). Please read that paper carefully.

Regarding to your ROI, usually people will not use such a small ROI. Common radius is 6mm as in most literates.

Best wishes!

Submitted by yhcha@mednet.uc... on Thu, 09/15/2011 - 13:31

Thank you so much Chao-Gan. Your expertise and explanations have been so helpful to me and have really helped my project. I was really confused about the z score and wondered whether I had processed the data incorrectly. Thank you for setting me straight.
Thank you SOOOO much.
Yoon-Hee